Evaluation of the Reforms to Summary Criminal Legal Assistance and Disclosure

Appendix B - Analysis of the economic impact of the reforms

We have assessed the wider economic impact of the reforms using decision modelling. The main benefit of using this approach is that it enables complex processes and activities to be broken down into component parts, each of which can then be further broken down and analysed in detail before being recombined in a logical, quantitative and transparent way to estimate - in this case - the economic consequences of different courses of action and different experiences. The ultimate purpose is to identify which of the options provides better value for money. The structure of the model allows pathways through the system and possible outcomes (fines, custodial sentence, acquittal, further court attendances, etc.) to be mapped out. The probabilities of following the paths of the model are estimated from routine data where available and the costs associated with them from national figures. There is of course a level of uncertainty around these probabilities and costs. To address this uncertainty we have conducted sensitivity analyses where individual parameters are altered and the impact on expected costs observed. This approach has previously been used to evaluate electronic monitoring for people on bail, fines enforcement and also numerous healthcare interventions.

The model is shown in Figure B.1. Individuals enter the model at the left and either attend a PD or have the case abandoned before this stage. However, as will be seen from the table on probability values we assume that all the cases we are interested in have a PD. At the PD there is a probability of a guilty plea, a not guilty plea leading to an ID or the case being deserted. Continuations and adjournments, whilst not displayed in the model, have been adjusted for in the calculations as will be discussed below. If a plea of guilty is made then the court either imposes a fine, a prison sentence, a community sentence or there is another disposal such as admonishment or absolute discharge. Similar events can occur following an ID, TD and sentencing diet.

Data sources

In order to compare criminal justice system costs pre- and post-reform we have run the model using Sheriff court activity data from two time periods. For the pre-reform period we have, where possible, used data relating to the period from April 2006 to September 2007, whilst for the post-reform period we have used data for the period July 2008 to March 2011.^[26] However, for some activity data (specifically on sentencing) we have had to use data for the first three-quarters of 2008 for the pre-reform period. Data on probabilities have largely been obtained from the Key Performance Indicators on the Criminal Justice Board Management Information (CJB MI) system and are shown in Table B.1.

Costs of court appearances have been obtained from data used in the Audit Scotland report "An Overview of Scotland's Criminal Justice System" published in 2011. Costs of sentencing options have been obtained from "Costs and Equalities and the Scottish Criminal Justice System 2005/06" produced by the Scottish Government. The model takes into account the occurrence of adjournments by weighting the court costs. For example, in the pre-reform model the cost of a pleading diet is multiplied by 1.17 to reflect the 17% of cases where an adjournment takes place. It has been assumed (based on trial diet data from Audit Scotland) that an adjourned diet will incur one-fifth of the costs of a single diet. Costs for prison stays after an evidence-led trial assume a duration of six months. Adjustments have been made to allow for the potential of sentencing discounts for early pleas. The unit costs contained in this report have largely been taken from the recent Audit Scotland report. Others have been obtained from previous work and inflated to 2009/10 costs where necessary using inflators published by the Personal Social Services Research Unit. We have decided to use unit costs relating to one year rather than using unit costs relating to the pre- and post-reform periods because the key interest is to see how changes in activity levels affect costs rather than investigating how unit costs change over time. The costs used in the model are shown in Table B.2. With the exception of defence costs these were changed by 50% upwards/downwards in sensitivity analyses.

Basecase results

From Table B.1 it can be seen that the main differences between the pre- and post-reform models are that in the latter a higher proportion of individuals are assumed to plead guilty at pleading diet, fewer proceed to an intermediate diet, and sentences are more likely to be fines.

The model run using pre-reform data results in an expected cost per person going through the system of £3047. Using data for the post-reform period the expected cost per person is £2642. This suggests savings (for the events modelled) of £405 per case (13%) on the pre-reform situation. It is likely that sentencing outcomes differ by type of diet but data were not available on this. This suggests that the cost savings may be underestimated. In 2009/10 83,276 went through the courts. Multiplying this number by £393 suggests cost savings of £33.7 million.

The above figures include the costs of sentencing and defence agents. If the sentencing costs are excluded then the costs per case pre-reform are £1364 and post-reform are £1366. If defence costs are also excluded then the costs per case pre-reform are £751 and post-reform are £753. The reduced likelihood of proceeding past the pleading diet post-reform is offset by an increased proportion of intermediate diet cases going on to a trial and a greater likelihood of adjournments post-reform compared to the pre-reform situation. These figures suggest an increase in court costs of £0.17 million.

Sensitivity analyses

In the post-reform model the costs were sensitive to changes in the probability of a guilty plea at pleading diet. If this was 0.1 then the expected costs would be £3035, for a probability 0.2 (i.e. the same as in the pre-reform model) the expected cost is £2864. If the probability were higher than assumed in the model then the cost would be lower than in the base case model, e.g. £2523 for a probability of 0.4 and £2352 for a probability of 0.5.

Figure B.2 shows that the cost saving of £405 was relatively robust to changes in the costs of legal system events upwards/downwards by 50%. Greatest uncertainty was around and prison costs (C2). Changes in this could result from different assumptions about plea bargaining.

Legal aid expenditure

In 2007/8 the total expenditure on summary criminal legal assistance - that is on A&A, ABWOR and SCLA - for cases going through the Sheriff Courts was £58.2 million^[27]. In 2009/10 the equivalent figure was £47.2 million, a decrease of £11 million.

Main limitations to the analysis

To summarise, the main limitations of the economic analysis are as follows:

• For some indicators, the pre-reform activity data extends to April 2006, for others it is only available from early 2008 when, arguably, some effects of SJR were already being seen.
• Court and prosecution costs are estimated from data published in 2005/06 and adjusted to take account of inflation.
• Allowance has been made for only one adjournment of any single diet stage. Qualitative evidence suggests that some cases may have multiple adjournments/appearances at each stage. However, data on number of appearances at each diet is not available.
• The likelihood of receiving a particular disposal - fine, community sentence, custody - is assumed to be the same at each potential stage of conclusion.
• Likewise, costs for prison stays after an evidence-led trial assume a duration of six months. Adjustments have been made to allow for the potential of sentencing discounts for early pleas. Data are not available on average length of custodial sentence by type of diet.
• The trajectory data and corresponding court and prosecution costs have been calculated to allow a precise comparison of pre- (up to June 2008) and post-reform periods (from June 2008 to March 2011). However, data on legal aid expenditure is only available by financial year as such the pre- and post-reform periods for each set of data do no match exactly.

The sensitivity analysis described above was undertaken to explore the impact of a significant change to any of the assumptions made which are related to the key limitations listed here. This analysis indicated that the key estimated costs were relatively robust to any changes.

Figure B.1 Decision model of summary criminal justice system.

Figure B.2 Sensitivity analyses around cost estimates.

Table B.1 Probabilities used in model.

Note: Default probabilities are 1 minus the other probabilities at that point. Unless otherwise stated, pre-reform figures refer to the average rate over the period April 2006 to September 2007. Post-reform figures refer to the average rate over the period July 2008 to March 2011.

Table B.2 Unit costs used in model.