Research and analysis

The impact of COVID-19 lockdowns on crime demand and charge volumes in England and Wales

Published 25 January 2024

Applies to England and Wales

Hannah Wharfe, Scarlett Furlong, Andy Feist and Alex Dunnett

We would like to thank Dr Samuel Langton (Postdoctoral Researcher at the Netherlands Institute for the Study of Crime and Law Enforcement) and Dr Aki Roberts (Associate Professor of Sociology at the University of Wisconsin-Milwaukee) for peer reviewing this report.

We are very grateful to the following individuals who also contributed to the report: Jennifer Wilcox, Katherine Hopper and Thomas Wray.

Executive summary

This paper examines the relationship between changes in crime demand and charge volumes in England and Wales during the COVID-19 lockdowns. It seeks to add to a modest evidence base on the relationship between changes in police workload and the detection of crimes.

Imposing restrictions on individuals and businesses to curtail the spread of COVID-19 had a profound effect on people’s behaviours, which in turn affected patterns of crime; police recorded crime (excluding fraud and computer misuse) fell markedly, down by 13% in the year ending March 2021, while crime weighted by severity also fell. This reduction was mainly attributed to limits on people’s movement, the closing of retail establishments and the night-time economy, and an increase in people working from home. Theft offences saw particularly large falls.

The overall impact on charges was less marked. In the 12 months to 31 March 2021, there was a 5% reduction in charge volumes. However, in the first quarter (April to June 2020) – the period corresponding with the most restrictive lockdown measures – there was an increase in charge volumes, up 7% on the first quarter (April to June 2019) in the previous year.

These aggregate changes in charge volumes mask large crime type variations. On the one hand, there were large reductions in charges for many theft offences, with shoplifting seeing the largest percentage fall of all offences in the year ending March 2021 (down 40%). On the other hand, there were also a wide range of other offences – both victim and state-based – that recorded marked increases in charge volumes during the same period. Offences which recorded increases in charges of 10% or more included rape, arson, trafficking of drugs, vehicle interference and other sexual offences.

Forecasting analysis (ARIMA) was used to explore whether the changes in monthly charge volumes during COVID-19 lockdowns diverged from expected trends, for both the first 3 months of lockdown (April to June 2020) and for the year ending March 2021.

The biggest impact on charge volumes was in the first 3 months, when lockdown restrictions were most severe; many offence types saw statistically significant increases in charge volumes for the 3-month period ending 30 June 2020, which were above the forecast trend. The list of offences with above forecast trends was diverse, covering both serious and less serious offences against both victims and the state. Shoplifting was the only offence sub-category to see charges significantly below the forecast in this quarter.

The impact on charges was less pronounced across the full year. Shoplifting was the only offence sub-category to see statistically significant changes for the 12-month period to March 2021, with charges below the forecast trend.

Discussion

This analysis has drawn on statistics on changes in police recorded crime and charge volumes, along with forecasting analysis to explore evidence of a workload effect. The analysis has taken advantage of the changing conditions created by COVID-19 restrictions and its impact on crime demand, as a ‘natural experiment’ to understand the impact of changes in police workload on charges. This builds on other work which has looked at the existence of a workload effect – that higher police crime workloads result in fewer crimes being cleared.

Qualitative evidence of the impact of demand changes on charge volumes – from various inspectorate reports covering how the criminal justice system (CJS) dealt with COVID-19 – provides helpful supporting evidence of a workload effect during this period. These find that both police forces and prosecutors reallocated resource to tackle their respective charging backlogs during COVID-19 lockdowns.

However, not all of the increases in charge volumes recorded between 1 April to 30 June 2020 were about backlog clearing. There is good evidence that some police forces used part of their ‘resource dividend’ to undertake more proactive policing, such as higher levels of stop and search. These are likely to have contributed to the substantial increases in drugs possession charges in the first lockdown quarter.

Fewer charges for shoplifting offences provide evidence of the importance of crime availability as a factor in determining changes in charge volumes. Reductions in recorded crime volumes for some offences are associated with lower charge volumes.

The analysis also provides evidence of a ‘reverse workload effect’ whereby short-term falls in crime demand are associated with short-term increases in charge volumes for selected offences. Unlike the ‘overload effect’, where increases in crime demand disproportionately suppress charges for less serious offences, the picture observed here is more complex. The ‘resource dividend’ from COVID-19 restrictions appears to have been spread across both serious and less serious offences, with the increase in charges in the latter partly driven by more proactive policing.

Inspectorate findings also suggest that the Crown Prosecution Service (CPS) reallocated resources to process cases referred by the police for a charging decision, due to a reduction in the court demands on prosecutors.

Conclusions

Overall, COVID-19 lockdowns presented a ‘natural experiment’ to better understand what happens to charges when crime demand, and therefore police workload, falls. The analysis significantly enhances the limited evidence base on the relationship between police workload and case clearance. Two contrasting effects played out in relation to changes in charge volumes.

Firstly, a ‘reverse workload effect’ appeared to take place in the first 3 months of COVID-19 lockdowns (April to June 2020), whereby marked, short-term falls in crime workload led to an increase in charge volumes. These took place across a diverse range of crime types covering violence, sexual, property and state-based offences (for example, drugs and public order).

The spread of crime types affected, and qualitative evidence, suggest 3 separate factors led to these changes. Firstly, the police reallocated resources to clear backlogs of open cases; secondly, CPS resources were reallocated to deal with police referred cases in the wake of court closures. Finally, more proactive policing – principally higher levels of stop and search – resulted in more charges for drugs possession offences.

Secondly, for some crime types, charge volumes fell sharply once COVID-19 restrictions were imposed due to precipitous falls in some recorded crimes. Shoplifting charge volumes recorded statistically significant falls compared with forecast volumes in both the first 3 months of COVID-19 restrictions (April to June 2020) and the 12 months to 31 March 2021. This highlights the importance of ‘crime availability’ in understanding changes in downstream pressures on the CJS

1. Introduction and aims of the study

COVID-19 lockdowns had important implications for crime and policing in England and Wales. Some COVID-19 impacts directly affected resources, for example, through officer abstraction due to infection or isolation (HMICFRS, 2021a). There were also substantial changes in demands placed on the police, and the way the police engaged with the public, with the police being made responsible for enforcing COVID-19 regulations (NPCC, 2021). It has also been suggested that how police agencies operated changed during the pandemic as a response to the risks posed by COVID-19, such as limiting their contact with the public through measures including reducing routine traffic stops (Neyroud et al., 2021). And as with many organisations, ways of working also changed, with more remote working. There were also marked changes in the nature of crime because of imposing restrictions (ONS, 2021a; Langton et al., 2021). As the Criminal Justice Joint Inspection (CJJI) noted, these had far-reaching impacts on the nature of policing demands (CJJI, 2021).

This paper, based on work supported through HM Treasury’s Shared Outcome Fund^{[footnote 1]}, examines changes in police recorded crime, charge volumes and charge-crime ratios during the first 12 months of the pandemic, to 31 March 2021. Previous research has suggested that changes in police workload may be one factor associated with changes in clearance patterns (see Prince et al. (2021) for an overview). Roberts and Roberts (2016) presented 2 linked hypotheses on the relationship between changes in demand and case clearance. The first is simply that police workload is inversely related to clearance. As more recorded crimes are assigned to each individual officer, each case is given less investigative time and has fewer allocated resources, leading to a reduced likelihood of securing a clearance. This has been labelled the ‘production-function’ perspective. The second – the ‘overload hypothesis’ – also predicts an inverse relationship between workload and clearance but additionally suggests that the police respond to this by reducing resources directed toward the high-volume, less serious offences. Resources are instead protected for the investigation of more serious offences (see Roberts & Roberts (2016) for a helpful summary of the theory and related evidence).

Given that COVID-19 had a marked influence on police workload – at least in terms of ‘crime demand’ – it is helpful to explore what happened to police recorded crime and charge volumes during the main lockdown periods. Examining data on charge volumes over the COVID-19 lockdown periods provides some insight – as a form of ‘natural experiment’ – into how changes in demand can affect police – and prosecutor – resourcing and, as a result, charge volumes. We also explore police recorded crime and charge volumes using auto-regressive integrated moving average (ARIMA) time-series analysis to identify whether changes in charge volumes due to COVID-19 are above or below forecast values, and whether the differences are statistically significant. In simple terms, ARIMA is a statistical technique that seeks to model forecast values using historic time-series data and helps establish whether actual and forecast values differ within different statistical parameters.

2. Data and analytical approach

The data drawn on in this paper is police recorded crime (Home Office, 2022a) and outcomes data (Home Office, 2022b) collected by territorial forces in England and Wales. The main outcomes covered in this paper are charges/summons, referred to herein as ‘charges’. Appendix A includes information on trends of out-of-court disposals over the COVID-19 lockdowns. The analysis excludes crime outcomes data for Greater Manchester Police (GMP) and one non-metropolitan force. The analysis also excludes fraud and computer misuse offences. See Appendix B for further information on the data used in the study.

Much of the descriptive analysis presented in this paper focuses on 12 months from 1 April 2020 to 31 March 2021. This period broadly coincides with what has been described as the first 3 national lockdowns (see table 1)^{[footnote 2]}, and some of the most marked effects on police recorded crime^{[footnote 3]}. Additional analysis was undertaken specifically on the 1 April to 30 June 2020 quarter (quarter 1, year ending March 2021)^{[footnote 4]}, which largely coincided with the most severe restrictions in the first national lockdown.

We conducted descriptive analysis to summarise what happened to crime and charge volumes during the COVID-19 lockdowns. This data was further broken down by offence groups and sub-category to explore offence type variation.

While descriptive analyses of changes in crime and charge volumes and ratios can generate insight into patterns and trends during pandemic restrictions, this approach has limitations. It is difficult to establish with certainty what changes in charge volumes might simply result from an already diminishing trend or natural variations and seasonal effects, which are well known to be present in crime. Statistical techniques are available to identify statistically significant shifts in time series of data, controlling for existing trend and seasonal patterns observed in the historical data. ARIMA analysis is one such approach that has been applied previously to crime patterns and trends (Payne & Morgan, 2020; Estévez-Soto, 2021; Langton et al., 2021). Historic data is used to construct forecast values for the period of interest (in this case, COVID-19 lockdowns). If the observed values in the period of interest sit outside the confidence intervals (CI) of the forecast values, we can be more confident that the change is not due to natural variation or seasonal effects. As part of this study, we conducted ARIMA analysis on police recorded crime and charge volumes derived from monthly Home Office data consistent with published figures (see Appendix C for further information).

Table 1: Timeline of COVID-19 lockdowns from March 2020 to July 2021

Months	Relevant dates and events
March 2020	23/03/2020 First national lockdown was announced for all countries in the UK (England, Wales, Scotland and Northern Ireland). All ‘non-essential’ high street businesses were closed. People were ordered to stay home, permitted to leave for essential purposes only (for example, buying food or medical reasons).
May 2020	10/05/2020 The first phase of lifting some lockdown restrictions; however, timings varied between England and Wales (the dates for the rest of the timeline refer to England, although Wales followed a similar pattern). The work from home order was eased for people who could not work from home (for example, manual workers).
June 2020	15/06/2020 Non-essential shops could reopen. 23/06/2020 More restrictions were eased (including relaxation of the 2m ‘social distancing’ rule).
July 2020	04/07/2020 More easing of restrictions across the rest of England. Hospitality venues could reopen (with some guidance in place).
August 2020	14/08/2020 Entertainment venues could reopen.
November 2020	05/11/2020 The second lockdown in England legally came into force.
December 2020	02/12/2020 The second national lockdown ended, returning to the 3-tier system across England.
January 2021	06/01/2021 Third and final lockdown in England was announced.
February 2021	22/02/2021 PM and other ministers began constructing a road map of lifting the lockdown.
March 2021	08/03/2021 The first stages of lifting lockdown began, such as schools reopening. 29/03/2021 Six people could meet up in private and public venues.
April 2021	12/04/2021 Non-essential shops, hospitality venues and entertainment venues could reopen (with guidance in place).
May 2021	17/05/2021 More restrictions were lifted, allowing more people to meet up in public, outdoor and indoor venues.
July 2021	19/07/2021 Step 4 was completed, lifting all lockdown restrictions, and officially ending the third national lockdown.

3. Trends in police recorded crime during COVID-19 lockdowns

Examining police recorded crime volumes before and during the COVID-19 period is a good starting point for showing how demand on the police changed during the various lockdowns. Crime demand is only one, albeit substantial, part of all the workload that the police deal with (see, for example, Laufs et al. (2021) for an overview of different types of police demand). However, it has particular importance in our understanding of the relationship with crime outcomes. Crime demands on the police have generally been the focal point of studies which have sought to explore the relationship between police workload and crime outcomes (for example, Roberts & Roberts, 2016).

Recorded crime has risen steadily since the year ending March 2016 due to many factors including: improvements in recording processes, the addition of new offences to police recorded crime, increased reporting for some offences (for example, for sex offences) and a ‘genuine’ increase in some types of crime (for example, serious violence) (see figure 1) (HMICFRS, 2021b, ONS; 2021a). However, between years ending March 2020 and 2021, there was a 13% decrease in police recorded crime.

Figure 1: Quarterly police recorded crime volumes (excluding fraud and computer misuse), year ending March 2016 to year ending March 2022

Total police recorded crime – excluding fraud and computer misuse – fluctuated across the period corresponding to COVID-19 restrictions (see figure 2)^{[footnote 5]}. There were substantial falls in April 2020 compared with April 2019 (down 27%), broadly coinciding with the start of the first national lockdown. Police recorded crime started to increase as lockdown restrictions eased, and remained stable over the summer months of 2020, when there were fewer restrictions in place. There were further declines towards the end of 2021 during the second and third lockdowns. Police recorded crime appears to have returned to, and remained close to, pre-pandemic levels since national restrictions ended in July 2021.

Figure 2: Monthly police recorded crime volumes (excluding fraud and computer misuse), year ending March 2020 to year ending March 2022

The initial fall in recorded crime has largely been attributed to COVID-19 restrictions which limited people’s movement, closed retail establishments and the night-time economy, and encouraged people to work from home. The suggestion is that changes in routine activities and a reduction in crime opportunities have driven the fall in recorded crime (Cohen & Felson, 1979). For example, fewer open shops limited opportunities for retail theft, and closure of bars and pubs reduced opportunities for alcohol-related crime and disorder. The decline in certain crime types, such as burglary and vehicle crime, is likely to reflect the greater presence of people in residential areas during the daytime, increasing the number of ‘capable guardians’ (ibid).

The more detailed impact of COVID-19 restrictions on recorded crime has been documented both in official statistics and in the academic literature (for instance, ONS (2021a), Estévez-Soto (2021) and Langton et al. (2021)). While total police recorded crime volumes reduced during the COVID-19 period, the pattern varied considerably across offence types. The COVID-19 effect on the recorded ‘crime mix’ is best summarised in terms of large falls in theft offences and robbery (both down by around one-third, year ending March 2021 compared with the previous year), with more modest reductions in sex offences (down 10% across the same period) and criminal damage. Violence offences generally remained flat, while modest increases were recorded for possession of drugs, trafficking of drugs and public order offences. Increases in possession of drugs offences are likely to result from increased stop and search by some forces over the first lockdown period. The Metropolitan Police Service (MPS) reported that drug-related stop and searches more than trebled (up 257%) from 20,147 in quarter 1 (April to June) year ending March 2019, to 72,008 in quarter 1 year ending March 2021 (MOPAC, 2021). The combined effect was a sudden and marked shift in the police recorded ‘crime mix’ during the year ending March 2021 compared to previous years, as illustrated in figure 3 (see Appendix D for figures broken down by offence sub-categories).

Figure 3: Percentage change in police recorded crime volumes by offence groups from year ending March 2020 to year ending March 2021

Existing analysis of police recorded crime during COVID-19 lockdowns

Langton et al. (2021) used ARIMA time-series analysis to compare expected and observed monthly crime rates in England and Wales for 13 different offence categories and anti-social behaviour (ASB) captured in police.uk data for the 6 months from when the pandemic began. Monthly police.uk crime data was collated from March 2015 to February 2020 to train the model to produce the ‘expected’ 6-month trend, with 95% CIs from March to August 2020. They found that 12 offence categories^{[footnote 6]} experienced a sharp decline in the first month of lockdown followed by a recovery as restrictions were eased. The observed trends for these offences fell below what ARIMA forecasted over the period. However, the pattern varied by offence category. For example, property crimes made a slower recovery, while violence and sexual offences returned to near pre-pandemic levels towards the end of the 6-month period. The 2 exceptions to these declining trends were drug offences and ASB incidents, which experienced increases during the lockdown period relative to the expected trends. Some forces recorded breaches of COVID-19 regulations as ASB incidents, contributing to the increase over the COVID-19 period (Halford et al., 2022). Langton et al. (2021) argue that the increase in recorded drug offences over the period likely reflects a shift towards proactive policing in the absence of typical levels of reactive police demand. For completeness, we have reported on ARIMA models trained on monthly Home Office police recorded crime data from April 2009 to February 2020 using 23 more detailed offence sub-categories (see Appendix E).

Weighting recorded crime for severity

Although changes in police recorded crime provide a helpful perspective on change in crime demand, arguably a more meaningful way of assessing change in police workload over time is to weight police recorded crime using the Office of National Statistics (ONS) Crime Severity Score (see figure 4)^{[footnote 7]}. Prior to the pandemic, crime workload – as measured by severity-weighted crime – increased at a faster rate than unweighted crime. Between years ending March 2014 and 2019, total recorded crime increased by 45%, but crime weighted by the ONS Crime Severity Score went up 69%. Although the increase reflects improved recording of crimes – particularly sex offences – some of the increase will reflect genuine increases in workload (for example, growth in recorded sex offences due to increased reporting in the wake of Operation Yewtree, and some genuine increases in serious violence). During the COVID-19 period, severity-weighted crime fell sharply, down 12% between years ending March 2020 to 2021.

Figure 4: Crime Severity Score weighted police recorded crime and unweighted police recorded crime per 1,000 population in England and Wales, year ending March 2003 to year ending March 2022 (ONS, 2023)

4. Trends in charge volumes

Charge volumes had been steadily declining in the period up to the pandemic, falling by 25% between years ending March 2015 to 2020 (see figure 5). That trend appeared to reverse during and – somewhat surprisingly – immediately before lockdown restrictions. Figures for quarterly charge volumes show a marked increase in charges in quarter 4 (January to March) of year ending March 2020^{[footnote 8]}, and that continued into quarter 1 (April to June) of year ending March 2021. However, these 2 consecutive quarters appear to be quite different in terms of what is driving their respective increases. The former was largely due to one large metropolitan force clearing a backlog of un-outcomed cases from earlier that year (a similar effect was found for out-of-court disposals – see Appendix A).

The latter, however, was a more widespread phenomenon. Almost half of forces (20 out of 41) saw net increases in charge volumes between quarter 1 (April to June) year ending March 2020 and quarter 1 year ending March 2021, and, as we shall consider below, also involved a marked change in the ‘charge mix’. This increase in charge volumes was not sustained and was followed by a marked fall in charge volumes in the remaining 3-quarters of the year ending March 2021. Comparing annual totals between years ending March 2020 and 2021, charge volumes fell by 5% (from 412,245 to 389,580).

Figure 5: Quarterly charge volumes, year ending March 2015 to year ending March 2022

COVID-19 restrictions appear to have brought about some major shifts in charge mix (see Appendix F for figures broken down by offence sub-categories). In total, 10 offence sub-categories recorded absolute increases in charge volumes during the year ending March 2021. However, 14 sub-categories showed reductions across the same period. Offence sub-categories that saw the largest absolute increases in charge volumes during the year ending March 2021 were violence without injury (+4,258), public order offences (+2,722) and drugs trafficking (+2,815). Offence sub-categories that saw the largest absolute decreases in charge volumes during the same year were shoplifting (-26,322), other theft offences (-2,703) and violence with injury ( 2,530). Shoplifting saw the biggest percentage and absolute reduction in charge volumes, and accounted for 68% of the total fall in charges in the year (see figure 6)^{[footnote 9]}. While theft offences saw the biggest percentage falls in charges, offences that recorded large increases were more diverse, and included offence sub-categories of rape, arson, drugs trafficking and vehicle interference. While offences that saw the largest percentage falls in charges also saw marked falls in recorded crime, that pattern did not hold for offences that saw large percentage increases in charges.

Figure 6: Percentage change for offence sub-categories that saw the largest increases and falls in charge volumes (10% or more) from year ending March 2020 to year ending March 2021

We consider what these patterns mean for understanding the relationship between changes in workload and changes in crime outcomes in the discussion section. However, what is apparent is that during the pandemic these changes in charge mix led to a shift towards typically charging more serious offences. This can be clearly seen in figure 7, which weights charge data for each offence sub-category by crime severity.

Figure 7: Charge volumes weighted by ONS Crime Severity Score, year ending March 2015 to year ending March 2022

Finally, in this section, we consider the changing relationship between charges and recorded crimes. Overall, the combination of increased charges and lower recorded crime during pandemic lockdowns led to increases in the ‘charge-crime ratio’ – the number of charges recorded in a period divided by the number of crimes recorded in the same period. The charge-crime ratio showed a modest increase during the year April 2020 to March 2021, up from 8.3 to 9.3. Looking at quarterly changes (figure 8), there was a pronounced peak in quarter 1 (April to June) of that year, when the charge-crime ratio reached 11.6. This was against a backdrop of year-on-year falls in the charge-crime ratio since the year ending March 2016. Appendix G provides details on changes in the charge-crime ratio by offence sub-category.

Figure 8: Charge-crime ratio – the number of charges recorded in a period divided by the number of crimes recorded in the same period – years ending March 2015 to 2022

5. Forecasting (ARIMA) analysis of charge volumes

Charge volumes during the COVID-19 period can be compared against forecast estimates (see Appendix C for more information). This allows for the identification of any statistically significant differences between actual and expected charges (that is, forecast), based on the properties of pre-pandemic time-series data. Analysis of the percentage shift in charge volumes against ARIMA forecasts were first undertaken for the first 12 months of the COVID-19 pandemic (April 2020 to March 2021). Annual forecasts for this period were generated using monthly England and Wales^{[footnote 10]} charge data for the period April 2009 to December 2019 for each of the 23 offence sub-categories. Quarter 4 2020 data (January to March) was excluded fully from the data used to train the model due to irregular patterns in the data for this quarter^{[footnote 11]}. The resulting forecasts covered the full period January 2020 to March 2021, but cumulative deviations from forecast were measured for the period April 2020 to March 2021 only.

Figure 9 illustrates the ARIMA forecast for shoplifting, showing actual monthly charge data, March 2016 to March 2021, the forecast charge volumes for the first year of COVID-19 lockdowns with 80% and 95% CIs. The red line represents the expected number of monthly charges for shoplifting forecast from historic data. The blue line shows the actual observed monthly charges before and after the first lockdown.

This analysis shows actual shoplifting charges largely below the 95% CI and wholly below the 80% CI during the year ending March 2021. Shoplifting was the only offence sub-category that saw a statistically significant percentage decrease from forecast charges cumulatively over the full 12-month period April 2020 to March 2021 (see figure 12).

The impact of lockdown restrictions on shoplifting charges was sudden, with a statistically significant decline of over 1,000 charges in the first full month of national lockdown (see figure 9). In April 2020, shoplifting charges volumes were 23% below the expected volume. This is due to there being fewer shoplifting crimes available to charge – a point we return to in the discussion section. Volumes remained well below expected figures in the 12 months following the first lockdown. There were some signs of recovery as restrictions eased. However, charges fell again in the second and third lockdowns.

Figure 9: Monthly observed shoplifting charge volumes from March 2016 to March 2021, and forecast charge volumes for shoplifting for the 12 months after first lockdown, April 2020 to March 2021

Legend:

• observed monthly charges (blue line)
• expected charges (red line)
• confidence (prediction) intervals at 80% (dark grey) and 95% (lighter grey)

Figure 10 illustrates the ARIMA forecast for the offence sub-category ‘other sexual offences’ (excluding rape), also using 80% and 95% CIs. In contrast to shoplifting, charges for other sexual offences were above expected levels in the first few months of lockdown. In fact, in April 2020, charges for other sexual offences were 31% above the forecasted level. As we consider in the discussion section, there is evidence to suggest that both the police and CPS were addressing the charging backlog over this period. And, unlike shoplifting, charges started to get closer to the expected trend over the 12-month period. Monthly ARIMA forecasts for charge volumes by offence sub-category are provided in Appendix E.

Figure 10: Monthly observed other sexual offences charge volumes from March 2016 to March 2021, and forecast charge volumes for other sexual offences charge volumes for the 12 months after first lockdown, April 2020 to March 2021

Legend:

• observed monthly charges (blue line)
• expected charges (red line)
• confidence (prediction) intervals at 80% (dark grey) and 95% (lighter grey)

We ran ARIMA models for all offence sub-categories for both 3 months and 12 months (figures 11 and 12). Over the 3-month period (April to June 2020), cumulative charge volumes relative to forecast were statistically significantly (95% CIs) above forecast levels for 11 offence sub-categories (see figure 11). Offences which saw statistically significant increases in charge volumes in this initial lockdown period were:

• other sexual offences
• bicycle theft
• arson
• criminal damage
• theft of a motor vehicle
• vehicle interference
• public order offences
• miscellaneous crimes
• violence with and without injury
• drugs possession offences

Weapons possession offences was close to achieving a statistically significant increase. Shoplifting was the only offence sub-category to see a statistically significant fall in charge volumes stemming from a collapse in police recorded crime.

Figure 11: Percentage shift in observed charge volumes to mean forecast position and 95% CI(s) over 3-month period April to June 2020

Notes:

1. Any offence to the left of the blue vertical line has lower than predicted charges.
2. Any offence to the right of the blue vertical line has higher than predicted charges.

Figure 12 shows the cumulative charges against forecast for the full 12-month period (April 2020 to March 2021). Shoplifting was the only offence that saw statistically significant (95% CIs) decreases against the forecast level. In contrast to the initial 3-month analysis (figure 11), no offence sub-categories saw statistically significant increases above the forecast level (although vehicle interference was borderline while several other offences such as violence without injury were close to statistical significance at this level).

Figure 12: Percentage shift in observed charge volumes to mean forecast position and 95% CI(s) over 12-month period April 2020 to March 2021

Notes:

1. Any offence to the left of the blue vertical line has lower than predicted charges.
2. Any offence to the right of the blue vertical line has higher than predicted charges.

6. Discussion

This paper has sought to draw together a combination of descriptive and forecasting analysis to better understand the relationship between changes in crime demand and crime outcomes during COVID-19 lockdowns. The period was chosen as it provides a ‘natural experiment’ to explore what happens to crime outcomes when crime demand falls. The observed pattern is complex, but it does provide some helpful insights into the factors that influence charge volumes, and in turn, demand on the downstream criminal justice system (CJS).

There is good evidence that in England and Wales, along with many other countries, COVID-19 restrictions had a profound effect on crime demand. Previous studies (Langton et al., 2021) and further analysis included here have found that for most victim-based crimes, these reductions were statistically significant, with particularly large reductions found for theft offences. The ARIMA forecasting analysis in this paper provides good evidence of how the COVID-19 restrictions also impacted upon charge volumes. Just under half of all the offence sub-categories showed statistically significant increases in charge volumes during intense lockdown measures in April to June 2020. Those impacts all but disappeared when the analysis was extended to 12 months to March 2021^{[footnote 12]}. Consistent across both periods, shoplifting was the only offence to see statistically significant falls in charge volumes. Table 2 summarises the main offence sub-category picture for all the key measures reviewed in this paper.

Table 2: Summary of changes in recorded crime, charge volumes, charge-crime ratios and ARIMA (forecast) analysis by offence sub-category, year ending March 2020 to year ending March 2021

Offence sub-category	Recorded crime	Charge volumes	Charge-crime ratio	Forecasted police recorded crime (4 months)	Forecasted police recorded crime (12 months)	Forecasted charges (3 months)[footnote 13]	Forecasted charges (12 months)[footnote 14]
Arson	Down	Up	Up	No significant change	No significant change	Above forecast	No significant change
Bicycle theft	Down	Down	Up	Below forecast	No significant change	No significant change	No significant change
Burglary	Down	Down	Up	Below forecast	Below forecast	No significant change	No significant change
Criminal damage	Down	Up	Up	Below forecast	No significant change	Above forecast	No significant change
Homicide	Down	Down	Up	No significant change	No significant change	No significant change	No significant change
Miscellaneous crimes	Up	Up	Up	No significant change	No significant change	Above forecast	No significant change
Other sexual offences	Down	Up	Up	Below forecast	Below forecast	Above forecast	No significant change
Other theft offences	Down	Down	Up	Below forecast	Below forecast	No significant change	No significant change
Possession of drugs	Up	Down	Down	Above forecast	No significant change	Above forecast	No significant change
Possession of weapons offences	Down	Down	Up	Below forecast	Below forecast	No significant change	No significant change
Public order offences	Up	Up	Up	Below forecast	No significant change	Above forecast	No significant change
Rape	Down	Up	Up	Below forecast	Below forecast	No significant change	No significant change
Robbery of business property	Down	Down	Up	Below forecast	Below forecast	No significant change	No significant change
Robbery of personal property	Down	Down	Up	Below forecast	Below forecast	No significant change	No significant change
Shoplifting	Down	Down	Down	Below forecast	Below forecast	Below forecast	Below forecast
Stalking and harassment[footnote 15]	Up	Up	Down	N/A	N/A	N/A	N/A
Theft from a vehicle	Down	Down	Up	Below forecast	Below forecast	No significant change	No significant change
Theft from the person	Down	Down	Up	Below forecast	Below forecast	No significant change	No significant change
Theft of a motor vehicle	Down	Down	Up	Below forecast	Below forecast	Above forecast	No significant change
Trafficking of drugs	Up	Up	Down	Borderline	Borderline	Above forecast	No significant change
Vehicle interference	Down	Up	Up	Below forecast	Below forecast	Above forecast	Borderline
Violence with injury	Down	Down	Up	Below forecast	Below forecast	Above forecast	No significant change
Violence without injury	Down	Up	Up	Below forecast	Below forecast	Above forecast	No significant change

Notes:

1. Statistical significance for all forecasts apply 95% confidence intervals.

The importance of crime availability in determining charge volumes

Although this seems like a rather simple, uncontroversial observation, evidence of the impact of COVID-19 restrictions on crime demand and, in turn, a partially suppressive effect on charge volumes, is a reminder of the importance of crime availability in determining charge volumes. The collapse in shoplifting charges is largely a reflection of the fact that there were fewer shoplifting offences available to detect during COVID-19 lockdowns, independent of any police action or intervention. This pattern is almost certainly more likely to be associated with high-volume offences like shoplifting and is unlikely to hold for more serious offences. One cross-sectional study has pointed to the importance of the ‘availability’ of recorded crimes as being a factor in determining variations in outcomes per officer for burglary/vehicle crime offences within different police force areas (Burrows et al., 2005). Other studies have concluded that reductions in crime – linked to the deterrent effect of an increase in officers – may be also linked to falls in arrests (for instance Owens, 2013).

Evidence of a ‘workload’ effect on crime outcomes: backlog clearance, response and proactivity

Although the current research evidence on the relationship between changes in crime workload and charges is mixed, stronger studies – which have sought to assess changes in likelihood of clearance over time – suggest the existence of a workload effect. Higher police crime workloads result in fewer charges, with the impact of increased crime demand most felt on clearance rates for less serious crimes (the so-called ‘overload’ effect) (Roberts & Roberts, 2016). This current analysis provides additional evidence of a workload effect, albeit one which assesses the impact of a reduction in crime workload, rather than an increase.

ONS Crime Severity Score weighted recorded crime data show a marked reduction in overall severity-weighted crime during COVID-19 lockdowns in 2020 and 2021. This fall came after a period of sustained increases in severity-weighted recorded crime. As severity-weighted demand fell, charge volumes and clearance rates increased for some offences. The suggestion is that, as a consequence, this generated a ‘resource dividend’ for police, which were shifted to help increase charge volumes. However, as we consider below, this was not all about police resources.

The wide range of offences which saw statistically significant increases in charge volumes suggest that several different mechanisms were playing out. That some more serious offences – other sexual offences is perhaps the best example – saw statistically significant increases in charge volumes during the first quarter of lockdown, implies that at least part of the increase was due to backlog clearing. Median days from recording to charge for other sex offences receiving an outcome in the year ending March 2021 was 240 days. So, for this offence, the increase in charge volumes is extremely likely to be for offences already in the system.

Considering some serious offences saw statistically significant increases in charges also raises a question whether any backlog clearance was driven by police or prosecutor action, or a combination of the two^{[footnote 16]}. Given that most courts had closed due to COVID-19, prosecutor capacity was reported to have been freed up, creating a further ‘resource dividend’ (HMCPSI, 2020). Other evidence points to a similar prosecutor effect.

Evidence from inspectorate reports

Reports by the various inspectorates on the impact of COVID-19 restrictions – an HMICFRS report on policing the pandemic, 2 HMCPSI reports on the CPS response to COVID-19 and a joint review from the Justice Inspectorate – provide supporting evidence on how both the police and CPS responded to changing demands on resources (HMCPSI, 2020; HMCPSI, 2021; HMICFRS, 2021a; CJJI, 2021). Taking these inspectorate reports together provides evidence of charging backlog clearing by both the police and CPS. In the initial stages of lockdown, some forces chose not to pursue any warrants for arrest, or any live investigations that could result in arrest. But that position was temporary and, overall, custody suite throughput was found to have remained steady throughout lockdowns (HMICFRS, 2021a). That inspection also found forces managed investigative demand without changing their resources (although many had contingency plans to deal with possible higher rates of COVID-19 absence). HMICFRS did, however, identify 11 forces that moved resources to investigative functions to ensure continued effective operation.

The HMICFRS inspection also noted that, because of changes in the nature of demand, investigators could focus on clearing their outstanding investigations. They not only reduced backlogs but, with increased supervision, specialist guidance and support, improved the quality of investigations. Other forces saw the changes in demand as an opportunity to ‘clear up’ cases with outstanding named suspects (HMICFRS, 2021a). This is supported by evidence from HMCPSI inspections that some CPS areas experienced increased requests for pre-charge advice from the police as ‘they cleared the decks’ during the first lockdown (April to June 2020) (HMCPSI, 2021; p.18).

An inspection report by HMCPSI on the CPS response to COVID-19 during the first lockdown revealed that, due to court closures and reduction in court sittings, there was increased CPS capacity and resource available for casework and charging (HMCPSI, 2020). It was reported that priorities were set on clearing backlogs of pre-charge cases, with an initial focus on cases where the charging request was more than 28 days old. As a result of this reallocation of resource, the backlog of pre-charge requests of these cases more than halved, from 1,800 at the end of March to 800 by early May 2020^{[footnote 17]}. Additionally, it was reported that magistrates’ court teams could clear tasks not usually addressed due to limited resources. In all CPS areas, HMCPSI reported task numbers as the lowest they had ever been.

As well as tackling their own pre-charge backlog, CPS reallocated resources to deal with the increasing number of charging receipts being received from the police in some CPS areas (HMCPSI, 2021). Crown advocate resources were also reported to have been redeployed from delayed Crown Court cases to charging reviews to manage the increase of receipts from the police. It found that “all CPS areas had prosecutors who could focus on clearing charging receipts and, in some, charging backlogs, because magistrates’ courts were closed or had reduced sittings” (HMCPSI, 2021; p.17). As a result, HMCPSI reported that the timeliness of charging decisions had improved.

The HMICFRS inspection also explored whether there were any plans for greater use of out-of-court disposals during the pandemic restrictions. The thinking here was that, where appropriate, out-of-court disposals (OOCDs) might facilitate less suspect contact, take pressure off the courts, and could lead to swifter justice compared to charging offences. The inspection identified some forces as actively evaluating this approach. At least one force worked with CPS colleagues to review cases awaiting trial to establish if any were suitable to be dealt with by an OOCD.

That charges for some less serious offences – where responsibility for the charging decision would have rested solely with the police – increased over the period suggests that there was a distinct ‘police effect’ helping to drive up charge volumes (that is, additional charges were not simply the result of reallocated CPS capacity). Public order offences, criminal damage and theft of a vehicle are all offences where the police would be expected to make most charging decisions. And all these offences saw statistically significant increases in charges in quarter 1 (April to June 2020) of the first COVID-19 lockdown.

It is also possible that, for some offences – especially those which are less serious – a critical ingredient driving up charges is simply a better initial response to, and investigation of, offences which were initially recorded during COVID-19 lockdowns.

Proactive policing

Finally, there is evidence for some offences that part of the increase in charge volumes arose from more proactive policing due in part to the ‘resource dividend’ from falling crime demand. Drugs offences are probably the best example of an offence which largely depends on police-led activity. Drugs trafficking and possession offences both saw statistically significant increases in charge volumes in quarter 1 (April to June) year ending March 2021. The likely explanation is that this increase resulted from some forces – MPS in particular – greatly increasing levels of stop and search during the first full three months of lockdown. In MPS, there was a reported 65% increase in stop and search from January to March 2020 to the first quarter of COVID-19 lockdowns - April to June 2020 (MOPAC, 2021).

Conclusion

The analyses presented here provide supporting evidence of the importance of changes in workload on charges and case clearance. But this remains a complex area. The COVID-19 restrictions did not solely affect crime demand on the police – they impacted police capacity through COVID-19 abstractions; the police were responsible for enforcing lockdown regulations; and, for many reasons, there were increases in incidents reported as ASB (sometimes for gatherings perceived as breaching regulations). Critically too, from the perspective of capacity and flows through the upstream CJS, COVID-19 had consequences for the operation of courts and, as a result, prosecutors. Figure 13 attempts to illustrate some of the demand-charge changes which took place during COVID-19 restrictions. Although there is still much to learn about the relationship between crime demand, workload and crime outcomes, the lockdowns provide some helpful insights into a poorly understood evidential area.

Figure 13. An illustrative model of the national changes in police recorded crime and its potential relationship with charges during the COVID-19 period

Appendix A: The use of out-of-court disposals for notifiable offences in England and Wales during COVID-19

This Appendix examines the use of out-of-court disposals (OOCDs) for notifiable offences during quarter 1 (April to June) of the year ending March 2021, which coincides with the first COVID-19 lockdown. It does not cover the use of Fixed Penalty Notices (FPNs) issued for breaches under the Health Protection Regulations (HPR) 2020, which are non-notifiable offences. OOCDs are outcomes which the police can use as an alternative to charges (and prosecution through the courts). They are used predominantly for less serious offences. Some OOCDs come with interventions and support for people showing early signs of criminal behaviour. They can potentially reduce re-offending (Home Office, 2023).

OOCDs can be formal or informal. Formal OOCDs are adult or youth cautions or Penalty Notices for Disorder (PNDs). Informal OOCDs are Community Resolutions or cannabis/khat warnings. A diversionary outcome involves diversionary, educational or intervention activity. To put in context, formal OOCDs accounted for 1.4% of total outcomes for the year ending March 2021 while informal OOCDs accounted for 3% of all outcomes in the same year (ONS, 2021b; table 2.1).

The use of OOCDs was examined during the first COVID-19 lockdown by comparing quarter 1 (April to June) of year ending March 2021 with quarter 1 of the previous year. This takes place against the background of the move to simplify OOCDs^{[footnote 18]}.

OOCD volumes increased markedly in the first COVID-19 quarter (April to June 2021). Overall, there was a 21% increase in OOCD outcomes^{[footnote 19]} recorded in April to June 2021 compared to the same quarter in the previous year. Most of this increase was due to an increase in informal OOCDs (Community Resolutions / warnings) for state-based offences (see figure A1). These increased by 68% across the 2 quarters.

Figure A1: Percentage change in formal and informal victim and stated-based OOCDs from Q1 (April to June) year ending March 2020 to Q1 year ending March 2021

In terms of crime types, the increase in OOCDs for drug possession offences during quarter 1 (April to June) of year ending March 2021 was especially marked, with a net increase of over 11,000. This represented an increase of 78% over the equivalent 3 months of 2020. The largest decrease in OOCDs was, predictably, found for shoplifting offences. These fell in by more than 2,500 across the period, a fall of 48%.

Figure A2: Volume change in formal and informal victim and state-based OOCDs from Q1 (April to June) year ending March 2020 to Q1 year ending March 2021, for selected offence sub-groups

In terms of the force breakdown of the change in informal OOCDs for drugs offence, all but 5 police forces saw an increase in the number of community resolutions for cannabis possession during quarter 1 (April to June 2020). However, it was MPS that accounted for the greatest share of the increase. Informal disposals for cannabis possession increased by 136% in MPS during this period, while formal OOCDs for cannabis possession increased by 119%. The first COVID-19 lockdown saw MPS substantially increase stop and search activity. There were 72,008 drug-related stop and search records in the first COVID-19 quarter (April to June 2020), an increase of 60% from the previous quarter (MOPAC, 2021).

Although some of the increase in OOCDs in April to June 2020 resulted from an increase in offences which had taken place during this quarter, part of the increase was due to OOCDs being given for offences which had taken place in previous quarters. This might reflect some backlog clearing similar to that identified for charges. Generally, these 2 measures – total offences receiving an OOCD in quarter and offences recorded in quarter receiving an OOCD – followed each other closely. But the pattern changed around this time, with a marked increase in outcomes recorded in the quarter (see figure A3).

Figure A3: Quarterly OOCDs by OOCDs recorded in the quarter and OOCDs for offences recorded in the quarter

Figure A3 also points to a curious increase in total OOCDs recorded in quarter 4 of the year ending 2020 (January to March, that is, pre-COVID-19). When explored in more detail, this increase appears to be most concentrated in MPS. Further investigation of MPS data on OOCDs suggests that the increase in quarter 4 reflected MPS recording OOCD outcomes for offences – principally drugs offences – that took place in the earlier part of year ending March 2020 (quarter 1, April to June, to quarter 3, October to December).

Appendix B: Further information on the data used in the analysis

Data on police recorded crime is collated by the Home Office from all 43 territorial forces in England and Wales. This data covers all recorded ‘notifiable’ offences, which are crimes that the police have reported to the Home Office to compile national crime statistics. Police forces record the data in accordance with Home Office Counting Rules and the National Crime Recording Standard to ensure consistency across forces (Home Office, 2022).

Non-notifiable offences typically cover low level offending and are dealt with by magistrates’ courts or by the police issuing a Penalty Notice for Disorder (PND) or a Fixed Penalty. They include offences which might be generally described as anti-social behaviour (ASB) such as littering, begging and drunkenness. Breaches of the Coronavirus Act 2020 and Health Protection (Coronavirus) Regulations 2020 were all non-notifiable offences.

All recorded crimes are, in time, assigned one of 22 outcomes which reflect how the police close the case (for example, charge/summons, caution, investigation complete – suspect identified, victim does not support further action). The full list of outcomes can be found in ‘Crime Outcomes in England and Wales: Technical Annex’ (see Home Office, 2021b). The Home Office introduced the current outcomes framework in April 2014. It has since been subject to minor changes.

Data on the number of charges used throughout the paper relates to the absolute number of charges recorded in any specified period (sometimes referred to as ‘old style’ charges). This covers the number of charges recorded in the quarter or year, regardless of when the offence took place. This way of presenting charge data best reflects current police activity and charging patterns in any quarter. Charge-crime ratios are calculated by dividing the number of charges recorded in a given period – regardless of when the crime was recorded – by the number of crimes recorded in the same period.

Recorded crime and charge data is also presented weighted by its offence severity score (based on the average sentence length or fine that an individual offence attracts)^{[footnote 20]}. Severity-weighted crime outputs for recorded crime, alongside the weightings by offence type, are published as experimental data by the Office for National Statistics (ONS, 2023).

Greater Manchester Police (GMP) were excluded from the analysis as it could not provide outcomes data for part of 2019, and one non-metropolitan force was also excluded as their outcomes data in 2020 was incomplete because of delays submitting returns to the Home Office. Fraud and computer misuse offences were also excluded as how fraud offences are reported and investigated differs markedly from other crimes. Fraud offences reported to the police are recorded and collected by the National Fraud Intelligence Bureau (NFIB) from Action Fraud and 2 industry bodies, Cifas and UK Finance.

Offences of domestic burglary and non-domestic burglary were combined into a single offence type of ‘burglary’, due to recording changes for these offences in 2017. Stalking and harassment offences were excluded from the statistical analysis due to a change in counting rules for these offences. From April 2020, all cases where a course of conduct is reported between a victim and their former partner must be recorded as stalking unless the police are satisfied that the matter amounts to harassment in law only. This was expected to cause an increase in offences recorded against ‘stalking’, therefore data for the year ending March 2021 is not comparable with previous years.

Appendix C: ARIMA analysis – approach and police recorded crime analysis

Auto-regressive integrated moving average (ARIMA) fits a time-series model to data comprising:

(a) varying weights being applied to past data points (the auto-regressive terms).
(b) a trend component.
(c) a residual ‘error’ component based on a simple moving average of the data.

As well as describing the properties of the time series being analysed, it is assumed that those time-series properties will persist into the future, allowing the model to generate forecasts for the time series. In this analysis, the shift in charge volumes during the COVID-19 period can be compared to forecast estimates based on data from the pre-COVID-19 period. This allows the identification of statistically significant differences to what would have been expected (that is, forecast), based on the properties of the pre-pandemic series (see Hyndman and Athanasopoulos, 2018).

The first step was to run models for each offence to determine whether ARIMA was an appropriate modelling approach for this type of data. Homicide was excluded. This is an example of an offence which is modelled very poorly by ARIMA approaches due to low volumes and a lack of predictable, seasonal variation in police recorded crime.

ARIMA models were fitted for each offence to forecast into the COVID-19 period.

For charges: the ARIMA models were trained on data from April 2009 to December 2019. This recognised that there were some inconsistencies in the January to March 2020 charge data. The forecast included 15 months from January 2020 through to March 2021, but the paper only reports on the ‘COVID-19’ deviation from the forecast for the period April 2020 to March 2021. This equates to when the impact of COVID-19 restrictions was at its greatest. For the 3-month period, the paper reports on deviation from forecast for the period April to June 2020.

For police recorded crime: the ARIMA models were trained on data from April 2009 to February 2020, recognising that COVID-19 was already affecting volumes through March 2020 (compared with April 2020, the approximate start date for COVID-19 effects in charge series). The 12-month forecast is from March 2020 to February 2021. The paper reports on deviation from forecast for the 4-month period March to June 2020. This includes March 2020 as this month had some COVID-19 impacts on recorded crime.

Police recorded crime analysis

Figures C1 and C2 show the measured deviations in cumulative recorded crime from ARIMA-generated forecasts over a 4-month and 12-month forecasting window from March 2020. Proportional differences shown reflect the percentage deviation of the central forecast and 95% prediction intervals (that is, confidence intervals (CIs) from the actual level as a proportion of the actual level). As such, these remain symmetrical around the central position. For the example of the central forecast for ‘theft from the person’, the gap between realised actuals and the central forecast was approximately twice the level of the actuals, or, in absolute terms, the actuals were only a third as large as forecast. For possession of drugs, the difference resulting actuals were approximately 1.1 times the forecast level.

During the 4 months March to June 2020 (figure C1), trafficking of drugs and drugs possession offences saw statistically significant increases above the forecast trend. Most victim-based offending saw lower volumes than forecast, with theft from the person offences seeing the largest decrease^{[footnote 21]}. The only victim-based offences not to see statistically significant falls were arson, homicide, death or serious injury – unlawful driving and miscellaneous crimes. A broadly similar picture held for the full year analysis (March 2020 to February 2021; figure C2), where trafficking of drugs was the only offence to see an above forecast increase, while several offences (bicycle theft, public order offences, drugs possession) moved to fall within forecast prediction intervals.

Figure C1: Percentage shift of police recorded crime volumes to mean forecast position and 95% CI(s) over 4-month period March to June 2020

Figure C2: Percentage shift of police recorded crime volumes to mean forecast position and 95% CI(s) over 12-month period March 2020 to April 2021

Appendix D: Police recorded crime, by offence sub-category, years ending March 2016 to 2022

Yearly totals	2015/16	2016/17	2017/18	2018/19	2019/20	2020/21	2021/22
Arson	20,327	22,129	24,097	24,665	24,452	21,538	22,931
Bicycle theft	81,304	88,759	94,677	93,112	83,555	72,940	72,960
Burglary	369,433	377,889	402,124	387,858	352,797	245,754	241,250
Criminal damage	480,116	499,573	520,228	508,059	478,134	414,465	469,631
Homicide	529	652	625	620	666	521	642
Miscellaneous crimes	59,455	73,781	86,548	96,231	98,800	102,349	104,078
Other sexual offences	66,204	75,259	90,915	97,204	97,561	85,981	115,600
Other theft offences	454,483	473,459	499,445	513,254	484,096	320,271	403,083
Possession of drugs	118,671	109,021	107,125	119,064	143,726	161,194	132,770
Possession of weapons offences	24,098	29,336	35,941	42,327	43,674	40,721	44,598
Public order offences	190,405	258,672	334,984	403,607	425,452	445,254	543,667
Rape	33,927	39,405	50,744	55,535	54,850	51,818	65,424
Robbery of business property	4,797	5,294	6,716	7,223	7,729	5,100	5,394
Robbery of personal property	42,754	49,262	63,426	70,849	75,436	49,986	55,238
Shoplifting	316,358	348,140	360,556	353,019	338,033	214,436	257,577
Stalking and harassment	150,588	202,459	274,617	417,338	468,408	591,084	668,538
Theft from a vehicle	221,469	240,229	259,900	262,201	264,754	179,636	185,878
Theft from the person	77,058	81,454	92,415	95,931	105,149	46,279	82,057
Theft of a motor vehicle	76,427	90,485	103,673	111,305	110,190	86,611	104,180
Trafficking of drugs	24,273	24,422	25,115	28,241	32,078	38,863	36,130
Vehicle interference	42,494	48,780	60,332	60,503	58,268	46,543	44,166
Violence with injury	406,489	440,981	482,257	515,842	509,725	438,686	528,889
Violence without injury	373,797	440,703	536,504	628,076	684,114	630,715	749,878
Grand total	3,635,456	4,020,144	4,512,964	4,892,064	4,941,647	4,290,745	4,934,559

Notes:

1. Data for Greater Manchester Police and non-metropolitan force is excluded from these figures.

Appendix E: ARIMA monthly forecast for police recorded crime and charge volumes by offence sub-category

Police recorded crime

Charge volumes

Appendix F: Old style charge volumes, yearly, by offence sub-category, years ending 2016 to 2022

Offence sub-category	2015/16	2016/17	2017/18	2018/19	2019/20	2020/21	2021/22
Arson	1,515	1,345	1,336	1,178	1,004	1,219	1,062
Bicycle theft	2,899	2,255	2,479	2,363	1,718	1,587	1,173
Burglary	24,895	24,005	23,316	20,994	19,884	16,308	13,096
Criminal damage	38,980	35,513	32,028	29,887	26,016	26,045	22,868
Death or serious injury – unlawful driving	558	631	609	624	630	626	540
Homicide	474	359	443	586	496	462	473
Miscellaneous crimes	20,097	20,171	18,841	16,862	16,693	17,612	15,437
Other sexual offences	10,978	10,301	9,510	7,646	8,113	8,888	8,523
Other theft offences	20,138	16,866	14,761	11,913	9,991	7,288	5,368
Possession of drugs	37,888	35,540	32,536	33,552	36,482	36,262	30,162
Possession of weapons offences	13,756	14,958	15,664	16,916	17,482	17,109	15,136
Public order offences	43,092	42,119	42,050	41,162	39,792	42,514	38,167
Rape	4,971	4,241	3,493	2,396	2,375	2,902	2,926
Robbery of business property	1,704	1,528	1,595	1,686	1,692	1,319	1,074
Robbery of personal property	6,019	5,719	5,864	5,564	5,949	5,377	4,336
Shoplifting	98,892	95,278	85,187	76,554	66,012	39,690	38,248
Stalking and harassment	20,355	21,865	23,270	24,681	23,363	26,690	24,669
Theft from a vehicle	5,891	4,768	4,701	3,933	3,895	3,093	2,157
Theft from the person	2,029	1,652	1,639	1,351	1,564	1,031	818
Theft of a motor vehicle	7,818	7,291	6,521	5,437	5,003	4,496	3,533
Trafficking of drugs	16,684	15,318	13,562	13,243	15,610	18,425	15,946
Vehicle interference	1,739	1,610	1,606	1,562	1,641	2,069	1,547
Violence with injury	87,739	79,822	72,214	67,433	57,727	55,197	47,092
Violence without injury	55,576	53,965	53,730	52,468	49,113	53,371	47,472
Grand total	524,687	497,120	466,955	439,991	412,245	389,580	341,823

Notes:

1. Data for Greater Manchester Police and non-metropolitan force are excluded from these figures.

Appendix G: Charge-crime ratio, yearly, by offence sub-category, years ending March 2016 to 2022

Offence sub-category	2015/16	2016/17	2017/18	2018/19	2019/20	2020/21	2021/22
Arson	7.5%	6.1%	5.5%	4.8%	4.1%	5.7%	4.6%
Bicycle theft	3.6%	2.5%	2.6%	2.5%	2.1%	2.2%	1.6%
Burglary	6.7%	6.4%	5.8%	5.4%	5.6%	6.6%	5.4%
Criminal damage	8.1%	7.1%	6.2%	5.9%	5.4%	6.3%	4.9%
Homicide	89.6%	55.1%	70.9%	94.5%	74.5%	88.7%	73.7%
Miscellaneous crimes	33.8%	27.3%	21.8%	17.5%	16.9%	17.2%	14.8%
Other sexual offences	16.6%	13.7%	10.5%	7.9%	8.3%	10.3%	7.4%
Other theft offences	4.4%	3.6%	3.0%	2.3%	2.1%	2.3%	1.3%
Possession of drugs	31.9%	32.6%	30.4%	28.2%	25.4%	22.5%	22.7%
Possession of weapons offences	57.1%	51.0%	43.6%	40.0%	40.0%	42.0%	33.9%
Public order offences	22.6%	16.3%	12.6%	10.2%	9.4%	9.5%	7.0%
Rape	14.7%	10.8%	6.9%	4.3%	4.3%	5.6%	4.5%
Robbery of business property	35.5%	28.9%	23.7%	23.3%	21.9%	25.9%	19.9%
Robbery of personal property	14.1%	11.6%	9.2%	7.9%	7.9%	10.8%	7.8%
Shoplifting	31.3%	27.4%	23.6%	21.7%	19.5%	18.5%	14.8%
Stalking and harassment	13.5%	10.8%	8.5%	5.9%	5.0%	4.5%	3.7%
Theft from a vehicle	2.7%	2.0%	1.8%	1.5%	1.5%	1.7%	1.2%
Theft from the person	2.6%	2.0%	1.8%	1.4%	1.5%	2.2%	1.0%
Theft of a motor vehicle	10.2%	8.1%	6.3%	4.9%	4.5%	5.2%	3.4%
Trafficking of drugs	68.7%	62.7%	54.0%	46.9%	48.7%	47.4%	44.1%
Vehicle interference	4.1%	3.3%	2.7%	2.6%	2.8%	4.4%	3.5%
Violence with injury	21.6%	18.1%	15.0%	13.1%	11.3%	12.6%	8.9%
Violence without injury	14.9%	12.2%	10.0%	8.4%	7.2%	8.5%	6.3%
Total charge –crime ratio	14.4%	12.4%	10.3%	9.0%	8.3%	9.1%	6.9%

Notes:

1. Data for Greater Manchester Police and non-metropolitan force are excluded from these figures.

References

Burrows, J. and Tilley, N. (2005) An overview of attrition patterns. London: Home Office.

CJJI (2021) Impact of the pandemic on the criminal justice system. Criminal Justice Joint Inspection. Available online [accessed 04/01/2024]

Cohen, L.E. and Felson, M. (1979) ‘Social Change and Crime Rate Trends: A Routine Activity Approach’. American Sociological Review, vol. 44, pp. 588-608.

Estévez-Soto, P.R. (2021) ‘Crime and COVID-19: effect of changes in routine activities in Mexico City’. Crime Science, vol. 10(1), pp. 15.

Halford, E., Dixon, A. and Farrell, G. (2022) ‘Anti-social behaviour in the coronavirus pandemic’. Crime Science, vol. 11(6).

HMCPSI (2020) CPS response to COVID-19: 16 March to 8 May 2020. HM Crown Prosecution Service Inspectorate. Available online [accessed 04/01/2024]

HMCPSI (2021) CPS response to COVID-19: dealing with backlogs. Crown Prosecution Service Inspectorate. Available online [accessed 04/01/2024]

HMICFRS (2021a) Policing response to COVID-19. HM Inspectorate of Constabulary and Fire & Rescue Services. Available online [accessed 04/01/2024]

HMICFRS (2021b) Reports – Rolling programme of crime data integrity inspections. HM Inspectorate of Constabulary and Fire & Rescue Services. Available online [accessed 04/01/2024]

Home Office (2016) Police recorded crime and outcomes: open data tables user guide. Available online [accessed 04/01/2024]

Home Office (2021a) Crime outcomes in England and Wales 2020 to 2021. Available online [accessed 04/01/2024]

Home Office (2021b) Crime outcomes in England and Wales: Technical Annex. Available online [accessed 04/01/2024]

Home Office (2022a) Counting rules for recorded crime. Available online [accessed 04/01/2024]

Home Office (2022b) Police recorded crime open data Police Force Area tables from year ending March 2013 onwards. Available online [accessed 04/01/2024]

Home Office (2022c) Outcomes open data year ending March 2022. Available online [accessed 04/01/2024]

Home Office (2023) Reforms to the adult out of court disposals framework in the Police, Crime, Sentencing and Courts Bill: Equalities Impact Assessment. Available online [accessed 04/01/2024]

Hyndman, R.J. and Athanasopoulos, G. (2018) ‘Forecasting: principles and practice, 2nd edition’. OTexts. Melbourne, Australia. Available online [accessed 04/01/2024]

Langton, S., Dixon, A. and Farrell, G. (2021) ‘Six months in: pandemic crime trends in England and Wales’. Crime Science, vol. 10(1).

Laufs, J., Bowers, K., Birks, D. and Johnson, S. (2021) ‘Understanding the concept of demand in policing: a scoping review and resulting implications for demand management’. Policing and Society, vol. 31(8), pp. 895-918.

MOPAC (2021) Quarterly Performance Update Report. Mayor of London Office for Policing and Crime. Available online [accessed 04/01/2024]

Neyroud, P., Maskaly, J. and Ivkovic, S.K. (2021) ‘A comparative study of police organisational changes in Europe during the COVID-19 pandemic’. Policing: A Journal of Policy and Practice, vol. 15(4), pp. 2372-2388.

NPCC (2021) Update on crime trends and penalty notices issued under COVID Regulations. National Police Chiefs’ Council. Available online [accessed 04/01/2024]

ONS (2021a) Crime in England and Wales. Office for National Statistics. Available online [accessed 04/01/2024]

ONS (2021b) Crime outcomes in England and Wales 2020 to 2021. Office for National Statistics. Available online [accessed 05/01/2024]

ONS (2022) Nature of fraud and computer misuse in England and Wales. Office for National Statistics. Available online [accessed 04/01/2024]

ONS (2023) Crime Severity Score (Experimental Statistics). Office for National Statistics. Available online [accessed 04/01/2024]

Owens, E.G. (2013) ‘Cops and Cuffs’, in Cook, P.J., Machin, S., Marie, O. and Mastrobuoni, G., eds, Lessons from the Economics of Crime: What Reduces Offending? Cambridge, MA: MIT

Payne, J. and Morgan, A. (2020) Property crime during the COVID-19 Pandemic: A comparison of recorded of recorded offence rates and dynamic forecasts (ARIMA) for March 2020 in Queensland, Australia. Available online [accessed 04/01/2024]

Prince, H., Lum, C. and Koper, C.S. (2021) ‘Effective police investigative practices: an evidence-assessment of the research’. Policing: an international journal, vol. 44(4), pp .683-707.

Roberts, A. and Roberts, J.M. (2016) ‘Crime clearance and temporal variation in police investigative workload: evidence from national incident-based reporting system (NIBRS) data’. Journal of Quantitative Criminology, vol. 32, pp. 651-647.

1. As part of the first round of Shared Outcomes Fund projects, a strand of work was undertaken on criminal justice demand, with a focus on improving understanding of the relationship between policing and prison demand. ↩

2. Certain powers are devolved to Wales including public health services and education. There was some variation between England and Wales in their approach to enforcing and lifting COVID-19 restrictions. ↩

3. The first national lockdown started on 23 March 2020. This means that 9 days of the first lockdown have been excluded from selected lockdown period and included in the pre-lockdown period data. ↩

4. This paper refers to financial quarters due to the nature of crime data reporting by the Home Office, meaning quarter 1 refers to April to June 2020, quarter 2 refers to July to September 2020, quarter 3 refers to October to December 2020 and quarter 4 refers to January to March 2021. ↩

5. Some of these fluctuations will be the result of seasonal effects on crime trends; however, most of the differences can be attributed to the introduction and subsequent easing of national lockdown restrictions throughout the period (ONS, 2021a). ↩

6. Bicycle theft, burglary, criminal damage and arson, other crime, other theft, possession of weapons, public order, robbery, shoplifting, theft from the person, vehicle crime, and violence and sexual offences. ↩

7. These scores provide a ranking of severity of offences based on the typical sentence length or fine. For example, homicide receives the highest score of 7,832, whereas possession of cannabis receives a score of 2 (ONS, 2023). The total Crime Severity Score is calculated as follows: “Sum [across all offences] (weight * number of offences) / (Area) mid-year population estimates” (ONS, 2023). ↩

8. Up 10% on the previous quarter, although this equated to a fall of 2% on the equivalent quarter of the year ending March 2019. ↩

9. The offence sub-category ‘stalking and harassment’ has been excluded from this graph due to a change in the counting rules (see Appendix B). ↩

10. Excluding GMP, as they did not provide outcomes data for part of 2019, and one non-metropolitan force, as their outcomes data in 2020 was incomplete because of delays submitting returns to the Home Office. ↩

11. This was due to the impact of backlog clearing by one metropolitan force that is thought to have driven up charges in quarter 4 (January to March) of 2020. Including March data would have resulted in concerns about the validity of any comparisons between March 2020 actuals and forecast charge estimates. This explains the different time periods used for the ARIMA (crime) and ARIMA (charge) models. ↩

12. Although some offence sub-categories were close to significance at the 95% level at 12 months. ↩

13. ARIMA analysis using 80% CIs for the 3-month period showed the same results as 95% CIs, but with possession of weapons offences and rape showing above forecast compared to the actual levels. ↩

14. ARIMA analysis using 80% CIs for the 12-month period showed 5 offence sub-categories as above forecast compared to actual levels – arson, criminal damage, public order offences, trafficking of drugs and vehicle interference. ↩

15. From April 2020 a change to the Home Office Counting Rules means all cases where a course of conduct is reported between a victim and their former partner must be recorded as stalking unless the police are satisfied that the matter amounts to harassment in law only. This was expected to cause an increase in offences recorded against ‘stalking’, therefore data from year ending March 2021 onwards is not comparable to previous years. Therefore, ARIMA was unable to be conducted on stalking and harassment due to a lack of consistent historical data to train models. ↩

16. For CPS-charged cases, it will still be the police who are responsible for building the case and securing evidence. ↩

17. The CPS noted that there will always be a number of cases that cannot be dealt with in less than 28 days because of the complexity of the case and the level of legal input required, and the CPS estimates about 500 cases will usually be in this category at any given time. ↩

18. In November 2014, 3 police forces (Leicestershire, Staffordshire and West Yorkshire) started a 12-month pilot of simplified outcomes. All 3 forces stopped issuing adult Simple Cautions (part of Outcome 3) and cannabis/khat warnings (Outcome 7) and limited their use of PNDs (Outcome 6). They used the adult Conditional Caution (part of Outcome 3) and the Community Resolution (Outcome 8) instead. These 3 pilot forces have continued to use the Simplified Outcomes since the pilot ended in November 2015. Leicestershire, Staffordshire and West Yorkshire will have a smaller proportion of Outcomes 6 and 7 and a greater proportion of outcome 8 than other forces. Since this pilot, some other forces have also limited their use of some OOCDs. This may lead to apparent falls in some of these outcome types (for example, cannabis/khat warnings) which may be accompanied by increases in Outcomes 3 and 8 which will have been used in their place. ↩

19. The data in this report excludes Diversionary Outcomes and GMP, one non-metropolitan force and British Transport Police (BTP). ↩

20. These scores provide a ranking of severity of offences; for example, homicide receives the highest score of 7,832, whereas possession of cannabis receives a score of 2 (ONS, 2023). ↩

21. Comparative reduction, all values here are normalised against actuals to provide symmetrical CIs, measured at -138% (that is, the gap between mean forecast and actual police recorded crime was 138% of the actual value). ↩