Performance Tracker 2022/23: Spring update - Methodology

Public services spending

To estimate the real cost of public spending, we deflate government spending figures using the GDP deflators published in the November 2022 autumn statement, available at www.gov.uk/government/statistics/gdp-deflators-at-market-prices-and-money-gdp-november-2022-autumn-statement. To better reflect the underlying inflation conditions present in 2020/21, we estimate our own figures by generating a mid-point that averages across values from 2019/20 and 2021/22. We deflate spending figures in our financial analysis across all Performance Tracker chapters.

Average annual increases in spending between 2021/22 and 2024/25 relative to demand under different inflation scenarios (Figure 0.1)

For the nine services we cover in this report, we project how much money the public sector would have to spend to meet demand. To estimate the cost of doing this, we project growth in underlying demand for each service as described in the Methodology chapter of Performance Tracker 2022.

For each service we also project how spending is likely to evolve over the course of the spending review (up to 2024/25). The 2021 spending review did not provide budgets for particular public services, only government departments (with the exception of schools and the NHS, which have their own budget lines).

For each service, we take the most relevant department’s settlement, implicitly assuming that all spending within those budgets will increase at the same rate. This means that we assume that spending on GPs and hospitals will increase in line with NHS spending, spending on courts and prisons will increase in line with MoJ spending, spending on the police will increase in line with Home Office spending, and schools’ spending will increase in line with the specific school funding line in the spending review. For the three local government services (adult social care, children’s social care and neighbourhood services), we take the government’s projections for local authority spending power, which  incorporate changes to grants and assumed increases in local taxes (council tax and business rates).

To compare the generosity of the cash-terms settlements set out in the 2021 spending review over time, we deflate these numbers using three iterations of the GDP deflator, a measure of economy-wide inflation that is widely used – including by the government – to assess the real-terms generosity of public service spending plans. We take the GDP deflator from the October 2021 spending review itself and then the GDP deflator at the spring statement in March 2022. Both of these come from the Office for Budget Responsibility (OBR).

The third column in the series looks at the effects on spending envelopes of the autumn statement from November 2022. That fiscal event provided some services – the NHS, schools, and adult social care – with additional funding for the remainder of the spending review period. This new funding total was then deflated using the deflator provided by the OBR to accompany the autumn statement.

1. General practice

Change in spending on general practice since 2009/10 (real terms)

Spending on GP services comes from the ‘Investment in General Practice in England, 2016/17 to 2020/21’ dataset. ¹⁶ NHS England, Investment in General Practice in England, 2016/17 to 2020/21, 2022, retrieved 7 October 2022, www.england.nhs.uk/publication/investment-in-general-practice-in-england-2016-17-to-2020-21 For 2020/21, this dataset splits out the amount that the NHS provided GPs for spending on Covid-related activity. This information allows us to plot a separate data point for Covid spending in 2020/21.

Appointments in general practice by provider, September 2018 to December 2022 and GP appointments by mode of delivery, September 2018 to December 2022

The NHS changed how it collects information on the number and type of appointments in primary care in October 2018. There is an overall time series going back to November 2017, but granular daily counts of appointments are only available from December 2018. There is, however, a consistent time series of the number of referrals that GPs have made that is available back to 2008.

GP appointments resulting in a specific and acute referral, October 2018 to December 2022 and Specific and acute referrals to NHS secondary care by source per month, April 2009 to December 2022

We start the referral rate time series from the point when there is consistent appointments data, as specified above. For the referral rate, we calculate the proportion of attended appointments that GPs conducted that resulted in a specific and acute GP referral. To calculate the number of attended GP appointments, we take the total number of attended appointments across all of general practice in a given month (as outlined in the ‘Appointments in General Practice’ dataset) and multiply that by the percentage of appointments that GPs carried out (using the SDS Role Group categorisation, rather than HCP categorisation). This step requires us to assume that the attendance rate of GP appointments is the same as the attendance rate of all primary care appointments, an assumption that is unlikely to be met in any month but which will be close enough to make this analysis meaningful. For the monthly number of referrals, we use the ‘GP Referrals Made (Specific Acute)’ data from the Monthly Referral Return dataset. ¹⁷ NHS England, ‘Monthly Outpatient Referrals Data’, retrieved 7 October 2022, www.england.nhs.uk/statistics/statistical-work-areas/outpatient-referrals/mrr-data

Direct patient care staff employed in primary care networks, March 2019 to March 2024

For the projected number of direct patient care (DPC) staff, we calculate the number of DPC staff that the NHS has added to the service per quarter since March 2019. We then extrapolate that forward to come to a total number of DPC employees if recruitment continues at its current pace.

General practice primary care workforce by role, full-time employment, September 2010 to December 2022

The number of GPs used in this chart is the ‘All regular GPs (excludes locums)’ line from the GP Workforce Bulletin tables. This is a combination of ‘All qualified permanent GPs’ and ‘GPs in training grades’ from the same dataset.

Change in GP numbers and patients registered with GP practices since September 2015

As with Figure 1.10, the change in the number of GPs refers to the percentage change in the number of ‘All regular GPs (excludes locums)’. The change in the number of patients comes from the ‘Total number of patients’ line in Table 5 of the GP Workforce Bulletin tables. The starting date for both of these time series is September 2015 because this is when the time series starts in that dataset.

2. Hospitals

Overnight general and acute hospital beds per 100,000 people, April 2010 to September 2022

For this chart, we take the total number of general and acute overnight beds from NHS England’s ‘Bed availability and occupancy’ dataset and divide this number by the number of people in England in the relevant year from the Office for National Statistics (ONS) ‘Mid-year population estimate’ dataset.

Change in doctor and nurse number since 2009

Nursing numbers include adult and children’s nurses who work in hospitals, but do not include community nurses from the ‘NHS workforce statistics’ dataset. Doctor numbers are taken from the ‘NHS workforce statistics, doctors by grade and speciality’ dataset. The total number of doctors includes consultants, associate specialists, specialty doctors, staff grade doctors, specialty registrars, F1 and F2 doctors and hospital practitioners/clinical assistants.

NHS workforce resigning in the previous 12 months, by reason, March 2012 to September 2022

For this chart, we group some of the voluntary resignation categories from the ‘NHS workforce, reasons for leaving’ dataset to make the chart easier to read. The groupings are as follows:

• Work–life balance: this includes just the ‘work–life balance’ category from the original dataset, as this is the category that we want to highlight.
• Working conditions: this includes the ‘better reward package’ and ‘incompatible working relationships’ categories.
• Career development: this includes the ‘lack of opportunities’, ‘promotion’ and ‘to undertake further education and training’ categories.
• Family/external: this includes the ‘adult dependants’, ‘child dependant’, ‘health’ and ‘relocation’ categories.
• Other/not known: this includes only the ‘other/not known’ category.

To calculate the proportion of the workforce that left in the previous 12 months, we first calculated the total voluntary resignations for a given year. This is the sum of the total number of people resigning in the previous four quarters. That amount is then divided by the average NHS headcount across the same 12-month period. The NHS headcount total comes from the monthly ‘NHS workforce statistics’ dataset, and can be found in the tab called ‘1. England’.

For example, for the final datapoint in this chart, 10.8% of the workforce voluntarily resigned in the 12 months to September 2022. The number of voluntary resignations is calculated by summing the last two quarters of 2021/22 (35,057 and 36,263) and the first two quarters of 2022/23 (34,909 and 42,411) for a total of 148,640. The average headcount in the NHS in the 12 months to September 2022 was 1,373,904. The proportion of voluntary resignations was therefore 10.8%.

NHS staff absence by reason, January 2015 to September 2022

NHS England has 25 categories for staff absences, but no specific number for absences due to Covid. We produced an upper-bound estimate for this number by combining the total number of absences listed under S13 Cold cough flu – influenza, S15 Chest & respiratory problems and S27 Infectious diseases. But in practice this will have caught a proportion of non-Covid absences. Our upper-bound estimate of mental health absences uses the numbers reported under S10 Anxiety/stress/depression/other psychiatric illnesses. We cannot preclude the possibility that some staff absences for either Covid or mental health were reported under S98 Other known causes – not elsewhere classified, or S99 Unknown causes / not specified, but for the sake of simplicity we disregarded this.

In all cases we calculate the number of full-time equivalent (FTE) days lost to sickness in a given month for each category as a share of the total FTE days available, and present this percentage as the sickness absence rate.

3. Adult social care

Pandemic-related local authority spending on adult social care, 2020/21-2021/22

This information comes from the ‘Covid-19 financial impact monitoring information’ dataset, released in 20 rounds by DLUHC. Quarterly totals are calculated by summing monthly totals where this is relevant. Spending details for some months are not available in the dataset and in these instances we impute those monthly amounts through comparisons between year-to-date (YTD) amounts in different releases. For example, Round 17 does not include a monthly amount for October 2021, and only shows the financial YTD total for April to October 2021. But Round 16 includes YTD data to the end of August 2021 and forecasts spend for September 2021. Combining the August YTD actuals with the September forecast means that we could create a YTD September total, which, when subtracted from the YTD October actuals, would give an estimate for the spending in October 2021. This means that the monthly totals are likely not to be completely accurate, as September actuals might have differed from the September forecast. But they are close enough for the imputation to be useful.

It should be noted that it was impossible to impute separate totals for July and August 2021 and November and December 2021 because the survey data was released too infrequently. Instead, we calculate a total for the two months. This should not be a problem as, in both cases, the months fall in the same quarter (Q2 2021/22 and Q3 2021/22 respectively) and therefore are shown in aggregate on the chart.

Change in spending on adult social care in England since 2009/10 (real terms)

We calculate the spending – excluding Covid support – data points on the chart by subtracting the additional local authority spending on adult social care (as laid out in Figure 3.1) from the total adult social care spending as outlined in Table 5 of Appendix B of the ‘Adult social care activity and finance report, England – 2021–22’ (ASCAFR) dataset. We use this time series rather than the spend on adult social care as outlined in the local authority revenue outturn dataset because it captures a wider range of spending on the service than just local authority related spending.

Change in people accessing long-term support since 2014/15, by age band

This is calculated as the number of people accessing long-term care during the year – from SALT table 36, in ASCAFR – divided by the number of people in the country in that year, as laid out in the ONS’s ‘Mid-year population estimate’.

4. Children’s social care

Change in local authority spending on children’s social care in England since 2009/10 (real terms)

We calculate the spending – excluding Covid support – data points on the chart by subtracting the additional local authority spending on children’s social care (as laid out in Figure 4.2) from the total children’s social care spending as outlined in the DfE’s ‘Local authority and school expenditure 2021 to 2022’ dataset. We use this time series rather than the spending on children’s social care as outlined in the local authority revenue outturn dataset because it captures a wider range of spending on the service than just local authority related spending.

Additional pandemic-related children’s social care spending, 2020/21–2021/22

This information comes from the ‘Covid-19 financial impact monitoring information’ dataset, released in 20 rounds by DLUHC. Quarterly totals are calculated by summing monthly totals where this is relevant. Spending details for some months are not available in the dataset and in these instances we impute those monthly amounts through comparisons between year-to-date (YTD) amounts in different releases. For example, Round 17 does not include a monthly amount for October 2021, and only shows the financial YTD total for April to October 2021. But Round 16 includes YTD data to the end of August 2021 and forecasts spend for September 2021. Combining the August YTD actuals with the September forecast means that we could create a YTD September total, which, when subtracted from the YTD October actuals, would give an estimate for the spending in October 2021.This means that the monthly totals are likely not to be completely accurate, as September actuals might have differed from the September forecast, but they are close enough for the imputation to be useful.

It should be noted that it was impossible to impute separate totals for July and August 2021 and November and December 2021 because the survey data was released too infrequently. Instead, we calculate a total for the two months. This should not be a problem as, in both cases, the months fall in the same quarter (Q2 2021/22 and Q3 2021/22 respectively) and therefore are shown in aggregate on the chart.

5. Neighbourhood services

Additional Covid spending on neighbourhood services and other local authority-provided services, Q1 2020/21 to Q4 2021/22

This information comes from the ‘Covid-19 financial impact monitoring information’ dataset, released in 20 rounds by DLUHC. Quarterly totals are calculated by summing monthly totals where this is relevant. Spending details for some months are not available in the dataset and in these instances we impute those monthly amounts through comparisons between year-to-date (YTD) amounts in different releases. For example, Round 17 does not include a monthly amount for October 2021, and only shows the financial YTD total for April to October 2021. However, Round 16 includes YTD data to the end of August 2021 and forecasts spend for September 2021. Combining the August YTD actuals with the September forecast means that we could create a YTD September total, which, when subtracted from the YTD October actuals, would give an estimate for the spending in October 2021. This means that the monthly totals are likely not to be completely accurate, as September actuals might have differed from the September forecast, but they are close enough for the imputation to be useful.

It should be noted that it was impossible to impute separate totals for July and August 2021 and November and December 2021 because the survey data was released too infrequently. Instead, we calculate a total for the two months. This should not be a problem as, in both cases, the months fall in the same quarter (Q2 2021/22 and Q3 2021/22 respectively) and therefore are shown in aggregate on the chart.

Neighbourhood services spending here includes all emergency spending on: cultural and related services; housing, environment and regulatory; and planning and development. ‘Other local authority spending’ is the remainder of local authority emergency Covid support, excluding spending on public health, local authority education support services, ‘other – costs associated with foregone savings/delayed projects’ and police, fire and rescue services. We exclude these items to make the totals in this chart comparable with other spending amounts in the chapter, where we also exclude public health, education and police, fire and rescue services. We exclude public health because this only became a local authority responsibility in 2013/14 while our time series for neighbourhood services spending extends to 2009/10, meaning that this spending would be incomparable with other spending amounts in the chapter. We exclude education and fire, police, and rescue services because local authorities do not have any control over the level of spending on these services.

7. Police

Change in gross police spending since 2009/10 (real terms)

Police spending figures for 2020/21 and 2021/22 are generated by combining DLUHC ‘Local authority revenue expenditure and financing England: 2021 to 2022 individual local authority data – outturn’, ‘Revenue outturn central, protective and other services’ RO6 tables for England ¹⁸ Department for Levelling Up, Housing and Communities, ‘Local authority revenue expenditure and financing England: 2021 to 2022 individual local authority data – outturn’, 2022, retrieved 17 January 2023, www.gov.uk/government/statistics/local-authority-revenue-expenditure-and-financing-england-2021-to-2022-individual-local-authority-data-outturn and StatsWales, ‘Revenue outturn expenditure summary, by service’ figures for Wales. ¹⁹ StatsWales, ‘Revenue outturn expenditure summary, by service’, Welsh government, 2022, retrieved 17 January 2023, https://statswales.gov.wales/Catalogue/Local-Government/Finance/Revenue/Outturn/revenueoutturnexpendituresummary-by-service

Incidents of crime excluding fraud and computer misuse, 2009/10-2021/22 and Victim-reported crime by type, 2009/10-2021/22

The Crime Survey in England and Wales (CSEW) suspended face-to-face interviews on 17 March 2020 due to the pandemic. From the year ending June 2020, data tables include data from the Telephone-operated Crime Survey in England and Wales (TCSEW). These do not include crimes experienced by children aged 10 to 15 years. Further details are outlined in ONS, ‘User guide to crime statistics for England and Wales: measuring crime during the coronavirus (COVID-19) pandemic’, 21 July 2022, retrieved 7 October 2022, www.ons.gov.uk/peoplepopulationandcommunity/crimeandjustice/methodologies/userguidetocrimestatisticsforenglandandwalesmeasuringcrimeduringthecoronavi…

Public perception that local police are doing a good or excellent job, 2009/10–2021/22

Supplementary tables were not published as part of the CSEW in 2020/21 or 2021/22, as the new telephone survey limited time and questionnaire length. As a result, not all of the usual questions were asked of all participants. Data for a similar question, on rating the local police, is available for each quarter, however. The 2022 figure represents the average of responses from each quarter.

8. Criminal courts

Backlog of cases in the crown court, Q2 2010 to Q3 2022

The latest official statistics for the backlog in the criminal courts are taken from the Quarterly Criminal Court Statistics up to September 2022. We calculate a backlog adjusted for complexity in three stages:

• We calculate the number of jury and non-jury disposals that are missing by assuming that the share of cases coming into the crown court since March 2020 that end up as jury trials is the same as pre-Covid. The ‘missing’ cases are then the gap between those assumed to be entering the courts system and those that are completed each quarter.
• We treat jury trials and other cases separately. We multiply the ‘missing’ number of both by [share of total hearing time]/[share of total cases] to get a complexityweighted increase in the backlog.
• An ‘ordinary’ backlog is more complex than the average of cases processed (specifically, more cases that will end up as a jury trial), so to adjust this number to be consistent with the pre-Covid backlog we multiply it by [average hearing time of backlog case mix]/[average hearing time of all cases].

9. Prisons

Change in spending on prisons since 2009/10 (real terms)

To project spending in 2021/22, for which the official total is not yet published, we uprate the 2020/21 spending figure in line with the increase in HMPPS spending between 2020/21 and 2021/22 published in the supplementary estimates laid before parliament in February 2022. ²⁰ HM Government, Central Government Supply Estimates 2021-22: Supplementary estimates, 2022, p. 151, https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/1056679/E02711189_HC_1152_Supply_Estimates_21-22_Web_…