Imagine a vast interconnected web of 300 million firms, connected through an estimated 13 billion supply chain links. These are the businesses that produce the goods and services that make up our global economy. In the UK alone, there are around 5.6 million registered businesses with millions of suppliers. This web of complex interconnections has been almost impossible to untangle on a company-by-company level—until now.
According to University of Cambridge Professor of Macroeconomics Vasco M. Carvalho—who operates at the frontier of economics as Director of the Janeway Institute—this blind spot has left our society ill-prepared to make rapid and well-informed decisions about how to respond to economic shocks. During the COVID-19 pandemic, for example, this led to shortages in critical medical supplies.
To address this gap, Professor Carvalho and his team have been working on a pioneering approach to help economists make sense of the web of data relating to businesses in the global economy by drilling down into financial transactions that can help to show company and industry-wide trends. In some countries, understanding firm-level interactions has been more straightforward because there are VAT records showing the value of goods and services traded between companies. Professor Carvalho refers to this as producing the “gold standard” of data. However, in countries like the UK and US, this gold standard doesn’t exist.
This means that Professor Carvalho and his team have had to take a different approach. Instead of VAT records from the tax authorities, he and colleagues from The Alan Turing Institute—where he is also a fellow—have been working together with the Office of National Statistics (ONS) to draw on data from the UK’s electronic bank payment systems network to generate the information they need to map economic trends.
Bank payment systems are the behind-the-scenes processes that allow money to be transferred between bank accounts in a series of steps. These digital systems now enable everyday financial activities like buying groceries, withdrawing cash, paying for a house deposit, receiving salaries or benefits, paying by direct debit, or transferring money via smartphone. Last year, these regulated processes handled over 22 billion transactions, amounting to around £75 trillion, according to the UK’s Payment Services Regulator.
One of the businesses that hosts these transactions on behalf of the banking industry is called Vocalink—a MasterCard company that processes all of the UK’s real-time payments, settlement, and direct debit systems, as well as networks of over 47,000 ATMs. They also deal with over 90% of salaries, more than 70% of household bills, and 98% of state benefits in the UK.
A Game Changer for Economists
Controlled by the banks, this huge ledger of financial transactions is a potential treasure trove of data for economists. However, until recently, it has not been available for anything other than fraud and security surveillance due to privacy concerns. This began to change during the COVID-19 pandemic when economists like Professor Carvalho were asked to look at innovative ways to access supply chain data in real time to help avert shortages in vital supplies. They turned to the banking payment system for some quick answers, and what they found was potentially game-changing.
“As we move away from a cash economy, there are all these immense ledgers that we can use for extremely valuable public statistics in real time with plentiful data,” explains Professor Carvalho. “Democratic society relies on public statistics. Banks have been generating these massive amounts of data as a byproduct of the financial system. And all this data can serve a broader goal of producing national statistics in a different way.”
Mathematicians, data scientists, statisticians, and economists came together with the ONS to work out how they could use machine learning to transform these huge ledgers of anonymized data into something that could map financial transactions on an aggregated sector level so it could be used by researchers. The findings on a sample of the data were surprisingly accurate:
“It turns out that there is comparable information to the gold standard type data that does not exist in the UK,” says Professor Carvalho. “It surprised me the extent to which this data looks like the gold standard we were trying to replicate. This data simply does not exist otherwise.”
This new approach has potentially wide-ranging implications for economic research, helping to map where UK goods and services are imported from and exported to and showing how different regions across the UK do business with each other. In turn, this can help to inform regional industrial policies and major infrastructure decisions across the country.
Confidentiality Concerns
Customer confidentiality and privacy have been crucial. Protocols for handling the data are incredibly strict: None of the raw bank-owned data ever leaves Vocalink’s systems. Only a limited number of ONS researchers have clearance to access the data, and findings from the data are only shared publicly in an aggregated way—for example, showing industry trends rather than company-by-company data. Even with these assurances, the banks—the data owners—have had concerns about the process.
“The biggest barrier by far has been that the data owner does not see any benefit,” says Professor Carvalho. “And in fact, they only see downsides in anyone touching their data—even a relatively neutral body like the ONS. At the beginning of the project, there were hopes that there would be more access and more freedom of working with individual data. We could have moved much faster and with much fewer error rates if we were allowed to know more stuff.”
Professor Carvalho says some external observers have expressed a dislike for this type of research, regarding it as an example of Orwellian oversight gone too far. But his argument is that the data is being gathered anyway, so why shouldn’t it be used for society’s benefit?
“There’s all this data and these ledgers that could revolutionize the way we as a society compile information about how we’re doing socially and economically,” he says. “And it’s our data anyway because we’re generating it. But it’s in the hands of the banking system. There needs to be a public conversation about what could prevent these barriers—and the effective use of these data for the global public good.”
Professor Carvalho’s research has already generated a lot of interest, both from the UK and overseas. The pilot ONS collaboration has been extended, and there are plans to make the team’s research methodology public so it can be replicated in other countries and using other payment systems too.
In the meantime, Professor Carvalho believes this type of research could help to revolutionize an antiquated national statistical system in the UK that has limited resources to produce the vital public statistics that underpin our democracy. He suggests a National Data Library in the UK could potentially act as a kind of ombudsman that could help to bridge the gap between sensitive data sources and individual researchers—providing a highly valuable resource for researchers while safeguarding data privacy.
Written by Vicky Anning
