Written evidence from Maximus (WTW0016)

Executive Summary

1. In this submission, MAXIMUS present the considerations that effectively calibrate the funding model and leverage risk and reward to better support the hardest to help on The Work Programme and ensure they receive a high quality service. We outline the importance of:

• Accurate forecasting of referral volumes and sharing volume risk
• Performance benchmarks and accurate measurements
• The “average unit price” as opposed to the “maximum unit fee”
• Suitably balancing risks and returns to deliver a vibrant competitive market place of high quality bidders
• Mitigating the risk of the emergence of a low performance/low investment model

2. We conclude with three recommendations:

1)      The Department for Work and Pensions (DWP) should consider how it can share volume risk with providers. Variations to volume forecasting are detrimental to both performance and cost effectiveness.

2)      DWP should focus on the average price received rather than total unit price when calculating how much money is available for interventions.

3)      DWP should properly calibrate performance expectations. This will enable the DWP to provide the right incentives to drive investment and exceptional performance on Work Programme Plus.

Introduction to MAXIMUS and purpose of submission

3. MAXIMUS is one of the DWP’s top performing providers of the Work Programme, delivering the scheme in West London, Thames Valley, Hampshire and the Isle of Wight, and North Yorkshire and the Humber. MAXIMUS delivers the Work Programme directly and through a network of local supply chain partners. Our network comprises a mix of voluntary, public and private sector providers, ranging from smaller, specialist community organisations to national partners. We recognise that supply chain partners are an integral component of our successful delivery and of the Work Programme overall. The recommendations set forth in this submission are particularly important to maintain the role and status of delivery partners in Work Programme Plus.

4. MAXIMUS has considerable experience providing support to help disadvantaged people back into the labour market and has partnered with governments around the world to provide critical health and human service programmes to a diverse range of communities for more than 40 years. MAXIMUS was recently chosen by the DWP to partner with disability employment specialists Remploy to enhance employment opportunities for disabled people.

5. MAXIMUS operates in the UK, US, Canada, Australia, New Zealand and Saudi Arabia. Through its subsidiary Health Management Limited (HML), MAXIMUS is also one of the largest providers of occupational health services in the UK and the provider of the DWP’s Fit for Work Service. MAXIMUS Centre for Health and Disability Assessments also provides the Health Assessment and Advisory Service for the DWP.

6. The purpose of the submission is to demonstrate ways in which the commissioning of Work Programme Plus can be improved to better support the hardest to help unemployed people. This will drive higher performance and better value for money. This submission is our contribution to helping the Government achieve this goal.

Information to support MAXIMUS conclusions

Calibrating the funding model and leveraging risk and reward

7. DWP have stated that they want to achieve the following in the commissioning and delivery of Work Programme Plus:

• Maintain the performance improvements that have been made over the course of the Work Programme to establish a new baseline for performance on which to build further improvements.
• Improve the efficiency and cost effectiveness of delivery and to reduce the overall cost of Work Programme Plus.
• Ensure that significantly improved performance is achieved with the ‘hardest to help’ groups, in particular ESA job seekers and those with disabilities.  Improved performance with these groups will generate a greater level of benefit savings for the Government.

8. Taken together, this is a challenging set of demands. However, these can best be realised if DWP correctly calibrates the funding model for Work Programme Plus. This can be achieved through focusing on two elements through the commissioning process:

1. Transparent and accurate modelling to ensure a better assessment of costs, revenue, risk and reward including:

• Accurate forecasting of referral volumes and sharing volume risk
• Clear and realistic performance benchmarks informed by a transparent and accurate assessment of historical and current performance levels: desired performance improvements need to be within reach
• Understanding that it is the ‘average unit price’ that matters, not the ‘maximum unit fee’

2. Providing the right incentives for high performance focused on:

• Suitably balancing risks and returns to attract bidders
• Mitigating the risk of the emergence of a low performance/low investment model

Accurate forecasting of referral volumes and sharing volume risk

9. Accurate and reliable forecasting of referral volumes is an essential part of achieving both a high performing programme and a cost effective one. On the Work Programme we have seen a negative impact where referrals have been either significantly higher than forecast or significantly lower than forecast.

10. For example, over the first three months of the Work Programme, referral volumes were over 200% higher than forecasts and they continued at a much higher than forecast level for the whole of the first year. Providers had infrastructure and staffing capacity set up to deal with half the actual volume of referrals. This meant jobseekers referred over the early months of the programme had less interventions and a less intensive service, which in turn meant lower performance was achieved. It took between three to six months to stabilise the resources and to ensure adequate resources were in place to meet volume demand. For example, time was required to reach full staffing levels on account of recruitment, training and security clearing.  Volume fluctuations had a disproportionately large impact on smaller and more specialist organisations who are more sensitive to commercial changes.

11. Undoubtedly this was a large contributing factor to the lower than expected performance in Year One of the Work Programme. The subsequent criticism took a long time to neutralise.

12. At the opposite end of the spectrum, current referral volumes are one sixth of what they were in Year One of the contract and are less than 50% of the original forecast.  This leads to poor cost-efficiency as money is currently being spent on fixed costs that are not required, such as premises and IT equipment. This inflates the cost of delivery and takes away investment from front-line services that could further improve performance.

13. Recognising the importance of volume assumptions to both performance and cost effectiveness, DWP can support higher performance at a lower cost by sharing some of the inherent volume risk. This can be achieved by committing to shifting referral trigger points when volumes are too low (or high) and/or reviewing unit prices in scenarios where volumes are outside a set tolerance level.

Clear and realistic performance benchmarks informed by a transparent and accurate assessment of historical and current performance levels: desired performance improvements need to be within reach

14. When designing outcome-funded models it is very important to accurately project the range of performance than can be achieved with a particular group — this should range from a base level of performance through to a good standard and a peak level. It is vital that performance benchmarks are informed by current performance levels and that there is transparency around levels of historical performance that have been achieved.

15. It is also important to accurately benchmark performance levels for groups based on levels of difficulty.  There will be clear differentials in performance that can be achieved for different groups based on the challenges faced when assisting them into work and the base level of performance that would be achieved if no service were offered. 

16. For PbR models to work effectively, providers need to be confident of being able to achieve targeted performance levels. Providers also need assurances that they can afford to pay the costs of delivery while achieving a reasonable margin, with larger rewards available for peak levels of performance. If performance levels are set unrealistically, then there will not be enough money available to cover the costs of delivery. This will ultimately lead to the degradation of service and performance.

17. Insufficient performance data for providers to be able to assess whether the performance benchmarks are realistic can result in risk aversion and under investment. Providers will not have the confidence of receiving sufficient outcome payments to cover the cost of delivery.

Understanding that it is the ‘average unit price’ that matters, not the ‘maximum unit fee’

18. In an entirely PbR model, the primary financial metric used by providers to financially model and create their business plan is the average price per claimant paid in each client group. This is calculated by multiplying the unit prices against the expected referral volumes and performance levels. It is this which dictates how much money is available to spend on interventions and what quality of service is achievable - determining the caseload sizes for advisers, level of training and specialist interventions. If the balance of risk and reward is set incorrectly because targets for certain groups are not achievable, this can result in an unintended and dramatic reduction in the average price available per claimant. This can greatly reduce the quality of services the business model makes it possible to offer, regardless of how high the theoretical maximum price per claimant.

19. The illustrative example below demonstrates that it is the “average unit price” that matters, as opposed to the “maximum unit fee”:

Figure 1: Average Unit Price versus Maximum Unit Price Illustrative Example

Average Price not Unit Price

20. As the Payment Schedule below shows, there is a risk to the service for the harder to help job seekers in Group B by entering into a negative low investment, low performance model. This group is much harder to help than Group A and therefore the cost of assisting back to work should be higher and caseload sizes lower. Although the maximum unit price per person at £8,000 is double that available for Group A, this would on the surface suggest more intensive interventions can be afforded.

21. However in this case the peak level of performance into work is only 20% and when the model is run through we can see that despite a unit price that is twice as high, the lower rate of outcomes actually mean that the price per referral is £1,623 as the Payment Schedule shows. This is actually lower than the average price for Group A. This means that caseload sizes for Group B would actually have to be higher than for Group A, which in turn makes it highly unlikely that the modelled success rates could be achieved. This leads to a lower average price per referral which in turn would lead to less money being available to be spent on interventions.

22. The same amount spent on per referral Group B generates far fewer jobs and has a higher price per job.

23. To ensure that an outcome funded model works for both government, providers and, most importantly, jobseekers, it is important to build a model in which the unit prices taken together with the target success rates possible with a group, lead to a price per referral that supports the level of intervention required to achieve the target success rates. This can get more challenging with hard to help groups where unit prices have to become very high to support the level of intervention required to make the programme work.

24. If we apply this thinking to the Work Programme we can see that the model has worked well and has supported high performance for some groups but for others it has not worked as intended.  The table overleaf sets out each of the payment groups alongside the maximum unit price which can be earned and our forecast of a potential average price providers can expect to receive.

Figure 2: Maximum Unit Price compared to Average Price Per Person by Payment Group

25. The funding model has worked well for the JSA groups in the main where high performance has been achieved, and the average price per person supports the required level of service to drive high performance.  However, we can see that for some of the hardest to help groups (ESA and prison leavers) the average unit price is actually lower than the unit price for the JSA job seekers.  The consequence is that there is actually less money to invest in these harder to help groups and this will inevitably impact on the performance that can be achieved.

Suitably balancing risks and returns to attract bidders

26. Creating a platform that balances risk and reward is a sign of good market stewardship.  With the right balance in place, providers can be appropriately rewarded for their performance in a risky environment. This means that exceptional performance is rewarded through the opportunity to reach high profit margins, and poor performance leads to the risk of financial losses.

27. In this model, Work Programme Plus providers will be incentivised to invest in service delivery as they know it will lead to a greater financial reward. Average performance will only permit modest financial terms, so providers drive to deliver above average performance and beyond. Providers who perform below average risk losing money.

28. If the financial structure is set such that profit is capped and incremental increases in performance do not result in an increase in profit, investment will be lower and fewer long-term unemployed people will be supported into jobs. There is an expectation in government contracting that providers can expect a margin of between 6-10%. However, in light of the risk and potential for low return from low investment, Work Programme Plus has the option to push providers towards the best performed achieved on employment services through increasing this percentage bracket. This is pertinent in PbR models which load the majority of risk to providers. For example, if providers know there is an option to earn double digit profit for strong performance, they will invest in delivery to achieve exceptional performance. This will help more jobseekers into work and reduce spend on benefits. Quality will suffer if the margin is set too low, as providers will not risk investing in delivery without the potential for return.

Mitigating the risk of the emergence of a low performance/low investment model

29. As we have seen in the Work Programme Pricing Table above, the average price received on the Work Programme for an ESA (PG6) job seeker is £1,144, compared to £2,009 for an early access JSA job seeker (PG3).  While both are hard to help groups, ESA job seekers are relatively more difficult to help and require more time and resources to be supported into work.  The lower amount of money means there is no option to invest in more costly health-focused interventions and there is a disincentive to invest and take risks in trialling new approaches because of the uncertainty of receiving that money back in outcome payments.  This scenario can be mitigated against if the recommendations above are followed to ensure contracts are properly calibrated.  However, if there is no clear or reliable historical performance track record (as was the case at the beginning of the Work Programme with ESA job seekers) the risk of under-investment can be mitigated by increasing the proportion of guaranteed funding, and building in review points as the contract progresses to ensure that the unit prices are properly calibrated to achieve the desired average spend per job seeker – this could result in unit prices increasing or reducing over time.

Recommendations for action

30. Based on the information provide above, MAXIMUS propose the following recommendations:

31. Recommendation 1) DWP should explore means to share volume risk with providers. Variations to volume forecasting are detrimental to both performance and cost effectiveness.
• How? DWP can use contractual mechanisms that allow flexibility within pricing models that will allow providers to effectively manage change. 

32. Recommendation 2) DWP should focus on the average price received rather than total unit price when calculating how much money is available for interventions
1. How? DWP can calculate the average unit price by multiplying the unit prices against the expected referral volumes and performance levels.
2. The average unit price can show DWP how much money is available to spend on interventions and achieve a particular quality standard.

33. Recommendation 3) DWP should properly calibrate performance expectations alongside the expected average per person. This will enable the DWP to provide the right incentives to drive investment and exceptional performance on Work Programme Plus
1. How? DWP can create a pricing model which ensures funding follows performance.
2. Average performance will only permit modest financial terms or cost recovery. Providers are driven to deliver above average performance and beyond. Providers who perform below average risk losing money.

34. These recommendations will drive higher performance and better value for money in Work Programme Plus. They are also particularly important considerations to ensure the viability of services for smaller, more specialist organisations, who play a critical role in all employment services.

August 2015