PublicTechnology sits down with ONS statisticians to learn more about the organisation’s Faster Indicators programme
It’s said that you can tell the mood of the nation – and the health of its citizens’ purses – by monitoring sales of champagne.
It probably comes as a surprise to none of us that those sales were markedly down in 2020, when there seemed so little to celebrate. In fact, the global industry reported a decline in sales of approximately $2bn – one of the biggest drops since the financial crisis of 2008.
Of course, governments traditionally place more stock in the GDP calculations of bodies like the Office for National Statistics (ONS) than in the sales reports of Dom Perignon.
But a new initiative from the ONS seeks to harness the power of some more unconventional economic indicators.
“We tried to look at how we could use new big and innovative data sets to tell us things that were actually advanced or more granular than our normal run of official statistics,” says Rob Kent-Smith, deputy director for national accounts, who helps lead the Faster Indicators team.
“When we have data from third parties, we work very hard to understand what the biases are, and to communicate that in a way that people can understand.”
Louisa Nolan, ONS
“When the pandemic hit back in March last year, it became clear that some of our regular official statistics were not going to be quite quick or timely enough to do what we needed them to do and for the decisions that needed to be taken in that context.”
While traditional methods of calculating the national GDP often publish months in arrears, the Faster Indicators team turned to a range of new data sets – including VAT returns, shipping movements, and road traffic sensors across the country – to paint a more responsive image of the economy.
They even checked publicly-available CCTV images of high streets in order to measure foot traffic during lockdowns.
Louisa Nolan, chief data scientist, says: “We wanted something that was as close to real time as possible, for the coverage to be as good as possible – or at least for us to understand if it wasn’t – and then we wanted something that was an economic concept.”
Like the aforementioned champagne sales, the number of lorries on the roads or ships in UK ports can shed light on the movement of goods into, out of, and around the UK – and therefore give an idea of economic performance.
“You would expect the case to be that if people are going to work, and goods are moving around the UK, you’d see high levels of traffic,” explains Nolan.
While the Netherlands have published similar work monitoring road traffic as a means by which to understand economic movement, the ONS claims to be the first national statistical institute to publish something like the Faster Indicators on a regular basis.
And by the time the pandemic hit, they had already been publishing Faster Indicators on a monthly schedule – but this output was soon to ramp up.
“We very quickly set about moving the monthly Faster Indicators suite to actually be in a weekly release, and really quite quickly expanded the range of indicators that were available,” Kent-Smith says.
The team also worked to set up a fortnightly business survey, and the range of indicators now available is a sign of how broad the interest in this kind of ‘real-time’ insight is – even outside of government.
“Decision-makers in government at various levels will have an interest in this data to understand and make their own policy, but also it’s widely reported in the media, and we find that businesses are using this data to see how they’re comparing to other people in similar fields,” Kent-Smith says. “So, the aim is to make the usership quite broad and wide.”
Unlike the official GDP statistics, the ONS does not have control over the data sources used to produce the Faster Indicators. The data instead comes from private providers, or through publicly available websites.
This means that they have to be cognisant of potential bias in the data.
“When we have data from third parties, we have what we get. So, we work very hard to understand what the biases are, and to communicate that in a way that people can understand,” says Nolan. “And the narrative around what the data is, is a really important part of that publication, particularly when you’re doing something which you know might not be representative, or might have some other kind of bias or uncertainty in it.”
Alongside the risk of bias in the data sets, there is also the chance that the method through which the data is being reported, collected or modelled could change in the future, and this would have an impact on its usability.
“At the moment that’s not happened,” Kent-Smith says, though he acknowledged that, “it could be a challenge on the horizon.
“I think it really involves working very, very closely with our providers in a partnership, rather than as a kind of customer-supplier, but actually real partnerships to build these types of things together.”
Another challenge posed by the nature of the Faster Indicators is the danger of over-interpretation.
According to Kent-Smith, while people are keen to see the insights that can be garnered from the indicators, they must bear in mind that these are indicators only, and cannot be accepted at face value as readily as the ONS’s usual statistical output.
The Faster Indicators are useful in illustrating wider trends rather than providing hard-and-fast answers, which is what the official GDP statistics are for – and Kent-Smith says it’s still “too early” to know for sure “how robust some of the indicators are.” Though he stressed that the team had done “a good job” of “putting some of those caveats around it.”
With all this in mind, how trustworthy is the data?
“That depends what you mean by trustworthiness,” says Nolan. “I think we’ve worked hard to make sure that our users trust our outputs – whether the normal measurements or these newer, faster indicators – and that they trust them for the things that they do, but not for the things that they don’t do.
“It’s not as simple as saying ‘this is an accurate measure of everything to do with GDP’… we’re saying: ‘you can use this as an indication of how things are going, but these are some of the shortcomings’. And once you accept that, that’s still useful, and can still help inform your decisions.”
The willingness of users to adapt has been something of particular note throughout the process of building and maintaining the Faster Indicators output.
This project has not only been useful as a way to develop the technical tools necessary to manage this kind of data – or even to do so on such a quick timescale – it has also been impactful as a way of highlighting the flexibility of end users.
“That capability building is a really important part of it,” Nolan says. “Learning that you can bring your users along with you to understand what you are doing and understand what the benefits and limits are.”
“When the pandemic hit back in March last year, it became clear that some of our regular official statistics were not going to be quite quick or timely enough to do what we needed them to do and for the decisions that needed to be taken.”
Rob Kent-Smith, ONS
It has also highlighted the possibilities of what the team can build – one reason why Nolan is so keen for the ONS Data Science Campus to be a space where experiments like this can be encouraged, regardless of outcome.
After all, she says, having the “space to fail” is “important for innovation”.
The Data Science Campus, which started as a team of six, has since grown to house eighty with a remit to build capability across the public sector, and is continuing to push ahead with new initiatives.
One such initiative includes experimenting with synthetic data sets, which Nolan hopes will “democratise data” by providing private, shareable data that organisations can experiment with.
“They’re safe to share, but are good enough that you could maybe build your code, build your pipelines, explore the data a bit, before you get access to the real data – whichh really opens up and democratises data,” she says. “We’ve started quite a major programme of work on that.”
There is still more to come from the Faster Indicators team, too.
So far, the primary focus has been on UK-level data – though there are some regional breakdowns within the releases. As focus shifts to the government’s Levelling Up agenda, though, this kind of insight can be used at a regional level.
“We will really think about how we can use some of these data sources to tell a more granular picture around what’s happening in different parts of the country,” says Kent-Smith.
More responsive statistics are one key outcome of the Faster Indicators project, but it has also had an effect on the overall way data is approached and accessed.
“I think the biggest step in wrapping this together with the ONS brand, with a regular publication, is that this type of work has now entered the mainstream and isn’t something that happens quietly in an academic community,” Kent-Smith says.
Not only is it more accessible for users, it can serve as a guide for future ONS projects too.
“It sets the direction of additional things that we can do as the public sector and as ONS,” Nolan says. “I think it enhances the way we deal with data.”
Although the possibilities for this kind of ‘real-time’ data now seem endless, Kent-Smith says that “there will always be a place for the official statistics”.
“Please,” Nolan adds, “do complete your census forms!”