Monday, September 3, 2018

How to damage Australia: don’t collect good data

You don’t have to be very bright to see that as we enter the information age, realise decisions need to be evidence-based, and glimpse the huge potential of “big data”, we need the Australian Bureau of Statistics to be at the top of its game. But you do have to be brighter than our econocrats and politicians.

They’ve been cutting the bureau’s funding every year for more than a decade – meaning both parties have been at it – in the name of increased efficiency. The Orwellian annual “efficiency dividend”, cutting up to 2.5 per cent off running expenses, is a flowing fount of false economy.

According to the bureau’s boss, David Kalisch, it has suffered a reduction in real resources of more than 20 per cent over the past decade. Meanwhile, funding from big users of its data – which now accounts for between 10 and 20 per cent of its total funding - has increased only slightly.

The majority of its social statistical collections are only possible through user funding, with budget funding devoted predominantly to its economic and population stats.

The cutbacks have obliged the bureau to “prioritise”. It has reduced or stopped a number of statistical collections, with Kalisch admitting it hasn’t undertaken a survey of the way Australians use their time, nor a survey of mental health, for more than a decade.

“If the [bureau] continues to be subject to efficiency dividends over the next decade, at the same trajectory as it has for the past decade, some of the core information currently taken for granted by governments, business and the community may no longer be available,” he told a conference last month.

“Our capacity to continue producing all of the detailed statistics around our labour market, industry activity and population would increasingly be at risk.”

It oughtn’t be necessary to remind politicians, bureaucrats, marketers, academics, journalists and ordinary citizens just how heavily we rely on our national statistical office for reliable, objective information about a hundred dimensions of what’s actually happening around us, including to the natural environment.

The bureau’s data inform “fiscal and monetary policy settings, social support programs and infrastructure spending . . . many pertinent public policy debates, such as housing affordability, income and wealth inequality, cost of living, energy prices, the quality of life in our cities and regions, education and health outcomes, needs-based school funding, immigration policy and much more,” Kalisch told a conference of economists.

That’s not to mention that official data are “key to the effective functioning of our democracy, with population data helping establish fair electoral boundaries and our official statistics informing choices by voters and political aspirants”.

But it’s not just that we’d be much more poorly informed if government spending cuts robbed us of any of the information we presently collect. Our economy, society and natural environment keep changing, meaning we need to measure more than we do at present, as well as improving the way we measure things because they’ve changed from what they were.

Kalisch says globalisation and the digital economy introduce new measurement challenges. Over the past 15 years, the services sector has grown at an average rate of 6 per cent a year, meaning it now accounts for 63 per cent of gross domestic product [and a much higher proportion of total employment].

Measuring services is more difficult – conceptually and empirically – than goods. Good measurement of two key industries – health and education – is particularly important.

“Policy-makers and service providers are confronting wicked [difficult or impossible] problems across social policy and the environment that require a more sophisticated evidence base,” he says.

The bureau was an early public sector adopter in using computers, but in 2013 Kalisch’s predecessor blew the whistle on its “fragile ageing statistical infrastructure”. In 2015 the government agreed to provide most of the additional funding to build new systems.

In 2016 the bureau struck trouble with its first go at having many people complete their Census forms online. At the start of the filing period, the system was offline for nearly two days.

It was a “teachable moment”, but the bureau “owned the process errors, has reflected upon the learnings from this experience" and has revised its operating arrangements across the bureau. As proof it has learnt its lesson, Kalisch points to its trouble-free conduct of the same-sex marriage postal survey.

And all this before we get to big data. Any fool can see its huge potential for improving our evidence base at relatively low cost. But it takes a bit more brain to see that if we barge on with little attention to the public’s concerns over privacy and Big Brother governance, we could derail the whole show before we even get going.

Just the right time to cut the funding of the national statistical agency and decide we can afford to do stats on the cheap.