Summary of the model
The unit level data which has been used in the analysis includes all CPOs with unpaid work requirements that were either:
1. Imposed between (both dates inclusive) 11 April 2017 and 31 March 2023, or
2. Imposed before 1 April 2017 but still in existence at any time on or beyond that date.
This amounted to 67,500 CPOs with unpaid work requirements over the six years. More technical information on the analysis and all assumptions made can be found in Annex A. The main body of the paper will discuss how the model was created; it has been split into 6 building blocks which are summed to create the final estimations.
Diagram text below:
Building block 1: Successfully completed : Completed on or before 31 March 2022
Building block 2: Unsuccessfully completed : Terminated on or before 31 March 2023
Building block 3: Estimated successful : Estimates with successful completion date after 31 March 2022
Building block 4 : Estimated unsuccessful : Estimates with termination date after 31 March 2022
Diagram text below:
Building block 5: Estimated successful : Estimates started after 31 March 2022
Building block 6: Estimated unsuccessful : Estimates started after 31 March 2022
The first building block counts CPOs where the unpaid work hours had been completed successfully. These formed 57% of unit level records for CPOs with unpaid work between 2017-18 and 2021-22. They are assumed to have been completed because the hours being completed follow a consistent pattern between the start and end date of the requirement. This creates a smoothing effect and assumes work continues every day including weekends and public holidays. This is very different from the way CPOs are run in reality and does not reflect the fact that unpaid work was not taking place during points of the Covid-19 pandemic because of national lockdowns and social restrictions.
The general calculation used in the model is illustrated using the example below. Different assumptions are made for start dates, end dates, etc.. These are explained in more detail in Annex A.
Example: An unpaid work requirement is imposed on 1 December 2018 with 50 hours. It ended on 28 February 2019, with all 50 hours having been completed.
The average hours assumed to have been completed per day is:
Total hours imposed divided by total days
Total hours imposed = 50
Total days till ended= 31 days (Dec) + 31 days (Jan) + 28 days (Feb)
= 90 days
So assumed hours completed (daily rate) = 50/90 = 0.556 hours per day.
Therefore, at close of business on 1 February 2019, the assumed outstanding hours are:
Total hours imposed minus hours completed (1 Feb)
Total hours imposed = 50
Hours completed at end of 01/02/2019 =
number of days worked multiplied by number of hours worked per day
= [31 days (Dec) + 31 days (Jan) + 1 day (Feb)] x 0.556
= 63 x 0.556
= 35 hours
So assumed outstanding hours = 50 – 35 = 15 hours
This was done for all unpaid work requirements that were still running at specific dates. The dates in question were chosen to line up with management information collection dates. These hours were then summed to give overall estimated totals for hours outstanding at different time points. This calculation has a smoothing effect on the data.
Figure 3 is based on CPOs with unpaid work requirements that were successfully completed before 31 March 2022 and shows that the overall number of outstanding hours fell between April 2017 and March 2022. This is to be expected as the closer we move to March 2022, the more likely that requirements which started in the year 2022 will not yet have been finished. Such requirements are not captured in this first building block but in the third. The analysis does also show a noticeably sharp drop in outstanding hours in early 2021. This was due to the introduction of the Community Orders (Coronavirus) (Scotland) Regulations 2021 which came into effect in March 2021. The regulations reduced the number of hours imposed in relevant orders (with exceptions for orders imposed in relation to domestic abuse, sexual offences, or stalking) by 35 per cent in order to reduce the overall volume of hours to be delivered and ensure the continued effective operation of the community justice system
The second building block is to estimate the unpaid work hours for requirements which ended but not successfully. For such requirements, there is no data from the unit level collection in the annual justice social work return on how many hours were completed before the order stopped. An end date is however supplied in the unit level collection. These requirements cover 29% of unit level records for CPOs with unpaid work between 2017-18 and 2021-22.
An assumption has been made that the unpaid work hours stay outstanding in the system until the end date. This model concentrates on unpaid work hours outstanding only. More statistics on completion of CPOs and types of termination can be found in Justice Social Work Statistics.
While the proportion of hours completed is unknown for unsuccessfully completed CPOs, it is possible to apply different estimated proportions to show the range of possible scenarios. For instance, let us suppose the assumption is made that 25% of hours are done for such requirements.
Example: An unpaid work requirement is imposed on 1 March 2019 with 220 hours. It ended unsuccessfully on 13 June 2019 and the person was assumed to have done 55 hours (25% of 220) of work.
The average hours assumed to have been completed per day is:
Assumed hours completed divided by days worked
Assumed hours completed = 55
Total days till ended = 31 days (Mar) + 30 days (Apr) + 31 days (May)
+ 13 days (Jun) = 105 days
So, assumed hours completed (daily rate) = 55/105 = 0.524 hours per day
Therefore, at close of business on 14 April 2019, the assumed outstanding hours are:
Assumed hours completed minus total hours worked
Total hours imposed = 220
Total hours worked = Total days worked multiplied by assumed average hours worked per day
= (31 days (Mar) + 14 days (Apr)) x 0.524
= 45 days x 0.524
= 24 hours
So estimated outstanding hours for this order = 220 – 24 = 196 hours, where it assumed only 25% of hours were completed.
Figure 4 shows the hours outstanding for this second group of orders (unsuccessfully completed on or before 31 March 2022) over the period April 2017 to March 2022 based on different assumptions. The lines relate to assumed completion levels of 0%, 25%, 50%, 75% and 100% for the proportion of hours being completed when the requirement is unsuccessfully completed. This chart illustrates that the higher the percentage assumed to be completed, the smaller the number of unpaid work hours outstanding. The assumption that 0% of hours were completed provides the maximum default for this model i.e., the upper limit. Where 100% of the unpaid work hours were assumed completed this provides the lower limit.
For this stage, the assumption that hours completed varied between 0% and 100% will cause the most variation in the model. This range is used as it provides the lower and upper limits of the model. The model gives us an illustration of possible outcomes if unpaid work hours could have run as usual over the pandemic.
It can be seen that Figure 4 smooths the effects of the pandemic i.e., there is a continual decline rather than a sudden drop. The nature of the calculations used in this analysis has assumed that work was still progressing regularly during the pandemic years but this will not have been the case in reality.
Like the first block, the pattern for hours fell sharply in the most recent time periods. This is also expected as requirements which started close to March 2022 will generally not have had sufficient time to be finished. These requirements are included in the fourth building block.
First stage of the model – Calculations on unpaid work requirements that ended by 31 March 2022
Figure 5 combines Figures 3 and 4, covering all the CPOs that ended by 31 March 2022. This is the first stage of building the model. The grey area represents the difference between the lower and upper limits for Figure 4, and gives the model an estimated range:
1. The top line of the grey area is the sum of unpaid work hours for those CPOs that eventually completed successfully and the unpaid work hours for unsuccessfully completed CPOs where it was assumed 0% of hours were completed. (i.e. the upper limit)
2. The bottom line of the grey area is the sum of unpaid work hours for those CPOs that eventually completed successfully and the unpaid work hours for unsuccessfully completed CPOs where it was assumed 100% of hours were completed. (i.e. the lower limit)
For this model the number of outstanding unpaid work hours at a specific point of time lies within the grey area.
As the requirements in the first and second blocks were completed by 31 March 2022, the outstanding hours decreases rapidly in the last two years.
Figure 5 shows that the number of outstanding hours fell across the period April 2017 to March 2019. This is to be expected as, over this period, not only did the number of CPOs imposed fall but the number of unpaid work requirements associated with those CPOs also fell. The outstanding hours were relatively steady over year 2019-20. This was consistent with the fact that, while the number of CPOs imposed in that year went up slightly, the prevalence of unpaid work requirements fell.
Across the first three years (2017-18 to 2019-20), the lower limit of the model was never less than 500,000 outstanding unpaid work hours. This is not an unexpectedly high level, as across these three years:
- Around 130,000 unpaid work hours were imposed by courts each month
- Unpaid work requirements which were successfully completed took an average of around seven to eight months to finish.
Second stage of the model – Estimates on unpaid work requirements that was not completed by 31 March 2022
Stage two of the model contains the third and fourth building blocks covering requirements which were started but had not finished before the end of March 2022. This covered 14% of CPOs with unpaid work requirements over the five years.
For these blocks, it was not known how many hours would be done nor whether the requirements would go on to be successfully completed. The requirements were assigned (randomly) to two categories: successfully or not successfully completed. The split used was 65%/35% respectively. This ratio was calculated from previous years data, and further assumptions were made as to how many hours were completed and by what date. More information on the assumptions made can be found in Annex A.
The same methodology of calculating an upper and lower limit has been applied to the unsuccessfully completed category as in the first stage. Again, this has created a range, shown by the grey area in Figure 6. The successfully completed hours estimated is the blue line in Figure 6.
Figure 6 shows that, for both categories, the number of hours outstanding increased across the period from April 2017 to March 2022. This rise was much more marked towards the end of this period than it was at the beginning. This is expected as, for example, a requirement which started in early 2022 was much more likely to have more hours outstanding than a requirement which started in April 2018.
There was a large decrease in both the line and the grey range area after the start of April 2022. This was expected as:
- All the requirements could only have finished on or after 1 April 2022, and
- Requirements which started from April 2022 onward are not included here but are in the final stage
Figure 6 shows that it generally takes less time for a successfully completed requirement to finish than it does for one which was unsuccessfully completed, as the outstanding hours are generally higher for those assumed to have finished unsuccessfully than for those assumed successfully completed.
Adding the four building blocks together, the overall picture with all outstanding unpaid work requirements can be seen in Figure 7. The model indicates that the total number of outstanding hours was highest in the pre-Covid-19 period, varying between 500,000 and 900,000. The average difference between the upper and lower ranges for the first three years was about 230,000 hours. The outstanding hours fell in the period after the start of the pandemic, particularly after the Coronavirus Regulations in March 2021, which reduced the number of hours imposed in some orders. The graph does, however, show a rise in late 2021 and early 2022, with the lower limit just above 450,000 hours in March 2022.
Although the model assumes that a daily rate of hours are being completed, it also considers the increase in time taken to process unpaid work hours during the pandemic in 2020-21 and 2021-22. The time taken to process unpaid work requirement during the pandemic increased. Between 2019-20 and 2020-21, there was an increase between 3.5 to 7 months depending on the length of unpaid work hours imposed. A similar increase in length could be seen when comparing 2019-20 to 2021-22. (Annex A).
Model in relation to management information on outstanding unpaid work hours data collection
Management information on unpaid work requirements and hours outstanding was collected on an ad-hoc basis during the early period of the pandemic. This allowed the Scottish Government and stakeholders to monitor outstanding unpaid work hours during this time. This information was not available through the National Statistics unit level collection.
This data collection took place at times when it was deemed most appropriate to establish the up-to-date situation during the pandemic. However, from May 2022, the decision was taken to conduct the collection every three months. The information supplied is administrative data from various management IT systems from local authorities and excluded outstanding unpaid hours relating to breaches. It will vary day to day, as information is added and closed within the IT system as part of usual administration processes. In a small number of cases, estimates were calculated for this data collection. The nature of this data was to provide a broadly indicative summary, rather than a precise measure of activity. Therefore, the outstanding unpaid work hours data collection has been classified as management data and not official statistics.
The management data is displayed in the blue line in Figure 8. This information shows what happened with Scotland’s outstanding unpaid work hours during the first year of the Covid-19 pandemic (2020-21). The line starts to increase as the first national lockdown (March 2020) and social distancing restrictions started to impact on justice social work processes. The hours steadily increased until March 2021. When this is compared to the estimated model, you can see the line is not within this grey area. This is to be expected, as the model does not consider the social restrictions and assumes the delivery of unpaid work hours continued as normal during that period, as the unit level data cannot provide this exact detail.
The model shows a decrease in hours between March and September 2020 which reflects the fact that very few CPOs were being issued by courts over this period, as a large number of courts were closed or operating at reduced capacity. Both the management data and the model show a decrease in hours outstanding because of the Community Orders (Coronavirus) (Scotland) Regulations 2021. This drop is more pronounced in the management data which fell by 250,000 hours between February and April 2021 from 820,000 to 580,000. For the model, this drop is much smaller, with both the upper and lower levels falling by around 120,000 hours. The estimated range for outstanding unpaid work hours in April 2021 was between 420,000 and 610,000. This is to be expected as the model is unable to show the true extent of social restrictions in the way it is calculated, this is instead what the management information shows.
Final stage of the model – Estimates on unpaid work requirements from 1 April 2022
The management data has continued to be collected from local authorities up to and including the collection for August 2023. However, the unit level data for 2022-23 will not be available for comparison until early 2024.
To try to understand what is happening from 1 April 2022, another source of data was used in the model. This data has been extracted from the management information used to publish the Scottish Government Justice Analytical Services Criminal Disposals Dashboard. This management information is based on the Scottish Courts and Tribunal Service (SCTS) administrative database (COPII). Information from this administrative database is based on management information. As such, results from the dashboard should be seen as providing a broadly indicative summary rather than a precise measure of activity. In order to indicate the change in the model due to the different data source, the grey area is patterned from figure 10 onwards).
Source: Scottish Government Justice Analytical Services Criminal Disposals Dashboard 2022-23
The management information used in the disposals dashboard shows the data at charge level, whereas the local authority data is at CPO unit level. For this model, it was adjusted to be nearer the unit level data by using only the maximum number of unpaid work hours per case-accused. Whilst this is not exactly how CPO unit level data is calculated, it did give a good approximation. In the pre-pandemic year (2017-18 to 2019-20) for example, this averaged maximum number of unpaid work hours per case-accused nationally about 3% lower than the actual CPO unit level data. The disposals dashboard management information was considered a good approximation for 2022-23. Without this data, 2022-23 would not be modelled, making the model 18 months behind the management information rather than six months.
From April 2022 to March 2023, the courts imposed CPOs containing an average of just over 106,000 hours of unpaid work per month.
The numbers in Figure 6 (based on blocks three and four of the model) estimated the dates the requirements ended, for outstanding unpaid work hours still in existence at close on 31 March 2022. That same technique used in building blocks three and four was also used to create blocks five and six on the disposals dashboard management information. An adjustment has been made to the average length of time to complete the order to reflect the moving away from Covid-19 restrictions. This was estimated by averaging 2019-20 and 2021-22. This reduces the average length taken to a value between the pre-pandemic and pandemic years (see Annex A).
Figure 10 shows the completed model, which includes each of the six building blocks, and indicates key events relating to the pandemic.
A: First lockdown – March 2020
B: Second mainland lockdown – January 2021
C: Coronavirus Regulations introduced – March 2021
D: Scotland moved to beyond level 0, when the legal requirement for physical distancing and limits on gatherings was removed. Some protective measures did stay in place such as the use of face coverings indoors – August 2021
E: Legal requirements to wear face coverings on public transport and most indoor settings ended – April 2022
In Figures 10 and 11, the introduction of the unpaid work requirements data from disposals dashboard management information has created an artificial dip on 1 April 2022. This is due to timing issues between the two datasets in the model. This corrected itself from May onwards.
Figure 11 is the same graph as Figure 10 without 2017-18 and 2018-19. Figure 10 is a simplified version with the event time points removed. It shows more clearly the use of the disposals dashboard management information in the model. This is the striped section.
The model is based on several assumptions. However, the overall trends for the model show the number of outstanding unpaid hours in the system were at their lowest estimate (range between 420,000 and 610,000 hours) after the Coronavirus Regulations were introduced in March 2021. Similarly at that time, the management data dramatically dropped from its highest point (820,000 hours) to its lowest point of 580,000 hours. The management data starts to increase steadily thereafter. Again, this is reflected in the model, from May 2022 to December 2022. The model then decreases in the first two months of 2023 but increases again in March.
The model is only able to estimate up to the end of March 2023 due to data availability. The outstanding unpaid work hours management data collection was last collected in August 2023. This analysis shows that the pandemic did affect outstanding unpaid work hours and there is still a backlog of hours in the system at the end of the financial year ending 31 March 2023. In February 2023, the management information had 748,000 unpaid hours outstanding. For the same date the model’s upper limit is 749,000. The upper limit assumes that no unpaid work hours were done for unsuccessfully completed unpaid work requirements. This would not be the case in reality. This indicates that the backlog created by the pandemic is still influencing the outstanding unpaid work hours but cannot be quantified exactly.
The management data increased to 781,000 outstanding unpaid hours in May 2023 and then dropped to 770,000 in August 2023. The increase in unpaid work hours in May 2023 is partially due to the above average number of hours being added to the system from March 2023 (Figure 9).
It would be advisable to check this model with 2022-23 unit level data to update the average length of order for 2022-23. It would also be advisable to continue collecting the management data from the local authorities, and making this publicly available, on a regularly basis for the foreseeable future, to help to monitor recovery from the pandemic. If there was no pandemic backlog remaining it would be expected that the management information would move away from the upper limit to sit more in the middle of the model.
There is a problem
Thanks for your feedback