Digital Transformation » Technology » Predicting the present

Predicting the present

The use of price information from the internet could be a huge resource for businesses to get ahead of the official data, writes Phil Thornton

IN THE MODERN age of email communication, real-time news and “flash trading” of shares, the release of official data on the economy seems stuck in the slow lane of the information superhighway.

The first estimate of 0.2% growth in the UK for the three months to June was published almost four weeks later, on 26 July. Figures on trade and unemployment have a six-week time lag. Yet the UK is a world leader in the speed and accuracy of official data. As a result, economists are now looking at how they can harness the immense volume of real-time online data to give businesses a head start.

Researchers at the Massachusetts Institute of Technology (MIT) have designed an index of inflation based on five million prices collected every day. The Billion Prices Project (BPP) scans web pages for prices of goods such as food and drink, household products and clothing.

The result is a real-time index of price changes. Roberto Rigobon, one of the founders, says the real-time nature of the data collected makes it valuable in the current economic environment.

“Our data provides the right level of detail and relevance required to not only track official statistics, but also forecast future trends,” he says.

The project scored an instant hit in September 2008 when the Lehman Brothers bank collapsed. “Our US average online price index started to drop only two days after Lehman and we were able to see prices recovering in early January 2009 – well before a trend could be seen in official CPI announcements,” adds Rigobon.

The index has now gone global. State Street, a US bank, provides inflation data on the UK, Germany, France and Brazil for its clients in a partnership with PriceStats, a spin-off from MIT. It has already picked up startling data in the UK: the index showed a one percentage point increase in food prices in the five days after 28 March when the government imposed large rises in alcohol and tobacco duty (see graph 1).

“It is almost as if the stores were timing their price increases in anticipation of the tax increases,” says Rigobon. “They seem to have used the fact that people were going to get upset about alcohol anyway, to use that day to increase prices of other things.”

Speaking towards the end of July, he says the BPP data did not show the slowdown in inflation in June reported by official data and shows inflation rising back to 5% in either July or August. “We expect the inflation rate to be close to 5%,” he says.

Real time reporting

p35-google-trends-v2The Bank of England has also dipped its toe in the water, looking at how searches for key terms on Google can act as a predictor of trends in the labour and housing markets, as well as in consumer confidence.

In an exercise dubbed “nowcasting”, it found that searches for “estate agents” would track house prices, while “JSA” – the acronym for the Jobseeker’s Allowance – pointed to moves in unemployment.

But the big issue is how useful this type of information will be to businesses, and to what degree they can rely on it when making strategic decisions. Rigobon agrees that it could be valuable for companies.

Turning online information into forecasts has been a hot issue since Hal Varian, chief economist at Google, published a paper in 2009, in which he claimed that the Google Trends function could “predict the present”.

He found that the volume of queries on one type of automobile during the second week in June could help predict the June sales report for that brand, which is not released until sometime in July.

“It may also be true that June queries help to predict July sales, but we leave that question for future research,” he wrote with co-author Hyunyoung Choi.

Louise Ross, head of corporate performance management at the Chartered Institute of Management Accountants (CIMA), says there is clearly potential.

“We live in a digital society, in which our online transactions – booking, buying, searching, feeding back – can be captured and reported in real time,” she says. “The size of the internet allows organisations to observe the behaviour of millions of people – it is an extremely rich data source.”

The International Federation of Accountants (IFAC) has launched a consultation into the field, which it calls business predictive analysis.

It claims that, if managed well, it should enable businesses to anticipate future events, forecast possible outcomes and select actions to improve performance. IFAC points to a PwC survey of more than 400 senior leaders in both the public and private sector that found high-performing companies made greater use of alerts, driver-based forecasting and data mining.

IFAC has set out key principles, the most important of which are that the data must be relevant, of reliable quality and able to show a strong cause-and-effect relationship.

Ross also mentions the example of Puma, the German sportswear company, which recently realised it needs to track water and agricultural conditions. The 2010 drought cut Russia’s wheat production by a quarter, which drove up the prices of wheat and, in turn, of cattle. This fed through to a spike in leather prices that hit Puma’s profits.

Information wranglers

Price index food pricesThe danger is that the explosion of data means that companies have to make sure they are picking the most relevant information. As CIMA’s Ross says: “This data is just that: data. It tells us what people have done, but not why, and it needs interpretation to add insight.”

She sees a need for “information wranglers”, such as management accountants who can analyse data for users: “They need to be able to draw conclusions from uncertain or incomplete data, and about the credibility of the data.”

For example, online retail searches may give more up-to-date information than retail sales data, but they could be distorted by cash-poor window shoppers who cannot afford to turn searches into sales.

The Bank of England highlights other drawbacks, such as the short back run of data and the fact that internet use is heavily dominated by younger, wealthier sections of the population. And even BPP does not cover the whole economy; it is focused on goods and services bought online.

However, Rigobon says BPP is working on producing indices for inflation rates and capital gains for residential and commercial property. His team is collecting data from almost 200,000 cities around the world but a launch may be two years away.

“These inflation rates we are reporting now took four years to produce,” he says. “It is not only collecting data but working out how to organise it so it can be used.”

He hopes that such data will be complementary to official figures, “but I think statistics offices need to change – otherwise, they will become dinosaurs”.

The famous economist Yogi Berra once said: “It’s tough to make predictions, especially about the future.” The challenge for economists, statisticians and financial directors is finding a way to harness the acres of data on the internet to ensure they can predict the present. ?


Was this article helpful?

Leave a Reply

Subscribe to get your daily business insights