Monday, March 7, 2016

Let’s stand against misleading modelling

Many people have been left with red faces following their part in last week's disastrous intervention into the negative-gearing debate by forecasters BIS Shrapnel. Let's hope they all learn their lesson.

This isn't the first time that "independent" modelling purchased from economic consultants has been used by vested interests to try to influence government decisions. Nor the first time the questionable results have been trumpeted uncritically by the media and misrepresented by the side of politics whose case it happens to suit.

But BIS Shrapnel's late entry into this dubious game has come at a time when the game's credibility is wearing thin and qualified observers are more willing to go public with their critiques of the quality of the modelling, the plausibility of its assumptions and the internal consistency of its findings.

As is common practice, various of the BIS Shrapnel model's findings were expressed in a highly misleading way. "Rents will rise by up to 10 per cent ($2,600) per annum", for instance, doesn't mean rents will rise by up to 10 per cent a year. It actually means that, by the 10th year, annual rents will be up to 10 per cent higher than they otherwise would be. Not nearly as bad as it was made to sound.

The first lesson for BIS Shrapnel is that when you publish commissioned modelling, but agree not to disclose who commissioned it, you attract a lot more criticism and scepticism. When it's not possible for those on the other side of the debate to say "they would say that, wouldn't they", they examine your assumptions and methodology a lot more critically.

Another lesson is that when what you're modelling looks like it's a party's policy but isn't, you should say so up front, not in mitigation after that party has denounced you from the rooftops.

Similarly, "unfortunate typos" saying $190 billion when you meant $1.9 trillion get you hugely adverse attention. Your "trust me, I'm an economist" line implodes.

I can't remember when so many economists of repute have gone out of their way to attack a modeller's findings, and done it so bluntly.

John Daley, of the genuinely independent Grattan Institute, referred to the report's "convoluted logic", "manifestly ridiculous predictions", "outlandish" and "fanciful" claims, and "implausible" and "unjustified" assumptions. It was "nonsense on stilts".

The lesson for other economic consultants is that the days when you could produce for a client a bit of happy advocacy posing as objective econometric analysis, and have the rest of the profession look the other way, are coming to an end.

There's now a far greater likelihood that other economists or economic journalists will subject your assumptions, methodology and findings to scrutiny and make their conclusions public.

There's now much greater familiarity with the standard tricks of the trade, such as misuse of the Bureau of Statistics' "input-output tables" to exaggerate the "indirect effects" of some measure; saying "employment will fall by X" when you really mean "the growth in employment will be X less than otherwise", or presenting effects that build slowly over many years as changes that occur fully in the first year and occur again in each subsequent year.

The lesson for relatively new treasurers trying to establish a reputation for economic competence, and the ability to explain complex economic concepts persuasively, is you'll never do it if you act like a political brawler and latch on to whatever third-party modelling seems to be going your way.

A treasurer looking for respect doesn't identify himself with any modelling before his experts – the economists in his department, not the ambitious young politicos in his office – assure him it's kosher.

If I was a subscriber to an Australian newspaper that led its front page with a wide-eyed account of BIS Shrapnel's findings as though they were established fact, only to have them exposed the same day as highly debatable, I wouldn't be impressed.

The lesson for the economics profession is that the modelling they value so highly is too often being used by other economists to mislead rather than enlighten. The reputation of models and modellers is being trashed, and with it the credibility of the profession.

If economists don't want to be regarded by the public as charlatans, they should consider the call by the Australia Institute – a noted debunker of misleading modelling – for a code of conduct for economic modelling. It would "require key assumptions to be revealed, context and comparison to be provided, and the identification of who, if anyone, commissioned the work".

Since the profession has failed to act, the institute wants the code implemented by governments.