Adult social care labour supply: pay increase impact assessment

Footnotes

1. https://www.gov.scot/news/pay-uplift-commitment-for-social-care-and-childcare-staff/

2. Employee headcounts are obtained from the Scottish Social Services Council (SSSC) Workforce Data and are as at December of each year. Residential care refers to care in a care home for adults. Non-residential care comprises the Care at Home, Adult Day Care, and Fieldwork Service (Adults).

3. Hourly pay data are obtained from the ONS Annual Survey of Hours and Earnings (ASHE). Residential care refers to Residential Care Activities (SIC: 87) and Non-residential care refers to Social Work Activities Without Accommodation (SIC:88). Real hourly pay is derived by deflating hourly pay (excluding overtime) for all workers by CPIH for 2022.

4. We did not provide a sectoral split by care setting between the two countries due to the lack of information on the split between private and voluntary providers by care setting in Skills for Care data for England. We also need to note that the figures only provide a rough comparison between ASC markets of the two countries due to differences in reporting. For example, the Scottish data reports headcounts while the English data reports filled job posts, which may differ from headcounts when individuals occupy multiple posts. We also excluded from the data for England 130,000 filled post by Personal Assistants (i.e., care workers employed directly by cared for people), as these have no equivalent in social care workforce data in Scotland.

5. We take the social care wage to be the employment-weighted average wage levels in SIC Q87 (Residential care activities) and Q88 (Social work activities without accommodation). As these industrial classifications include childrens’ social services, they are rather an approximation of the ASC sector we analyse.

6. The vector of variables includes time-variant, time-invariant (e.g., gender) as well as time-dependent variables (e.g., age and tenure).

7. We also run estimations with log transformed wages, but these did not fit data as good as the model with 3-degree polynomial of hourly wages.