National Statistician’s Independent Review of the Measurement of Public Services Productivity

Published:
13 March 2025
Last updated:
14 March 2025

Chapter 12: Public Order and Safety, including the Criminal Justice System

As discussed in Chapter 10, the Public Order and Safety (POS) Classification of the Functions of Government (COFOG) category is a complex mix of different services currently grouped under a single COFOG category. This report therefore discusses productivity measurement for the policing and immigration & citizenship services separately (see Chapter 10 and Chapter 11 respectively) given the distinctive nature of these services. The remaining services included in the POS measure are law courts (which itself has five further sub-components: Magistrates’ Courts, County Courts, The Crown Court, Crown Prosecution Service, and legal aid), prisons, probation, and Fire & Rescue.

In 2017, the Office for National Statistics (ONS) introduced a new model for measurement of the Criminal Justice System (CJS) components of the POS COFOG category – primarily courts, legal aid, prisons and probation, and supporting activities such as electronic monitoring and prisoner transportation.

This new model implemented methodological innovations around weighting quality adjustments and heralded consideration of the criminal justice system as a model of integrated service delivery. Quality adjustments currently included in the POS measure are:

  • Re-offending.
  • Courts’ timeliness.
  • Prison safety.
  • Custody escapes.

In the context of this Review the Ministry of Justice (MoJ) proposed a number of potential further improvements, which this chapter discusses. In particular, the impact of the coronavirus (COVID-19) pandemic and the consequent staff furlough scheme resulted in reduced court services being offered that impacted on the measurement of productivity estimates for the courts and the quality adjustment based on re-offending rates.

12.1 Core methodological and data challenges identified

Even after separately dealing with Policing and Immigration & Citizenship services, together with the various components of the CJS, there remains a final service, Fire & Rescue, which is accommodated in this service grouping. Fire & Rescue services have been historically captured in this area and the Review considered this continued to be appropriate, pending any update of the international COFOG structure.

In 2020, due to the coronavirus pandemic, court services were significantly impacted and reduced due to the furlough scheme. Consequently, fewer convictions were passed down. Given the re-offending metric used to quality adjust the CJS is, in reality, a ‘re-conviction’ metric, this meant that the ONS was forced to suspend the re-offending section of the quality adjustment of this service during the pandemic. The Review explored options to re-apply this.

The inputs of the CJS productivity measure are currently calculated indirectly (Labour, Intermediate Consumption (Goods & Services) and Capital Consumption), using annual UK National Accounts data and a range of deflators. As in other services, the Review explored moving to direct measurement by using appropriate full-time equivalent (FTE) workforce data and associated average salary information.

Direct measures of quantity output have already been included in the total public service productivity (PSP) series for, as this chapter will henceforth refer to them, the four primary CJS components (courts, legal aid, prison, and probation). However, the Review has identified improvements given the availability of updated data sources to bring measures of the court service up to date, alongside methodological improvements to reflect changes in the composition of prisons by category. The MoJ also queried the pertinence of some of the current quality adjustments to the court system. The rationale for the current model and the reflection of the Review on this are explained in the Improvements to Quality Adjustments section.

Except for Fire & Rescue services, all output is quality adjusted using a range of metrics. There are several metrics used to derive quality adjusted CJS outputs and the Review identified within these that updating data sources for the output of criminal courts and updating the quality adjustments with the best data on case timeliness in courts, were priorities. Identifying and sourcing unit cost data to facilitate output methods improvement across the POS services is particularly challenging.

12.2 Improvements to inputs estimates

The inputs element of the POS productivity measure uses an indirect approach of deflated expenditure to derive the volume of Labour, Intermediate Consumption (Goods & Services) and Capital Consumption inputs. An aggregate index is then compiled for total POS inputs and weighted using expenditure levels for the four primary CJS components.

The Review focused on assessing the latest availability of workforce data to explore the feasibility of moving to a direct measure of labour inputs for each of the four primary CJS components, as per Annex F. Where data permitted, the Review compared the direct approach against the existing indirect series, as recommended in the Atkinson Review.

Criminal Justice System

Data were sourced from a variety of areas including the ONS Annual Survey of Hours and Earnings (ASHE) and published management information. The Review also investigated options to obtain unpublished data directly from relevant organisations. Data availability varied, with only a limited timeseries available in some cases. Certain services have also implemented changes to grades or job roles over the years, which results in issues of comparability over time.

Work undertaken to derive direct measures of labour for the four primary CJS components was advanced as far as data availability and time constraints allowed. Following engagement with MoJ officials the Review feels that progress in these areas is possible (see 12.5 Recommendations for further work).

In addition to work on improving the measurement of labour input, the ONS has improved the central and local government deflators used to determine volume growth in the Intermediate Consumption (IC) measure. Due to the different types of expenditure across services, a composite bespoke IC deflator has been generated for each services, with the exception of the Fire services, where IC expenditure will be deflated using the headline CPI-based deflator, both because of relative scale and absence of data.

Recommendation 79:

The ONS should apply bespoke intermediate consumption deflators to expenditure data in all Public Order and Safety service areas, with the exception of fire.

Fire & Rescue

The Review developed a direct labour input measure for Fire & Rescue services. This relates to local government labour, which accounts for more than four-fifths of Fire & Rescue service labour expenditure.

To account for the different FTE volumes by rank or grade the ONS matched the FTE in each rank with relevant average salary data from ASHE. ASHE was deemed the most suitable data source with a long time series, as this currently uses aggregate ranks and grades. The Review considers that given the comparably small size of this service, ASHE provides sufficient detail and further work to find more granular data is feasible, but not necessary or recommended at this time.

From ASHE, the ONS extracted the data series for the following Standard Occupational Classification codes with availability back to 1997, consistent with the treatment of other areas in taking account of generic administrative functions:

  • 3313 – Fire service officers (watch manager and below).
  • 1163 – Senior officers in fire, ambulance, prison and related services.
  • 4113 – Local government administrative occupations.

A comparison of the new direct measure and pre-existing indirect approach showed they follow a pattern of negative growth between 2011 and 2016, but the rates are more pronounced in the indirect measure. Using the direct approach, negative growth extended to 2018. The fall in direct labour captured in this analysis was in line with previous ONS published analysis indicating a fall in workforce across Fire & Rescue services.

Based on the data available, the Review recommends implementing this change from 2011 onwards into published ONS estimates. Home Office has developed a comprehensive Fire salary model, but at present there is an insufficient time series for incorporation as an alternative to ASHE. The available Home Office data on FTE relates to England only. The ONS would seek to expand coverage of the direct labour measure to the devolved governments should data be available.

Recommendation 80:

A new method implementing a direct measure of fire service labour inputs into Public Order and Safety should be introduced, expanding in future years to cover devolved governments.

12.3 Improvements to outputs estimates

Prisons

The activity measure used for English and Welsh prisons output is MoJ published monthly estimates of the prison population. This does not contain data on differences in the cost of different categories of prison and therefore the effect of changes in the composition of prisons. The Review has identified data to improve the estimation of prisons output to enable this heterogeneity to be reflected.

The new prisons output measure will weight activity by their security categories, with higher weights applied to higher cost prisons. For example, prisoners in ‘Dispersal’ prisons (those prisons which hold Category A prisoners, alongside other prisoners) will have a greater weight than those in ‘Male open’ (establishments where prisoners are lowest risk).

The current prisons output measure uses total monthly prison population figures for England, Wales and Scotland. Northern Ireland is not included in the current output measure for prisons. The proposed output measure will continue to cover England and Wales, although these will now be split by security category. Scotland will also be captured as a single category due to the different design of Scottish prisons compared to the English and Welsh dispersal, local, trainer or open prison model. Appropriate new data sources may exist for Scotland.

Recommendation 81:

The ONS should explore possible new prison data sources for Scotland to allow for differential cost weights for different categories of prison, and data for Northern Ireland in the ambition of creating a UK-wide output measure.

The cost weights come from prison performance data which provides the expenditure of individual prisons annually for England and Wales. These are published by security category which are used to split the activity data. For England and Wales, the ONS uses the categories provided in the annual cost publication to allocate cost weights for each of nine categories:

  • Male dispersal (category A).
  • Male reception (category B).
  • Male Trainer (category B).
  • Male category C & Young Offenders Institute (YOI), Trainer and Resettlement.
  • Male open (Category D).
  • Female closed.
  • Female local.
  • Female open.
  • Male YOI young people (ages 15 to 17).

As the ONS currently lacks data disaggregating the prison population in Scotland by category, the average cost for English and Welsh prisons is used to weight the Scottish prison population figures. These matching activity and expenditure figures will then be used to create the Cost Weighted Activity Index (CWAI) for prisons output.

Probation

The current probation output measure uses number of offenders being supervised to capture activity. This is taken from data published by the MoJ. Data are currently only available for England and Wales. Currently only a single activity measure is used with all probationers given the same weight.

However, as this measure does not reflect differences in the cost of different groups of probationers, the Review developed a new probation output method to improve measurement of productivity by splitting the activity into two categories of probationers:

  • Community order and suspended sentence order.
  • On licence.

Community order and suspended sentence order probationers will be summed into a single activity. Relative weights will be assigned to these two activity categories using the marginal unit cost of each group of probationers provided by MoJ, allowing the ONS to create a CWAI for Probation. However, as these data are available only for financial year ending 2024 the ratio of unit costs for the two groups will be fixed over time. The cost weight for probation output relative to other aspects of POS will continue to come from HM Treasury data, meaning the average unit cost of probationers will still vary over time.

Recommendation 82:

The ONS should continue to work with the Ministry of Justice to obtain expenditure data that enable the relative weights of different groups of probationers to vary with time in future years.

Recommendation 83:

The ONS should explore the data available on probation services for Scotland and Northern Ireland with a view to creating a complete UK measure for probation output.

Criminal Courts

The output measure used by the ONS for criminal courts is a CWAI covering the activity of the courts of first instance for criminal offences, that is, the Magistrates’ Courts and Crown Court. Magistrates’ courts deal with cases sanctionable with up to 12 months in prison and motoring offences. The Crown Court deals with more serious or complex cases, such as serious assault or large fraud. The activity of appeals courts is not included in the output measure as firstly the number is small, but secondly, if the desired outcome is successfully completed trials, including the costs of appeals but not an additional output metric allows the ONS to see the system cost of delivering that outcome.

The Crown Court

The current output measure for the Crown Court uses three activity categories: net committals for trial, net cases for sentence and net appeals. These activity series have been forecast since 2018 and the relative weights for the three activity categories uses a set of unit costs from 2000. The Review has developed an updated cost weighted activity measure for the output of the Crown Court. The new output measure uses the criminal court statistics quarterly published by MoJ, who have recommended using disposals as the activity measure.

The analytical tool ‘Pivot Table Analytical Tool for England and Wales‘ published alongside the activity data includes data on the average number of hours taken for each type of court activity (hearing type, offence type and plea). By splitting by the hearing type and plea the ONS is able to weight the different types of activity with their own cost weights and is therefore able to better capture changes to the landscape of the criminal courts system. The data are available from 2014 and, where unavailable, the existing activity data series will be used to generate the output index for earlier years.

Note that splitting by offence type as well as hearing type and plea has also been explored. The resulting index varied little from that split by hearing type and plea alone, yet the resource demand of splitting by offence type was significantly larger. As such, the Review has discounted this option.

Table 6 shows the four hearing types (triable-either-way, indictable only, committal for sentence and appeals) with their four respective plea options. Where a plea is not applicable (non-trial hearings), the hearing category will be taken as a whole leading to a total of 10 activity categories.

Table 6: Hearing types with plea types options for Crown Court disposals

Hearing typePlea type
Triable-either-way trialsGuilty plea
Not guilty plea
No plea entered
Dropped case
Indictable only trialsGuilty plea
Not guilty plea
No plea entered
Dropped case
Committed for sentenceNot applicable
AppealsNot applicable

To weight these ten categories, the Review has used the analytical tool ‘Pivot Table Analytical Tool for England and Wales’ discussed to estimate overall hourly cost for each of the hearings – plea type using overall average hearing time divided by overall expenditure, using HM Treasury data, which includes both labour and non-labour costs. Other MoJ expenditure data provided was considered, but only contain labour costs. With hourly cost calculated, the differential in the number of hours per hearing type or plea can be used to get a unit cost for each category.

Magistrates’ courts

For magistrates’ courts only a single activity measure is used, defined as caseload. This activity series has been forecast since 2011. As with the Crown Court, the ONS used disposals as the activity measure, with this taken from the criminal court statistics quarterly published by MoJ. Due to data limitations, magistrates’ courts activity will be taken as an aggregate measure.

The activity data used to improve these criminal court output measures covers England and Wales only. Further work is required to identify suitable data for Northern Ireland and Scotland to produce a complete UK measure.

Cost weighting

Weights to aggregate the various components are derived from the Online Systems for Central Accounting and Reporting (OSCAR) dataset, HM Treasury’s expenditure dataset used for Central Government spending in the UK National Accounts and much of PSP inputs. COFOG 03.3 refers to ‘public order and safety: law courts’, with a further breakdown derived by the ONS for the Crown and magistrates’ courts expenditure respectively. As this breakdown is coded using OSCAR segment codes according to a methodology set up when OSCAR started, it is believed these breakdowns are out of date and likely result in both the Crown Court and magistrates’ courts being underweighted relative to the other services in POS. Work is ongoing with MoJ to obtain more suitable expenditure data to address this issue.

Recommendation 84:

The ONS should continue to work with the Ministry of Justice to obtain expenditure data to derive improved weights for individual law courts services.

Recommendation 85:

The ONS should explore the data available for Scotland and Northern Ireland with a view to creating a complete UK measure for criminal court output.

Legal Aid

Legal Aid is the provision of legal services to those who require them in and around court hearings. In this area the ONS only measures public provision of Legal Aid, rather than privately provided legal advice. This is because the consumer of these privately provided services has ‘opted out’ of using the public model (either voluntarily or involuntary, such as eligibility criteria). This is the same as not including private schools within the measure of education services: the opting out has removed the function from public services.

The Review has developed a new measure of Legal Aid output. This greatly increases the coverage from four lower crime legal aid activity categories to:

  • All lower crime categories which represents Legal Aid work carried out in police stations and magistrates’ courts.
  • All higher crime categories which represent Legal Aid work carried out in the Crown Court and above.
  • All of civil legal aid such as Special Children Act Proceedings.

The new output measure continues to use the MoJ’s Legal Aid statistics tables quarterly publication. This source publishes both activity and expenditure using matching categories, which enables the construction of a CWAI. The ONS is expanding the scope of use of this source to encompass almost all legal aid output activity. The new method will take the most disaggregated data available, increasing the number of categories from 4 to 72, as this provides the greatest level of granularity and means the CWAI is more sensitive to changes in higher-cost activities and less sensitive to others.

Recommendation 86:

The ONS should explore the data available for the Scotland and Northern Ireland with a view to creating a complete UK measure for Legal Aid output.

Alongside this, the Review is assessing the concept of Legal Aid being included as intermediate consumption in the wider criminal justice system. Whilst the ONS treats Legal Aid as a standalone function with its own outputs, increasingly the Review has accepted the argument that Legal Aid cannot operate outside the wider court system: if the Legal Aid solicitor does not arrive, the case is adjourned, in the same way that if the judge is unavailable the court session will not commence.

The Review therefore recommends the ONS considers the possibility of linking legal aid to criminal courts data and using expenditure on this within the weighting of the delivery of justice services more widely, removing the distinct Legal Aid output measures. Feedback on this proposal would be welcomed by the Review.

Recommendation 87:

The ONS should undertake further research into whether Legal Aid should be considered an intermediate stage in the provision of justice as opposed to an output in itself.

12.4 Improvements to quality adjustments

The Review focussed on the existing quality adjustments for criminal court (Crown and Magistrates’ Courts) output within POS. The quality adjustments, which are given equal weight, applied at present are:

  • Timeliness – The timeliness adjustment relates to the average time taken for criminal cases to reach completion, on the basis that the delivery of a sentence in a timely manner is favourable. This is measured through the reciprocal of mean time until completion.
  • Recidivism – The recidivism (or re-offending) measure is based on the number of proven re-offences, adjusted for the ‘offender mix’ of the cohort (i.e. their likelihood to re-offend), weighted to account for the severity of the proven crime types, and expressed as a rate per person. This is calculated as the observed reoffending rate for a particular cohort plus any difference between the Offender Group Reconviction Score (OGRS) 4/G score in that cohort and a baseline cohort in 2011.

Areas considered for improvement were:

  • Improvement of the granularity of the timeliness quality adjustment in line with the new criminal court output measures being developed through the Review (see Improvements to outputs estimates).
  • Introduction of an additional quality adjustment to account for change in the case backlog in the criminal courts service.
  • Re-introduction of a recidivism quality adjustment to account for the period when the adjustment was held constant due to the impact of the coronavirus pandemic on the source data.
  • Review of weighting levels for both adjustments.

Improving granularity of the timeliness quality adjustment

Data for the existing timeliness adjustment are available from 2011; the current data for Magistrates’ timeliness have been used since inception, while Crown courts changed to an improved method in 2014. The adjustments for both court types are:

  • Magistrates’ courts: Average number of days between charge and completion.
  • Crown courts: Average number of days from case being sent by Magistrates to the Crown courts, to completion in the Crown courts.

These adjustments are then indexed and applied to the respective courts output data.

Criminal cases do not always take a linear path through the courts system. Depending on their severity, Magistrates can send cases to Crown Courts either before trial, dependent on the nature of the crime and the plea, or after a trial, if they believe the sentence should exceed their sentencing powers, so the contribution of both courts to the timeliness of cases can be intertwined and difficult to fully reflect in this adjustment. To ensure the ONS factor this complexity into its adjustment, the Review identified relevant data from the MoJ ‘timeliness tool’ which puts substantial data into the public domain. The ONS and MoJ have collaborated to ensure that the most appropriate source for the variables are included into the quality adjustment.

The MoJ ‘timeliness tool’ also allowed the opportunity to improve the granularity of the timeliness quality adjustment in line with the new case disposal data being used to derive the criminal court output measure. As a result, the ONS will now be able to ensure the timeliness adjustment is reflective of the underlying mix of case disposal activity at Crown Court (i.e. aligning with the relative proportion of both case type and plea), and incorporate Triable-either-way, Indictable, Committed for Sentence and Appeals case types.

Both indictable and triable cases are further disaggregated by plea types (not-guilty, guilty plea, no plea entered, dropped case). Average timeliness was used for appeals and those cases where no plea is entered (as these are excluded from the MoJ timeliness calculations). This composite timeliness adjustment better accounts for the differing case complexity and more serious cases by considering the respective timeliness and relative expenditure as opposed to using a simple average based on the number of cases alone.

The Review also analysed whether a similar, granular approach to timeliness quality adjustment is possible for Magistrates’ Court output (see Recommendations for further work).

Recommendation 88:

Timeliness data by case type should be used to produce a more granular quality adjustment that aligns with the improved Crown Court output measure.

Consideration of case backlog

While the timeliness adjustment is used to account for changes in the speed at which cases progress through the courts system, the Review has also considered whether an additional adjustment should be made to account for change in the volume of outstanding cases (backlog). It was felt that the term ‘backlog’ implies that all cases are somehow past their due date, whereas the published data refers to the ‘open caseload’, and as such reflects both ‘work-in-progress’ as well as backlog.

The Review has concluded that there is a clear relationship between backlog and timeliness (i.e. where backlogs rise, timeliness generally gets worse) and that including both risks compounding the impact of the quality adjustments. The Review does not recommend including an additional backlog quality adjustment.

Re-introduction of recidivism adjustment

The proven re-offending data used within the PSP measure were materially impacted because of the disruption to court proceedings during the coronavirus pandemic as re-offending is measured through re-convictions, which requires courts holding cases. With courts furloughed, the ONS made the decision to hold the recidivism quality adjustment constant at the level applying to the last cohort unaffected by court closures over the lockdown periods. The rate has now returned to a level within historical ranges, and the Review considered how best to address the reintroduction of the adjustment, over the coronavirus pandemic impacted period.

Several options were explored including (1) retaining the current approach of keeping the adjustment constant, (2) removing the adjustment completely during the impacted period and using the published re-offending data, and (3) re-introducing the adjustment based on published data for the 2022 cohort and using linear interpolation to estimate the impacted period (2018 to 2021).

The Review concluded that option (3) was most appropriate. As the resulting trend is one of gradual decline in the severity-adjusted rate of re-offending the ONS uses in its measure, this will result in a positive impact on the quality adjustment through the impacted period. This option was seen as preferable to the volatility which would have been introduced to the series from removing the adjustment completely during the impacted period.

Recommendation 89:

A straight linear interpolation should be applied to the re-offending quality adjustment applied for 2018 to 2021 where the coronavirus pandemic interrupted the normal provision of data.

Review of weighting levels

In light of the re-introduction of a recidivism quality adjustment, the Review considered the relative weights attributed to each of the components of the quality adjustments applied to criminal courts output. When considering a baseline framework for capturing a quality adjustment measure for Courts productivity, the requirement was a measure of intrinsic core values which hold stable across time for a fair measure, resistant to the changes in policy of the government of the day.

This framework is provided by the five principles of sentencing, introduced in the 2003 Criminal Justice Act, which lists the legal purposes of sentencing by the court as the following:

(a) the punishment of offenders.
(b) the reduction of crime (including its reduction by deterrence).
(c) the reform and rehabilitation of offenders.
(d) the protection of the public.
(e) the making of reparation by offenders to persons affected by their offences.

Two of these principles (b and c) relate to recidivism while the remaining principles can be considered to have a greater association with the timeliness component of the measure. Noting this, there is no reason in the 2003 Criminal Justice Act to assume any of these five purposes has greater importance than others. The Review, therefore, recommends a revision to the equal weighting of the quality adjustments, which treats the five purposes as equivalent to each other and will be incorporated into the Spring 2025 annual ONS statistical release as follows:

  • 40% weighting to Recidivism.
  • 60% weighting to Timeliness.

Recommendation 90:

Court quality adjustment weights should be revised to be 40% recidivism and 60% timeliness.

12.5 Recommendations for further work

Development of direct labour measures for Criminal Justice services

Prison and Probation: Initial investigations and discussions with MoJ highlighted published HM Prisons and Probations quarterly workforce statistics. Based on this, the ONS extracted FTE data for HM Prison and Probation employees dating back to 2010 for England and Wales.

Accurate historic salary data prior to 2018 were not available due to a change in payroll systems, and sufficient historical data are not available to produce average costs. It was therefore not possible to obtain bands that aligned appropriately with published workforce statistics. Ongoing engagement with MoJ indicates that an extended back-series of FTE data and more granular management information on the prison workforce may be available in the future, if so this could be used to develop an enhanced direct measure.

In terms of developing a direct labour measure for the Probation service, a lack of disaggregated salary data, combined with the part-privatisation and renationalisation of services in the recent past resulted in the Review concluding that probation labour input should continue to be measured indirectly at present.

Recommendation 91:

Work should continue on identifying appropriate data to develop a direct measure of labour input for the prison service component of Public Order and Safety.

Courts: This is another area where the Review considers a direct labour measure will be possible in the near future. MoJ publish management information detailing FTE for services, which covers England and Wales. MoJ has advised that data prior to 2011 are not readily available, may not be robust and would require significant resource to obtain.

While the Review derived a measure of direct labour using FTE and available salary band data, other labour components within HM Courts and Tribunals Service were not captured, and there were expenditure items which may already be captured within the intermediate consumption inputs measure.

A comparison of the direct measure the Review derived with the existing indirect measure showed that while the long-term trends for both indices are broadly similar, the direct labour Index had more pronounced negative annual growth between 2012 and 2017. The Review recommends retaining an indirect labour approach while engaging with MoJ to ensure no double-counting in both labour and intermediate consumption by moving to a direct labour approach.

Recommendation 92:

Work should continue on developing a direct measure of labour input for the courts service component of Public Order and Safety, ensuring no double-counting with intermediate consumption input in this area.

Back to top