Official statistics need a culture change

Despite the integrity of its data, the Australian Bureau of Statistics is hardly renowned for its user-friendliness or its accessibility. It could learn a lesson or two from the private sector.

Financial difficulties and concerns about its ageing technology highlight the need to re-think the role of the Australian Bureau of Statistics. There is scope for much greater private sector involvement in the collection and dissemination of both official and unofficial statistics.

The ABS is well regarded internationally for the quality and integrity of its data, although there have been occasions when key users such as the Reserve Bank of Australia have disputed the accuracy of some important indicators and have been vindicated.

The ABS is not as well regarded for the user-friendliness and accessibility of its data. It's hardly alone among statistical agencies in this regard, but there is no reason why a focus on the collection of data should detract from the equally important task of data dissemination.

A good example of what can be achieved in terms of accessibility and user-friendliness are the comprehensive economic and financial databases, associated apps and software maintained by the Federal Reserve Bank of St. Louis in the United States, known as Federal Reserve Economic Data or FRED.

The distinctive charts generated by the FRED web interface have become ubiquitous in presentations, media and other contexts -- a testimony to their accessibility.

The FRED database also seamlessly integrates with popular statistical software packages. There is no reason why ABS data should not do the same.

One of the remarkable things about FRED and the associated software is that they are maintained by only half-a-dozen full-time St. Louis Fed employees working in a culture that is said to be more like that of a start-up web business than a branch of a central bank. The St. Louis Fed has shown what should be possible even with limited resources.

The Quandl search engine for numerical data is another example of what can be done to make data more accessible and useable.

Australia, along with New Zealand, is one of the few countries in the world to release its consumer price index at a quarterly rather than a monthly frequency. Even New Zealand manages to produce a monthly food price index, an important component of the CPI.

The ABS has argued that the extra $15 million a year it would cost to collect the CPI at a monthly frequency outweighs the potential user benefits, an assessment the Reserve Bank of Australia would certainly question.

The quarterly CPI introduces a backward-looking bias into monetary policy decision-making, with changes in official interest rates most likely to occur at the RBA board meeting immediately following the CPI release.

By contrast, the Melbourne Institute has been able to produce a monthly inflation gauge with a low tracking error to the official CPI at relatively low cost. Google’s ‘billion prices’ project suggests that private real-time measures of consumer price inflation may not be far off.

Savings could potentially be made by reducing the frequency of the monthly ABS labour force survey to a quarterly release. The labour force data are good for generating media headlines, but they are a noisy and unreliable set of figures at a monthly frequency. Australia previously released its labour market data at a quarterly frequency until the current monthly series began in 1978.

The combination of a monthly labour force release coupled with a quarterly CPI may even distort the debate over economic policy by increasing the frequency of media and financial market attention on the labour market relative to inflation outcomes.

Few people mourned the demise of the ABS’s volatile and revision-prone monthly current account release in 1997. By allowing the monthly trade balance to be reported independently of the net income balance, the move to a quarterly release for the current account probably improved the accuracy of public perceptions of Australia’s trade position.

Widespread dissatisfaction with ABS data on house prices led RP Data-Rismark to develop a superior set of measures that have improved our understanding of the housing market. These measures have increased the scope for housing to become an investable asset class on the part of institutional investors and may eventually lead to the development of fully-fledged derivatives markets that can be used to hedge housing market exposures.

There is considerable scope for the private sector and other official agencies to improve on the work of the ABS by making greater use of so-called ‘big data.’ For its part, the ABS should not expend resources collecting data the private sector has a strong incentive to produce and disseminate.

There is scope for greater private sector involvement in the generation and dissemination of official statistics as well. To fully realise this potential will require a significant change in the organisational culture of official statistical agencies like the ABS.

Dr Stephen Kirchner is a research fellow at The Centre for Independent Studies.

Related Articles