Communities and Local Government CommitteeSupplementary written evidence from Greater Manchester

Thank you for the opportunity to provide evidence on Monday, 3 June about community budgets.

As requested, I am enclosing information about the randomised control trial we are carrying out in Manchester. The evidence included is based on an incomplete sample, derived mid way through the trial. We are continuing to monitor the families and will report again in July this year and again at the end of the year, tracking the families’ progress over time.

The trial has been designed to test the assumption that an integrated approach to planning and delivering public services—our new delivery model, is more effective both in terms of outcomes and financially than “business as usual” in which services are provided in an uncoordinated way. We are following 240 families that have randomly been selected to receive the new model and 240 that do not. They are the control group.

The findings to date relate to 105 families that received the new model for at least six months and 90 families that were in the control group for a similar period of time.

The following diagram summarises the findings:

We have seen better results on debt management, safeguarding, anti social behaviour and crime in the new delivery model.

We have seen no difference on absence and exclusions, mental health and worklessness.

Looking more closely at the outcomes on the right we discovered that on absence and exclusions from school both groups performed well because we have an effective service in Manchester. For this outcome it is likely that the new integrated model is not required.

However, in both mental health and worklessness we observed poor performance in both groups. This has led to a review of mental health commissioning for troubled families and discussions with DWP about a post Work Programme intervention for this group, which is likely to include skills development.

The evidence collected from the trial has been fed into the cost benefit analysis and financial model.

This shows an estimated saving of £147k per year for the cohort of 105 new delivery model families.

This is the incremental benefit expected, meaning that the “business as usual” results have been used as the “deadweight” to represent what would have happened regardless of the intervention.

The largest savings are the reduction in Income Support at £45k and reduced fixed term exclusions at £37k.

It should be noted that no negative benefits have been included—ie if business as usual outperformed the new delivery model or the results worsened following the intervention, the result is assumed to be 0. The next iteration of analysis will address these shortfalls.

If we scale this up to look at the savings these 105 families may achieve over the five years following the intervention, we can estimate that there will be a positive net financial benefit of £607k

The model assumes that some families will slip back, hence the projected 5 year benefit is smaller than £147k times five years.

This does not take into consideration the cashability of the potential savings. Some of the results would not provide the required level of scalability to support withdrawal of resources. This is the reason our strategic focus is on Greater Manchester, where we can address dependency at scale.

Average benefits per head have also been calculated and show that the Family Intervention Project (FIP) is the most effective intervention. However, it is also the most expensive and this would be reflected in the benefit cost ratio.

The actual level of deterioration in effectiveness is not known so this analysis remains a high level estimate and will be refined as new evidence becomes available.

Given the total cost of the cohort over five years is £399k, the benefit cost ration is 1.52 meaning that for every £1 invested we can expect a return on the investment of £1.52.

The cost benefit analysis methodology we use was developed jointly with economists across Whitehall and is designed to be cautious. It includes discounts when we are not confident about the evidence and addresses optimism bias to mitigate against the risk of overstating benefits.

I trust this addresses the request for more information. Please let me know if additional information is required.

June 2013

—— END ——