Job vacancies are an early indicator of economic activity. When a firm’s demand for labour increases, it may post new vacancies some time before these vacancies are actually filled; conversely, when labour demand falls, a firm may stop recruitment efforts before an effect on employment is felt. Tracking the evolution of vacancies allows us to see where labour demand has fallen most during the pandemic, and where it is tentatively picking up again, much faster than official employment statistics allow.
We present real-time information from a daily census of all job adverts posted on Find a Job (FAJ), a recruitment website maintained by the Department for Work and Pensions (DWP). Details about the data source are given in the appendix. We aim to provide a detailed and up-to-date picture of the current state of labour demand across the UK economy, as it emerges from lockdown.
We compare the number of new vacancies posted each day (using a backwards-looking 7-day moving average) in 2020 to those posted on the same day in 2019. Figure 1 shows the number of daily vacancy postings in 2019 and 2020. The remaining figures show new daily vacancy postings in 2020, split by occupation, job characteristics and geographic areas. On each graph, we also highlight key dates on which announcements of the easing of lockdown were made: 10th May when the first restrictions easing was announced, 9th June when it was announced non-essential retail would reopen the following week, and 23rd June when it was announced the 2-metre rule would be relaxed and much of the hospitality sector could reopen on 4th July.
We will continue to update these charts on a weekly basis, as lockdown eases further and the number of vacancies continues to evolve. A more detailed analysis of vacancy postings between the end of March and early May 2020 can be found in a previous IFS briefing note.
This analysis uses data on job adverts posted on Find a Job (FAJ), a recruitment website maintained by the Department for Work and Pensions (DWP). To advertise a job on the website, a firm must create a profile and enter key information on each job opportunity, including job title, location and the closing date. Unlike other recruitment websites, no attempt is made to ‘scrape’ job adverts from other websites and therefore the FAJ data should be interpreted as a subset of all job opportunities. As of 29 June 2020, there were 70,290 vacancy postings listed on FAJ.This is 22% of the 326,759 vacancy postings listed on Indeed.com, a recruitment website that aggregates vacancies across company websites and job boards and aims to include all jobs listed online. FAJ is likely to over-represent public sector vacancies, as these are required to be listed on FAJ.
Data on job adverts posted during 2019 were made available by the DWP following a Freedom of Information request. These data contain information on all job adverts that were posted on the website between its launch in May 2018 and July 2019. Data on job adverts posted during 2020 were obtained using a web-scraping algorithm that records information on all job adverts that have been posted on the website in the last 24 hours. This algorithm has been run on a daily basis since 23 April 2020 (at approximately 10am each day). This daily time series is combined with data on all job adverts that were ‘live’ on the website on 23 April 2020, obtained using a modification of the web-scraping algorithm. This means that adverts posted in 2020 before 23 April that were removed from the website before that date are missing from the data. This would affect job adverts with an advertised closing date before 23 April 2020 and those that were manually removed from the website by the firm – for example, because the job position had been filled. Although this risks systematically understating job vacancies in early April 2020 relative to 2019, it appears that the overwhelming majority of job adverts record a closing date that is exactly one month after the date they were posted. This gives us confidence that the 2020 data are representative of adverts posted on the website from 23 March onwards.
Information on when job adverts were removed from the website is missing from both the 2019 and 2020 data. We are therefore unable to construct an accurate measure of the stock of vacancies on the website at any given time and so we confine our analysis to the flow of new postings added to the website on a particular date. Each job advert contains a unique identifier, which we use to identify and remove duplicate job postings.
We have supplemented the information available in the 2019 and 2020 data with occupation codes and various geographic identifiers. Occupation codes are derived from a cleaned version of the job title using the ‘SOC search’ operation available at http://api.lmiforall.org.uk. Geographic identifiers are derived for the subset of vacancies that contain a complete postcode in the location field (approximately 63% of adverts in 2019 and 68% of adverts in 2020), using the ‘postcodes’ operation available at https://findthatpostcode.uk/#api.