Breaking Down Data Barriers – A Smart Data Foundry Case Study

Income Volatility in the UK Next steps References Related Items

In our modern data-driven world, new and innovative data sources are increasingly becoming available to accredited researchers and decision-makers to build evidence and drive positive change. From the administrative data created when people interact with public services to the smart data generated through engagements with digital systems, devices and sensors, this data provides an incomparable wealth of information that could be used to help people. However, we need to ensure that these new data sources are used for the public good and respect the privacy of our fellow citizens
At Smart Data Foundry we have been working with large private sector financial datasets and adapting them for scientific use since 2020, with our data partnerships underpinned by a Trusted Research Environment (TRE) and access under GDPR Legitimate Interest. Our deidentified, granular, and longitudinal datasets are regularly updated to facilitate timely and detailed research into peoples' or businesses' changing economic circumstances in response to economic shocks such as COVID-19, the cost-of-living crisis, and the current challenging economic environment. This gives the potential opportunity for further analysis of how this data could correlate with other ancillary sources in ways that could not be done before, such as linking the data to climate or health data.
Smart Data Foundry have partnered with the Joseph Rowntree Foundation to create a data infrastructure that looks closely at the scale of income volatility—and thus a large component of economic insecurity—across the UK. This infrastructure consists of a freely available aggregated data dashboard, and individual-level data available in a Trusted Research Environment for those with further questions.

Income Volatility in the UK

Initial Findings

The scale of the gig economy in the UK is a highly contentious issue, with varying estimates of its size based upon differing definitions from government and campaigning bodies. The UK government's working definition covers 'participants who trade their time and skills through online platforms (websites or apps), providing a service to a third party as a form of paid employment'. Based off this definition, the CIPD estimates that only 1.4% of the workforce, or roughly half a million people are employed in the gig economy. However, this greatly differs to TUC research which estimated roughly 14.7% (4.9m) of working people had taken work within the gig economy, or even government research via the NatCen Panel in 2017 that estimated the proportion at 4.4% (2.8m).

Given that concerns about the gig economy stem from a perception of it having low pay and precarious working conditions, we propose a series of metrics that instead examine income volatility and variance across Great Britain. These were developed from NatWest Group Banking data, who supply us with de-identified aggregate transaction data for 1.2m customers dating back to 2019 and updated weekly. This infrastructure is open for researchers to access, with individual level data open to researchers with specific project use cases.

What does this data show?

Four subplots showing the variation in male and female payments interval through time.

Here we have a grid of subplots showing the variation in payment interval through time. The four subplots cover fortnightly, monthly, other and weekly payment intervals. Each plot has two lines, one showing the percentage of females receiving a given payment interval female and the males.

The most common payment interval is monthly with over 60% of NatWest customers receiving this. The least common interval is weekly payments, but fortnightly and other intervals are similar at 10-20% of customers. It is more common for females to be paid monthly, and males are most likely to be paid weekly or fortnightly. The number of people paid weekly and fortnightly has declined slightly through time. The most striking features from the plots are the interplay between monthly and other payment intervals. These occur in early 2022 and 2023 and are likely indicative of the early pay in before Christmas and a long gap until the end of January.

These plots are based on salary classifications of NatWest customer data using a 13-week rolling window.

Two maps of Great Britain overlain with postcode area boundaries showing low or medium and high-income variability.

This plot shows two maps of Great Britain overlain with postcode area boundaries. The postcode areas are coloured by the percentage of people who have low or medium and high-income variability. Low-income variability is defined as those with less than 5% variation, medium and high are 5% or greater variation.

Areas with the highest variation in salary are located in rural areas like south Wales and the Scottish islands. These could relate to jobs with shift work (e.g. manufacturing) or production-based rural employment. The lowest variation is in areas like SE England and SE Scotland. These may have above average concentrations of commuters with stable employment contracts.

These maps are based on salary classifications of NatWest customer data using a 13-week rolling window. Income variability is calculated using the coefficient of variation.

Next steps

Smart Data Foundry exists for researchers at the University of Edinburgh and beyond to be able to access this data, and ask wicked questions of it. The dashboard described above can be accessed through signing up and completing your profile for our portal, and the Trusted Research Environment can be accessed through completing an application form accessible here. Further advice on applying can be found within our portal.

For those interested in accessing our other data assets for research, please contact research@smartdatafoundry.com to discuss your proposal. We look forward to hearing from you.