Safety and Cost Effectiveness of Private Prisons

Tabled: 29 March 2018

2 Private prison contract management and performance

The Act requires DJR to monitor the performance of all correctional services to achieve the safe custody and welfare of prisoners. DJR is accountable to the community for the operation and cost of Fulham and Port Phillip. Contracting private companies to operate prisons does not lessen the state's duty of care to prisoners.

The contracts for Fulham and Port Phillip transfer significant operating and financial risks to the private operators and require them to meet specified service standards and performance expectations. These contracts provide the state with broad monitoring and access rights, as well as penalty and intervention options if the operators fail to meet expectations.

This part of the report examines DJR's approach to performance monitoring and the actual performance results for Fulham and Port Phillip. The performance of the private operators needs to be considered in the broader context of the challenges facing the entire corrections system, outlined in Parts 1 and 3.

2.1 Conclusion

GEO and G4S largely met the state's service delivery expectations for correctional and accommodation services over the life of the original contracts for Fulham and Port Phillip.

CV rigorously monitors the performance and contractual compliance of the private operators. It works collaboratively with the private operators to improve their performance and applies contractual penalties in response to serious performance concerns.

CV collects a large amount of data on prison performance, however outdated information systems limit its ability to produce meaningful analysis. CV's limited public reporting on prison performance restricts the transparency and public scrutiny of the prison system.

2.2 CV's scrutiny of operator performance

CV has multiple ways of gaining insight into and scrutinising the operational practices and performance of the private operators—see Figure 1J in Part 1. Reliable and accurate performance data and effective contract management is crucial for CV to meet its performance monitoring obligations and to hold private prison operators to account.

Contract management branch

CV's contract management branch (CMB) is responsible for managing the contract for each private prison. CMB's key functions include:

  • assessing the operators' service delivery against requirements
  • managing the state's oversight of the prisons' operations manuals and ensuring effective review of operating instructions
  • managing the process for authorising private prison staff under the Act
  • managing and monitoring prisons' reporting requirements and preparing reports for the Commissioner on prison performance
  • validating and verifying performance data reported by the prisons
  • managing contract variations, including analysis of operators' proposals to ensure a value-for-money outcome
  • managing payments to private operators under the contracts, including payments linked to performance.

CV is efficiently administering its contract management responsibilities. Since our 2010 audit Management of Prison Accommodation Using Public Private Partnerships, CV has increased resources in CMB including a dedicated contract manager for each private prison. It also implemented a new information management system for contract administration. CMB works collaboratively with the private operators and has a strong presence at both private prisons.

Operating instructions in private prisons

Operating instructions guide day-to-day operations in the private prisons. Prison safety and security rely on their effective implementation.

The Commissioner reviews and endorses operating instructions submitted by the private operators. In 2012, CMB began supporting the Commissioner's reviews by checking operating instructions to ensure their consistency with contractual, legal and other requirements and to gain assurance about private prisons' operations. Prior to this, CV conducted an annual high-level review of the prisons' operating manuals.

We examined endorsement processes for 114 operating instructions and found that the review process is rigorous. There were some minor record-keeping issues identified in 12 of the processes.

CMB undertakes regular audits to test the implementation of operating instructions. The audit schedule and methodology is risk based and, importantly, these are not desktop audits—they involve CMB staff talking to staff and prisoners, observing day-to-day prison operations, and checking source documents and registers. Prison management must immediately rectify any critical noncompliance. There is no equivalent audit program in public prisons.

Contractual responses to poor performance

The initial contracts enabled the state to address poor operator performance by:

  • reducing the fees operators received for accommodation and/or correctional services if any services were not satisfactory
  • reducing the annual performance-linked fee, a payment based on the state's assessment of the operator's performance against the SDO performance thresholds
  • issuing default notices
  • ending the contract and appointing a new operator if an operator was no longer capable of operating the prison satisfactorily, or if it became insolvent.

Imposing contractual sanctions is not necessarily an appropriate first response to poor performance by the operators. This is particularly so when system-wide challenges and risks, outside the control of individual operators, contribute to performance failures. CV's approach of applying graduated contract responses provides notice and time for the operators to improve their performance. For example, in early 2017, CV responded to a number of serious assaults at Port Phillip with a peer review to assist the operator to improve safety and security. Section 2.5 includes more information on the peer review process and results.

Figure 2A shows the default notices for correctional services issued to the operators of Fulham and Port Phillip, between 1997 and 2017. While public prisons are not subject to contractual defaults or similar processes, it is important to note that similar serious incidents also occur in public prisons.

Figure 2A
Contractual defaults at Port Phillip and Fulham

Period

Incident

Port Phillip

1997–2006

  • Six defaults over this period including for:
    • death in custody
    • serious staff misconduct
    • systemic noncompliance with operating procedures

April 2016

  • Breach of firearms licensing process

July 2016

  • Death of a prisoner after ingestion of drugs

March 2017

  • Prisoner escape from St Vincent's Hospital

Fulham

April 2016

  • Prisoner escape

August 2016

  • Infant cannabis seedlings found on prison grounds

Source: VAGO based on information from CV.

Default notices trigger a requirement for operators to develop 'cure' or improvement plans within specified time periods. CV reviews the adequacy of these plans and is rigorous in tracking operator actions to implement the agreed improvement plans through onsite validation and monitoring at the quarterly performance meetings.

Determining performance against SDOs

The private prison operators self-report performance against SDOs to CV, and CV then determines the performance outcomes. The operators' initial reports can change over time based on the outcome of other processes and investigations, including:

  • data adjustments initiated by operators following their own review processes, investigation of particular incidents and finalisation of pending drug test results
  • validation reviews of operator performance data by DJR
  • coronial investigations into deaths in prisons which can extend over years
  • determinations issued by the Commissioner based on detailed examination of particular incidents.

This means that there is often a delay between the reporting of an incident and finalisation of the performance outcome for a particular SDO. As an example, the SDO relating to assaults on prisoners by prison staff requires the prison operator to refer the incident to Victoria Police and investigate the alleged assault to either prove or dismiss the allegation. Port Phillip reported no assaults by staff on prisoners in 2010–11. CV later assessed Port Phillip as failing this SDO following an OCSR review of an incident.

The new contracts address the lengthy delays in determining contractual outcomes for prisoner deaths by allowing the Commissioner to issue a financial penalty, known as a 'charge event', when the Commissioner has sufficient information to establish that the operator's failure to meet contractual requirements contributed to the death. The Commissioner does not need to wait for a coroner's finding or other investigations to issue a charge event for a prisoner death.

The Commissioner and Deputy Commissioner also receive requests from both public and private prisons for special consideration when they assess final performance against particular SDOs. This is often the case for SDO 3 (out-of-cell hours). Consistent with the SDO guidelines, prison operators regularly request retrospective approval from the Deputy Commissioner for instances where they have restricted prisoners to their cells or units, known as a 'lockdown'. The Deputy Commissioner must decide whether the request is reasonable—for example, after a serious incident to maintain safety—and, if the request is approved, CV excludes the lockdown from SDO calculations.

CV gave both operators time to phase in the use of a new method for measuring performance against SDO 23 (case management) in 2016–17. This is because CV had imposed a new measurement tool for use across the system without properly considering the contractual implications for the private operators—see Appendix C. This situation demonstrates how important it is for CV to consider the contractual implications of any planned substantive changes in system-wide standards, initiatives or performance measures.

CV's oversight of occupational health and safety performance

Victoria's Occupational Health and Safety Act 2004 and associated regulations provide the framework for ensuring a safe workplace. In the context of prisons, the obligation for ensuring the health and safety of workers, prisoners and others resides with the employer, in this case G4S and GEO. However, the contracting of correctional services to private operators does not completely remove the state's obligations under the Occupational Health and Safety Act 2004.

Although CV has an SDO measuring occupational health and safety (OHS) performance, it does not apply it to private prisons. We found that CV had minimal oversight of the private operator's OHS performance. In addition, CV did not collect key information on how the private operators manage their OHS obligations, for example, the number of notices issued by WorkSafe Victoria under the Occupational Health and Safety Act 2004.

We raised these issues with CV during the audit and it has now increased the OHS reporting requirements for private prisons. This is a positive step to improve CV's oversight and better understand whether private operators are meeting their OHS obligations.

2.3 Prison performance data

Accurate and reliable performance data is important for transparent reporting on prison operations and costs. We reviewed CV's processes to obtain assurance on the accuracy and completeness of this data, with a focus on safety and security performance data.

Validation of incident data

The Prisoner Information Management System (PIMS) is CV's system for recording incidents in Victorian prisons. In 2016–17, there were 24 335 incidents recorded in PIMS for the men's prison system, a rate of 3.8 incidents per prisoner. This has increased from 10 400 incidents recorded in 2010–11, a rate of 2.4 per prisoner. Of the 2016–17 incidents, 5 per cent were notifiable, that is, the most serious incidents, described further in Part 3.

Every incident entered into PIMS, across both public and private prisons, is classified according to the nature and severity of the incident. CV staff manually check the details of each incident, which provides assurance that incidents have been accurately classified and recorded. It is important for the private operators to classify incidents correctly because many of the incidents recorded in PIMS are relevant for assessing the operator's performance which, in turn, has financial implications for the state and the operators.

Between December 2016 and May 2017, Port Phillip had 272 incidents which required reclassification (9.1 per cent of its incidents), and Fulham had 133 incidents (8.1 per cent of its incidents). System-wide in men's prisons, 8.7 per cent of incidents required reclassification. Reclassification of PIMS incidents does not always mean there is an error. Rather, the classification may change if further information becomes available, for example, a prisoner requiring hospital admission after an incident.

SDO data validation

CV regularly validates the SDO performance data self-reported by the private prison operators to ensure the accuracy of the results.

To do this, CV visits each prison and checks a sample of reported SDO data against source records. These records include the daily logbooks held in every prison accommodation unit, training and work attendance registers, and lab reports supporting urinalysis results.

CV is introducing a new approach to its validation of SDO data for public and private prisons in response to an external review that it commissioned in 2016. The review found that:

  • there was too much focus on data verification, rather than data and trend analysis
  • the length of time taken to collect, report and verify data prevented real‑time responses to issues
  • prisons should assume more responsibility for data accuracy
  • there was inconsistency in validation efforts across public and private prisons
  • a streamlined data collection method was required because of the different data systems in use
  • there was a need for a new risk-based validation method.

A CV reform project intends to establish a new risk-based methodology for validating self-reported data, including the frequency of validation required. The new contracts also include KPIs related to data accuracy, which shifts responsibility to the private operators.

Initial contracts

Under the initial contracts, the former OCSR validated SDO data annually. CV took over responsibility for data validation in 2015–16 and now performs monthly validation checks.

CV's validation of Port Phillip's SDO data for 2015–16 resulted in changes to reported performance data for five SDOs. The validation of Fulham's SDO data for 2015–16 resulted in changes to reported performance data for three SDOs. The changes did not affect the overall pass or fail rate of either prison.

CV's SDO calculator

CV's SDO calculator is a complex macro-driven Excel spreadsheet developed as the primary analytical tool to generate SDO and KPI performance reporting across all prisons. The spreadsheet performs both data storage and calculation. We reviewed the 2015–16 SDO calculator and found that it accurately computed the performance outcomes based on the input data.

However, there are numerous procedures required to manually input, verify and modify the source data. We are also concerned that the SDO calculator is highly customised, with complex design and bespoke back-end scripting. While there are basic usage guides in place and staff who are familiar with its operation, the primary capability and knowledge of the design and coding behind the calculator is limited to a single CV employee.

This dependence on one staff member potentially creates a single point of failure for the entire SDO performance reporting process. In addition to this, there are risks of untracked data changes to the data and issues with the functionality of the data.

Data analysis

Easy access to and effective use of relevant and reliable data is critical for CV to make informed decisions about current risks, emerging trends, operational requirements and strategic planning decisions in the corrections systems. A 2016 review of CV's SDO validation and review process found that CV has limited time to undertake meaningful analysis for the quarterly performance meetings.

CV regularly collects a significant volume of data on the operation and performance of both public and private prisons. However, the various information systems do not easily facilitate data analysis and the multiple systems do not readily interface, making it difficult to integrate data and produce meaningful analyses.

E*Justice holds a range of prisoner information and overlaps with PIMS. It records prisoner risk ratings, daily counts of prisoners (musters) and prisoner movements among other important information. Victoria Police uses E*Justice data via a different interface.

Consistent with our previous audits Managing Community Corrections Orders (2017) and Administration of Parole(2016),we found that CV's data is managed across several divisions and is contained in disparate legacy systems. Data is stored in CV's two main systems—PIMS and E*Justice—as well as over 200 locally managed 'shadow systems' such as Excel spreadsheets. Vendor technical support has expired for both PIMS and E*Justice.

There are clear opportunities to improve CV's data analytics capability, to better understand the performance challenges facing the prison system. For example, using data analysis to identify trends and potential correlations between prisoner profiles, the proportion of prisoners with psychiatric conditions, and incident trends would help CV to develop strategies for managing risks, particularly the higher instances of violence across the system. Another example is CV's current inability to analyse and use data to produce evidence on the impact of 'double-bunking' on prison incidents and risks. Double-bunking is when a cell designed for single occupancy is changed to accommodate two prisoners on a set of bunks.

We have seen isolated evidence of recent proactive data analysis by CV. For example, in August 2017, CMB analysed data at Port Phillip and found a potential correlation between 'shivs' (improvised knives) and the number of positive urinalysis results and tobacco-related incidents in the same location. As a result, CMB issued a contract administration note requiring Port Phillip to develop and implement mitigation strategies. Port Phillip provided an action plan to monitor and mitigate these issues, and CMB is monitoring its ongoing implementation.

CV recognises the weaknesses of using outdated legacy systems. However, it has been unsuccessful in securing funding for its preferred replacement—an integrated offender management system. In lieu of an integrated system, CV has developed an Excel-based incident-analysis tool that integrates some of the data between PIMS and E*Justice. The private prison operators find this tool valuable as it allows them to interrogate incident data in detail.

2.4 Prison performance reporting

Performance analysis and discussions

CV meets quarterly with the private prison operators to consider a wide range of performance information and data. These meetings are open and collaborative. They identify key issues and trends, and allow CV to hold the operators to account for their performance. CV also provides the Commissioner with a monthly report highlighting key performance data and trends.

In 2014, CMB implemented the Private Prison Reporting Framework (PPRF) to capture a range of data not directly related to SDOs or KPIs. The PPRF provides CV with greater oversight of prison operations and performance, including the following information:

  • workforce data—sick leave, staff turnover, staff disciplinary actions
  • prisoner movement data
  • prison searches for contraband and drugs
  • prison visits
  • prisoner disciplinary processes.

CV also contributes information to the PPRF, including the results of operating instruction and other compliance audits. The PPRF allows CV and the prisons to consider a broader range of information when monitoring and analysing trends.

Public reporting on private prisons

Transparent public reporting on the performance of the prison system is important for making DJR and private prison operators accountable and for building public trust in the system.

The limited information reported in DJR's annual report and Budget Papers provides little real insight into the operation of the prison system and no detailed information on the relative performance of individual prisons, including private prisons. The exception to this is the detailed performance information that DJR publishes on urinalysis drug results in individual prisons.

There is limited Victorian public reporting on private prison contracts, operations and performance. Other jurisdictions provide more detailed information, for example, the Department of Corrective Services in Western Australia publishes an annual report on the performance of each private prison, including detailed information about its service and financial performance.

Aside from ad hoc reviews or investigations by other integrity bodies such as the Victorian Ombudsman, there is very limited information in the public domain about the performance of private prison operators. JARO reports are not publicly available because they provide internal assurance rather than external oversight. Western Australia and New South Wales have independent inspectorates that monitor prisons and report publicly on this work. There is limited public transparency about the performance of Victoria's prisons.

In 2014, DJR and DTF advised government that they would work together to identify more suitable Budget Paper 3 measures to improve the transparency of the corrections system. To date, there have been no significant changes to those measures. CV advises that it is currently working on options to improve performance measures but that their implementation is dependent on a new integrated offender management system.

Australia's ratification of the Optional Protocol to the Convention against Torture and other Cruel, Inhuman or Degrading Treatment or Punishment (OPCAT) in December 2017 will increase the scrutiny of all places of detention in Victoria, including prisons. OPCAT is an international human rights treaty that aims to prevent abuse of people in detention by opening these places up to independent inspections by United Nations experts and independent local inspection bodies. Victoria is yet to finalise its implementation of OPCAT, but the November 2017 Victorian Ombudsman report Implementing OPCAT in Victoria: Report and inspection of the Dame Phyllis Frost Centre demonstrates the increased public scrutiny that all prisons should expect.

2.5 Performance results

GEO and G4S largely met the state's service delivery expectations for correctional and accommodation services over the life of the original contracts for Fulham and Port Phillip.

Accommodation services

The private operators met the state's requirements for accommodation services over the life of the initial contracts. However, these contracts included no SDOs directly related to facilities maintenance and specified only high-level obligations for repair and maintenance. The contracts required the operators to keep the facilities 'in good and substantial repair and condition' but did not define what this meant.

DJR advised government during negotiations for the new contracts that the assets and facilities at Port Phillip and Fulham were structurally sound and fit for purpose. DJR provided this advice based on its detailed asset condition surveys that also identified many defects in the facilities. The defects were rectified prior to commencement of the new contracts at the operators' expense, or under the new contracts as part of the asset life cycle management regime funded by the state.

Port Phillip's SDO results and performance payments

G4S has consistently underperformed at Port Phillip against the state's expectations between 2010–11 and 2016–17 in three key measures:

  • SDO 2—prisoner assaults on staff or other persons
  • SDO 6—prisoner-on-prisoner assaults
  • SDO 8—positive urinalysis results.

See Part 3 for more information on these performance results.

Figure 2B shows G4S's performance in meeting SDO thresholds since 2010–11 and the performance payment outcomes for each year.

Ten of Port Phillip's SDOs in the old contract require 100 per cent compliance and failure can result from a single instance of noncompliance, regardless of the overall performance rate.

Figure 2B
Port Phillip SDO performance

Year

SDO thresholds achieved

Failed SDOs

Performance payment outcome

2010–11

17 of 19
89.5 per cent

  • Assaults on staff or other persons
  • Assaults on prisoners by staff

Reduced by 3 per cent

2011–12

15 of 19
78.9 per cent

  • Assaults on staff or other persons
  • Assaults on prisoners by other prisoners
  • Assaults on prisoners by staff
  • Timely prisoner disciplinary hearings

Reduced by 13 per cent

2012–13

18 of 21
85.7 per cent

  • Assaults on staff or other persons
  • Prisoner participation in pre‐release programs
  • Provision of chronic health care plans

Full performance payment

2013–14

16 of 22
72.7 per cent

  • Assaults on staff or other persons
  • Assaults on prisoners by other prisoners
  • Assaults on prisoners by staff
  • Random general urinalysis
  • Prisoner participation in pre‐release programs
  • Timely prisoner disciplinary hearings

Reduced by 22.5 per cent

2014–15

19 of 22
86.4 per cent

  • Assaults on prisoners by other prisoners
  • Random general urinalysis
  • Provision of chronic health care plans

Reduced by 3 per cent

2015–16

18 of 22
81.8 per cent

  • Assaults on prisoners by other prisoners
  • Random general urinalysis
  • Prisoner medical screening within 24 hours
  • Timely completion of prisoner health care plans

Reduced by 17 per cent

2016–17

17 of 21
80.9 per cent

  • Escapes
  • Assaults on prisoners by other prisoners
  • Random general urinalysis
  • 'At risk' assessments within two hours

Reduced by 20 per cent

Source: VAGO based on information from CV.

G4S has only received the full performance payment once in the last seven years. There are a number of examples where CV has worked cooperatively with G4S to address its performance.

CV regularly meets with G4S to discuss performance, both formally and informally. Along with these structured and ad hoc conversations, there have been two occasions since mid-2014 when the Commissioner has formally 'called in' Port Phillip management to discuss particular concerns about its operational performance.

G4S responded to CV's concerns, whether raised formally or informally, by developing strategies and plans. CV rigorously monitored the implementation of agreed actions.

CV's increased oversight and scrutiny of private prisons is positive, however it is not possible to determine whether it has resulted in improved performance, as CV and the prison operators do not routinely evaluate the outcomes and impacts of their improvement strategies. Section 3.5 discusses an example of this relating to violence-reduction strategies.

In May 2017, after an escalation in serious incidents, G4S agreed to a peer review process at Port Phillip initiated by CV, designed to review the current operations and identify areas for improvement. This collaborative process sought to identify issues and share lessons learnt from the broader prison system. Through the process, CV's subject-matter experts and management from other prisons reviewed a range of different areas including:

  • the spread and effectiveness of middle management
  • barrier and perimeter security
  • prisoner cohort management, including placement and vacancy management
  • management of external escorts
  • culture, including staff and prisoner interactions
  • training, including for new recruits, refresher training, de-escalation, use of force and incident management
  • compliance management
  • implementation of the action plan to improve Port Phillip's PIU.

The reviews completed to date identified some areas of positive performance, including the management of visits, external escorts and the relationship between the prison and CV's sentence management function. They also identified about 40 areas for improvement, which included:

  • communication between management and general duties staff
  • development of supervisors, availability of supervisors in the unit and visibility of senior management
  • balancing compliance and prisoner management activities
  • use of resources in the PIU, particularly dog handlers and staff escorting prisoners outside of the prison
  • development and succession planning for staff.

It is positive to see Port Phillip and CV working together to identify issues and share the lessons learnt across the system.

Fulham's SDO results and performance payments

Fulham has generally performed well in meeting the state's expectations over the life of the initial contract, apart from some issues in 2012–13. There were also issues in 2016–17, the first year of the new contract. The prison has accepted remand prisoners since September 2015, and in December 2017 they made up around 25 per cent of all prisoners at Fulham.

Figure 2C shows GEO's performance against the SDO thresholds since 2010–11 at Fulham and the performance payment outcomes for each year.

Figure 2C
Fulham SDO performance

Year

SDO thresholds achieved

Failed SDOs and KPIs

Performance payment outcome

2010–11

16 of 17
94.1 per cent

  • Unnatural death

Reduced by 15 per cent

2011–12

16 of 17
94.1 per cent

  • Assaults on staff or other persons

Reduced by 10 per cent

2012–13

13 of 17
76.5 per cent

  • Assaults on staff or other persons
  • Assaults on prisoners by other prisoners
  • Completion of education and training programs
  • Prisoner participation in pre‐release programs

Reduced by 15 per cent

2013–14

All 18 SDOs
100 per cent

Nil

Full performance payment

2014–15

All 18 SDOs
100 per cent

Nil

Full performance payment

2015–16

16 of 18
88.9 per cent

  • Prisoner escapes
  • Timely completion of prisoner health care plans

Reduced by 15 per cent

2016–17 (new contract which includes KPIs)

17 of 18
SDOs 94.4 per cent

  • Assaults on prisoners by other prisoners

Reduced by 3.7 per cent

13 of 16
KPIs 81.3 per cent

  • KPI—Provision of timely primary health services
  • KPI—Performance data
  • KPI—Providing plans and reports
 

Source: VAGO based on information from CV.

GEO received the full performance payment twice in the last seven years. It made a submission to CV for reconsideration of the 2012–13 outcome, based on the fact that one incident in June 2013 resulted in 13 assaults on staff. GEO asserted that this one incident disproportionately affected otherwise good performance, but CV did not alter the performance outcome. GEO did not receive the full payment for 2016–17, the first year of operation under the new contract.

Back to Top