Adult social care labour supply: pay increase impact assessment

Technical Appendix A1. Data

Adult Social Care Workforce Data Set

The Adult Social Care Workforce Data Set (ASC-WDS) is the main source of ASC staff data in England. Information is provided by ASC employers on both establishment characteristics (e.g., type of service provided, sector, establishment size, count of employees and job roles, starters, leavers and vacancies, etc.) and care staff (e.g., age, gender, nationality, qualifications, training, hourly pay, job role, contract type, etc.). Public employers provide data on a mandatory basis, while independent employers provide data voluntarily, being incentivised through access to workforce development grants. All data in the ASC-WDS is updated or confirmed to be up to date within the last two years, with about 80% of employers having updated their data in the past six months. ASC-WDS data is validated at source and undergoing rigorous data quality checks (Skills for Care, 2023).

We included seven annual cuts from the dataset (Oct 2016 to Oct 2022), matched at individual level. We excluded all establishments with records not updated withing the last six months and establishments with missing IDs for more than 25% of their care staff. We included establishments offering ASC services to adults (i.e., people aged 18 and over), and being owned by either public (i.e., statutory local authority), private (i.e., for-profit) or voluntary (i.e., not-for-profit) sector care providers. Employees were included in the sample if aged between 16 and 59 (i.e., employees close to state retirement age excluded). We excluded observations for care staff without a unique ID (as these could not be traced over time; 7%), for those who erroneously had multiple entries per year with the same establishment (1%), and for staff with two or more jobs in any year (6%).

Care establishments and staff have each unique and permanent IDs, allowing staff to be traced over time and identifying job transitions/separations. Following Vadean et al. (2024), job transitions inside ASC were identified as a dummy variable equal to ‘0’ if the employee was still with the same employer one year later (t + 1) and equal to ‘1’ if the employee could be identified as working for another ASC employer in the sample at t + 1. Overall job separations were defined as a dummy variable equal to ‘0’ if the employee was still with the same employer one year later (t + 1) and equal to ‘1’ if either: a) the employee could be identified as working for another ASC employer in the sample at t + 1; or b) the employee left the sample at t + 1, but their employer at time t was still in the sample. For a small number of employees information was missing at t+1, but we could use the information from a subsequent year to identify the job separation status. Employees for whom the job separation status could not be identified because both they and their initial employer dropped from the sample in all subsequent years, were excluded from the analysis (about 14%).

The final sample consisted of 903,690 observations (job-spell-years) of 395,281 care staff employed in 13,291 care establishments; see Table A1. A large number of job spells in our sample ended with a job separation: 31%. Out of these job separations, we could identify about 30% to be transitions to other ASC employers in the ASC-WDS sample. As the ASC-WDS only covers about half of the ASC market in England, the remaining job separations were to destinations outside the ASC-WDS sample, and not identifiable whether inside or outside the ASC sector.

More accurate information on the job transitions between ASC jobs was obtained from a question related to the source of new recruits, this showing that about 62% of recruits were previously employed in other ASC jobs.

From the job spells in our sample, about 52% were in adult residential care (with or without nursing), 38% in adult home care, and 10% in other adult community care. Moreover, a large majority of job spells were in privately owned care establishments (72%) and smaller shares (14% each) in the public and voluntary sector.

The UK minimum wage policy and in particular the introduction of the National Living Wage (NLW) in April 2016 led to a significant increase of wages for many ASC employees. As illustrated in Figure A1 (b), when inflated by CPI to 2023 prices, the wage distribution in 2021 (the last year in our panel) is significantly above those from previous years. This shows that wages in ASC have increased more that consumer prices. Figure A1 (c) shows that a more suitable inflation factor for wages in this study is the percentage change in NLW, leading to a better overlap of the wage distributions between the years included in the sample.

Annual Survey of Hours and Earnings: wage distribution

Information on the distribution of wages in the Scottish social care sector is taken from the Annual Survey of Hours and Earnings (ASHE), the most comprehensive source of information on the structure and distribution of earnings in the UK. Table A2 presents the deciles and quartiles of the distribution of hourly wages (excluding overtime) in Scotland in 2023 for: a) residential care employees (SIC Q87) and b) employees involved in social work activities without accommodation (SIC Q88). We combined these in an overall distribution for the ASC sector using as weights employee headcount in the two care settings from the Scottish Social Workforce Data (Scottish Social Services Council, 2023). The ASHE wage data includes wages from employees in children social care services and is, therefore, an approximation of the wage distribution in ASC.

When comparing the wage distribution in ASC in Scotland in 2023 to that of the analysed ASC-WDS sample, we note that ASC wages are higher in Scotland. There are likely two reasons for this difference: a) the introduction in April 2023 of a sector minimum wage of £10.90 (which increased wages of the lowest paid ASC workers above the level of their counterparts in England), and b) the substantially higher share of ASC staff employed by public and voluntary care providers (see Figure 5), which are known to pay higher wages.

To bring the wage distribution of our analysed sample closer to that of the ASC workforce in Scotland, we applied a scaling factor equal to the percentage difference between the Scottish ASC minimum wage (i.e., £10.90) and the NLW (i.e., £10.42). As shown in the last row of Table A2, this brought the lower part of the two wage distributions (below the 40^th percentile) quite close to each other (within a 5% range), and with the 20^th percentiles almost coinciding. Wages above the 50^th percentile are more different (above 10%) and most likely due to the differences in ASC sectoral split between Scotland and England.

Office for National Statistics and Annual Population Survey labour market statistics

To compare the labour market in Scotland to the labour market in England (and parts of England) we used Office for National Statistics (ONS) data on job density (i.e., the number of jobs divided by the resident population aged 16-64) and Annual Population Survey (APS) data on economic inactivity (i.e., people not in employment who have not been seeking work within the last 4 weeks and/or are unable to start work within the next 2 weeks). The data for Scotland and English regions for 2021 is presented in Table A3.

Scottish Social Service Sector (SSSC) report on workforce data

All employee headcount and registered services (establishment count) data are obtained from the Scottish Social Services Council’s 2022 Annual Report on Workforce Data.