Follow up written evidence from Institute for Government

Thank you again for inviting me to provide evidenc on court capacity on 12 January 2021.

In that evidence session, I said that I would provide two pieces of information in writing.

  1. In answer to Q11 – the cases processed by the magistrates court – in the four weeks up to 20 December, the magistrates processed 101,000 cases, 75% of their pre-pandemic caseload.[1] However, because the flow of cases into the system is slowed, this was sufficient to reduce the backlog by 8,500.


  1. In answer to Q23 – the additional resources required to increase court capacity – our report in April 2020 states that compensating for the additional demand as a result of 20,000 extra police officers will require an extra £125m per year by 2023/24 in our central scenario.[2] We estimate that every 10% increase in caseload in the criminal courts requires an extra £110m per year.


In addition to this specific information, below I have provided our latest modelling based on HMCTS data published in January and covering up to the end of December 2020.

  1. At the end of December, the backlog in the magistrates’ court was 15% higher than before the crisis, but 11% lower than at the end of July. The backlog continues to fall as cases processed outpaces cases received.


  1. In the Crown, the volume of cases processed continued to increase in December. At the end of December, the backlog was 54,115 cases, a very similar number to the end of November. This backlog level is 38% higher than before the pandemic. Both disposals and receipts are now at, or similar to, pre-crisis levels.


  1. However, the situation remains much worse than these headline numbers suggest because jury trials take much longer to hear and they have been more difficult to stage throughout the pandemic. This is still the case. In December, on average 230 jury trials were processed per week.[3] This is higher than previous months, but still much lower than the 350 jury trials processed per week before the pandemic.


  1. Adjusting the backlog to account for the greater complexity of jury trials shows a troubling picture. Our modelling implies that the current backlog, when adjusted to be the same complexity as the pre-pandemic backlog, is equivalent to 70,000 cases. This is a 78% increase on the pre-pandemic period and is far worse than any backlog since at least 2000.


  1. This analysis rests on assumptions about the composition of cases the Crown court is receiving (as this is not provided in the weekly data). Specifically, we assume that the share of cases received requiring jury trials pre-pandemic was the same as the share of cases processed that require jury trials. Further, we assume that the share of case receipts requiring jury trials has not changed over the pandemic even as the total number of case receipts has changed. This is a neutral assumption, but could be wrong. If, for example, a smaller share of cases received during the pandemic require jury trials, the problem would not be as bad as our analysis suggests. But the opposite is true if the share of cases requiring jury trials has increased.


  1. We calculate that eliminating the current covid-related element of the backlog would require an additional 32,000 sitting days on top of the sitting days required to process as many cases as the courts are receiving. 24 months running the courts at 2016 levels (110,000 sitting days) would likely be sufficient to return the backlog towards pre-pandemic levels.


  1. However, the (complexity-adjusted) backlog is still getting worse. Even in December, when the overall backlog did not grow, the backlog became more complex because jury trials are still not being processed at pre-pandemic levels. Every month with the courts operating at December 2020 levels would add an additional 1,500 sitting days required for recovery.


I would also like to explain why the Institute for Government model[4] (and the MoJ model[5]) have very different predictions to the Crest Advisory model[6] about how demand on criminal courts is likely to change over the next few years. It is our view that a number of the assumptions in the Crest model serve to over-estimate the increase.

  1. The Crest model overstates the impact of the pandemic on backlogs because it assumes the very low number of cases processed in April and May persisted until September 2020. In practice, the courts have recovered much of their capacity since the depths of lockdown 1.


  1. The Crest model also includes a set of assumptions which contribute to the ‘normal’ flow of cases into the courts doubling over the next four years. We discuss the merits of these assumptions, and the alternative assumption that is in our model, in the following paragraphs.


  1. Assumption 1: the number of cases charged by police officers increases in line with police recorded crime. In effect, this assumes that the same amount of police resource will be able to lead to more charged crimes when the number of crimes reported increases. However, in 2019 less than 10% of reported crimes were charged. In both the IfG and MoJ model, the alternative assumption is that the number of cases charged by police officers increases in line with the number of police officers. This is closer to what we have observed historically. For example, since 2014 recorded crime has increased but the total number of charges (and the total number of charges per officer) has fallen.


  1. Assumption 2: recorded crime will increase as quickly as it has since 2014 over the next few years. This leads to a big increase in recorded crime up to 2024 because recorded crime has increased substantially since 2014. However, the increase in police recorded crime is largely an artefact of the data and does not represent a ‘real’ trend.[7] Much of the increase represents improved reporting by police, and there is no increase in crime in the crime survey of England and Wales. The IfG and MoJ models do not need to make an assumption about how much crime there will be because number of charges does not depend on number of crimes in those models.


  1. Assumption 3: recorded crime will grow even faster due to unemployment. The Crest model assumes that some types of crime (violence, sexual offences, public order offences and theft offences) will increase more quickly due to economic effects from 2021 onwards. These increases are speculative at best, and in practice we know that government support means that unemployment will not reach the heights the Crest model anticipates. There is also a further assumption that there is a spike in violence against the person offences for three months in September-December 2020 “due to anticipated gang/drug distribution power struggles”. As far as I’m aware this increase has not arisen. The IfG and MoJ models do not need to try to model how reported crime will evolve, and so there is no equivalent assumption in those models.


  1. Assumption 4: more police officers will lead to an increase in the charge rate: the fraction of reported crimes that are charged. This assumption means that the fraction of reported crime cases that are charged recovers to levels seen in the early 2010s. However, it is unlikely that this fraction will recover to pre-2014 levels because (i) cases have become more complex for the police and (ii) the denominator of the charge rate calculation – reported crime – has grown due to improved reporting by the police (see paragraph 13). This assumption, combined with the ones above on recorded crime, is what drives the very big increase in inflow into the courts. IfG and MoJ models assume that charges per police officer stays constant over time. This means that the new police officers do lead to more cases being charged, but the increase does not depend on how police recorded crime changes.


  1. The combined effect of these assumptions is their prediction that criminal court capacity will need to double by 2024. In effect, this means that they expect the number of cases being received by the courts will double in five years from 2019. We view this increase as implausible. In our modelling, an extra 20,000 police officers would require an 11% increase in court capacity (16% in the Crown). Even in our high scenario, in which charges per officer increases substantially, HMCTS would need to increase court capacity by 22% (27% in the Crown). This is orders of magnitude below the Crest model’s prediction. As noted above, the MoJ’s own analysis is very similar to our own. 

[1] HMCTS, ‘Weekly management information during the coronavirus pandemic March-December 2020’, January 2021,


[3] This figure excludes the week ending 27 December, when volumes were lower due to Christmas.

[4] Pope T, Davies N and Guerin B, ‘The criminal justice system. How government reforms and coronavirus will affect police, courts and prisons’, Institute for Government, April 2020,

[5] Ministry of Justice, ‘Prison population projections 2020 to 2026’, November 2020,

[6] Crest Advisory, ‘Impact and legacy of Covid-19 on the CJS Modelling overview’, October 2020,

[7] Elkin M, ‘What’s happened to the volume of crime recorded by the police?’ in Crime in England and Wales: year ending March 2019, Office for National Statistics, 18 July 2019, retrieved 22 April 2020,