Over 1.2 m people lost their jobs during the pandemic, with bartenders, pub staff, baristas in hospitality and thousands of high street workers affected. Official figures show 2.6m claimants in Feb 2021, a rise of 1.2m since Covid went nuclear on us.
This figure, frighteningly high as it is, does not show the full depth of Job Armaggedon – just in retail, job losses are still running at 30,000 per month. We all know many close friends and family who lost jobs but are not claiming unemployment, hoping the situation is temporary, so they are not officially counted.
It is also estimated that 4.7 m UK jobs are currently being protected by furlough payments, jobs that may be lost once government support is over.
The brunt of job losses, as always in a crisis, has fallen on the young with a disproportional number of jobs cuts falling on 18-24 year olds, mainly in high street and hospitality, coffee shops and the pubs sectors.
To get a grip on the real figures behind the headline ONS numbers, Newspeak House and Campaign Lab called a hackathon to get a deep dive into the jobs data, to reveal the true depth of suffering and surface spots where government support will be specially needed. To improve data transparency and to share facts with everyone, Campaign Lab data science detective squad gathered on a Spring Saturday in March to shed light on the UK jobs story.
Timing of our hackathon was not coincidental. In normal times, local citizens have no real influence on government actions with the UK 2 party system precluding any effective democracy in action. But on May 6th there are local elections, for 145 county and district councils, as well as 13 directly elected mayors in England. Importantly, 39 Police and Crime Commissioners in England and Wales will be up for election, a hot political spot in a time of rising tension on the streets.
All residents can vote and impact what is going on around them by influencing initiatives to help jobless people to get back into work. We are all in it together, but we need data to bounce off for sensible, practical proposals.
Squad Report from Business Deaths Team
The hackathon was an all-day affair, via Zoom, WhatsApp, Docu+ and GitHub.
We first gathered on Zoom, ran thru’ all the challenges, allocated ourselves into squads per challenge depending on skills and interests, and dug in for the day in breakout rooms.
Zoom chat was used to hold all the relevant links to datasets, data visualisation templates sources and images. Zoom dropped out on us on a few occasions, kicking out one or two people at a time, it’s not really a tool intended to keep the session open for the whole day…
To mitigate, we used WhatsApp as a back channel. Next time we will use Signal for backchat as a number of squad members voiced dislike of sharing our work with Facebook, WhatsApp’s mother company, a sentiment we agree with.
New Data Sets
My squad focused on a closer look at Business Deaths (company closures) as they tend to tell a more accurate picture on job losses than the actual employment data.
We were able to use a new, experimental dataset from ONS, provided in quarterly snapshots. It is experimental, as it has been designed for speed rather than accuracy but has offered an up-to-date news on the pace of Jobmaggedon that the pandemic is leaving it its wake.
Mark Williams, (the kind soul looking after this data for the ONS) provided tireless data support for queries, you can reach him at IDBRDas@ONS.gov.uk – worth a shot particularly if you need data from the very last quarter, which he may be still working on.
The situation with business deaths is a lot worse than unemployment figures would have led us to believe. The closures of businesses in Q4 in 2020 was 37% higher than Q4 2019, a record number after flatlining on a relatively low level in 2017-2019. It does not bode very well for a speedy recovery in jobs, coming on top of news that 4.7 mln people were still on furlough in Jan 2021. Forecasters expect over 20% of those jobs held in artificial limbo by furlough to be axed at the end of the scheme.
In absolute terms London and the South East have suffered the most, but in percentage terms the catastrophe has affected most regions to a similar degree. The highest number of business deaths was followed by a high number of business openings, but the net figure was still a 17% drop in the net number of businesses operations, with the added negative that new businesses registered significantly less employees than the year before, dropping from 3.4 to 2.7 on the average.
A Large proportion of newly registered units were by one-man bands or sole traders. It reflects a reactive action, when people lose jobs and set up a new entity by themselves or with a family member to kick start activity – often with no income stream to back up the registration.
What we also noted, taking data directly from Companies House datasets, was that pub closures are still running high, with owners and leaseholders giving up the ghost during Covid, a situation worsened by high business rates already affecting the industry before this andemic.
Data is the new oil but comes with challenges
We had a number of different issues with the data, not least that the periods of gathering of the data don’t match in all categories, sometimes data range matches but sometimes they don’t or are not up to date on one or two business categories, rendering the whole set unusable.
With Companies House the issue is the tagging of what type of business is in which code – some of the groupings are peculiar to say the least. Often the datasets are regional which means that the visualisation on maps is challenging, as UK regions are not standardised – in some sets UK regions are more granular than others.
That requires a translation from one set to another, limiting the usability for mapping. USA data sets are easier to use in the sense that they are based on states with the same boundaries and standard definitions amongst many datasets. The UK would benefit from a standardised geographical agreement regarding the base UK regions which would make the comparability of the existing sets massively better and save time and effort on conversions. It seems easier for now to stick to datasets that have postcodes, at least you will be able to output them in an accurate map display.
Our analysis will continue over the next few weeks, feeding the findings to Local Election Campaigners around the country and providing focus on the areas that need special love from post-Covid money bags. Investing in London would clearly help the biggest amount of businesses as the closures in the capital were the highest, but it is clear that all the country has suffered job losses and business closures to a similar degree. We really are in it together!
Digital citizenship means many things but being critical of what the official jobs data is showing is certainly a good place to start. Interrogate and examine what is being published and compare with other datasets to get to the fact. It allows us to step up lobbying for more help for London and the South East as clearly the business closures were heaviest in the capital, affecting the most people in absolute terms.
In the next instalment of the blog we will cover what other squads were doing, examples of Facebook Messenger bots to improve comms for the Local Elections and Tech Tools Grotto for sharing tech solutions for campaigning.
Email firstname.lastname@example.org to join the next session, learn about data wrangling, maps and campaigning
A big Thank You to
Nathan Lewis (BBC)
Dominic Norton (Peabody Trust)
Hannah O’Rourke (Newspeak House, Campaign Lab)
James Moulding (Newspeak House, Campaign Lab)
Maciek Ziarkowski (Colouring London team for geo advice)
Edward Saperia for lending us DocuPlus tool
Our families for feeding us and not complaining during this long weekend!