CBA taps Google search data to forecast household spending

The Commonwealth Bank of Australia is using Google search activity to help it forecast household spending trends.

The bank’s new Household Spending Intentions Series combines its own data – which is “one of Australia’s largest financial data sources” the bank said – with relevant, publicly available search data from Google.

“Employing near real-time spending readings from CBA’s household transactions data and combining them with relevant search information from Google Trends was used to map the data results on consumer spending,” the bank said.

“These searches provide insights into what consumers are doing/researching on the Internet and what their spending intentions are,” it added.

The approach to forecasting uses spending data derived from a sample of more than 2.5 million households who are CBA customers. The bank made decisions about what Google search terms to investigate that “pass the pub test as well as meet the more rigorous statistical standards”.

The bank’s economists used Google Correlate to match the economic data to search patterns.

Not all patterns were relevant, the bank states in the report, for example Google Correlate found that searches for ‘green bean casseroles’ were strongly correlated with house sales with a three month lead.

“We could posit that those thinking about buying a house are busy saving the deposit and looking for inspiration for a cheap meal. But in the end correlation does not mean causation. We should keep looking for more relevant search terms,” the report says.

“Instead, combining CBA loan application data with relevant Google searches generates a home purchase intentions indicator that leads the actual sales data by four months. The statistical metrics are quite good – the correlation coefficient comes in at 0.83,” the report explains.

The Household Spending Intentions reports are intended for use by small and large businesses, policy makers and investors. They will be published monthly to help Australians “stay ahead of emerging economic trends” said CBA chief economist Michael Blythe.

“Households are the dominant part of the economy and drive much of its activity and volatility – anything that gives an early warning sign on how the consumer is behaving is very useful,” Blythe added.

The reports focus on seven areas of consumer spending: retail, travel, health and fitness, entertainment, education and motor vehicles. These areas cover about 55 per cent of total consumer spend.

Combining the bank’s data with Google’s enable’s CBA to predict purchase activity up to four months in advance.

In this first report, the bank found that overall, household spending intentions remains negative.

“Weakness persists in the dominant retail and motor vehicle sectors. But other sectors appear to be bottoming out – including travel and entertainment, or picking up – including health and education. The turn in home buying intentions is particularly noticeable in recent months,” Blythe said.