Calculating a positive future: using big data to manage demand and make savings

11 Sep 15
Using data to predict risks can sound like science fiction, but it has the potential to help the public sector manage demand more effectively and improve targeting of scarce resources.

In Hackney, our continuing efforts to manage limited resources and think creatively have taken us into the world of ‘big data’ and predictive analytics. It sounds like science fiction – the idea that you avoid something that is highly likely to happen, and highly likely to be disastrous if it does, by acting on a prediction – but this type of risk model has the potential to help the public sector manage demand more effectively and target scarce resources. However, a major challenge for finance professionals is how to quantify the savings.

Predictive risk modelling is widely used in the private sector, particularly by financial and retail organisations, and is gaining traction in the public sector. HM Revenue & Customs is modelling pension demand, local government is looking at benefits fraud prevention models and the NHS is using modelling to predict the likelihood of A&E admissions. In Hackney, we have secured funding via the Capital Ambition London Ventures Investment Board for Xantura Ltd to work with us to pilot a risk profiling model in children’s services.

The premise is that academic research and the government’s Troubled Families Programme identify a number of risk factors, both distal and proximal, related to child maltreatment. These can include benefit receipt, a history of offending, poor educational attendance or issues with parental capacity.

The model we aim to pilot will identify these risk factors in a given family and, where they present collectively, an alert will be sent to a children’s services practitioner or our multi agency safeguarding hub to investigate further. The model will help us to identify children who are at risk of maltreatment and target our interventions more intelligently, to prevent escalation into statutory social care.

It’s true this will create more demand in the system, but the demand will be created upstream, where less costly interventions are required, in order to realise savings downstream, by avoiding high-cost interventions, such as children being taken into care. The model prioritises and promotes early intervention, while ensuring that those children who are under the radar are identified and supported.

In addition to generating high-risk referrals for children and young people, the system will also support cluster analysis of which interventions work with different family groups or types, and identification of the families and young people that are least receptive to service interventions. These may then be fast-tracked through the system, negating the need to invest significant resources at the front end.

To quantify the savings we must demonstrate that, but for the action taken, the avoided events would have been very likely to occur, and the avoided events need to be monetised by the finance department. Not all high-risk referrals will result in a very costly care placement being avoided, so both prudence and realism are required.

‘Cost avoidance’ is the most complex part of any business case or financial model and is definitely an issue we have explored at some length, by looking at academic hypotheses and the modelling in the Troubled Families Programme business cases. Getting buy-in to our assumptions from service managers and the finance director has been vital and we have developed high, low and mid-range scenarios which allow for different risk tolerances and give us a banded business case where we can say with some confidence that savings will be between X and Y. We are also working on a unit cost for the suggested intervention so that it can be compared with the cost of the avoided events

Providing robust and rational financial input to the creation of complex new risk-profiling models is absolutely key, but my experience so far has been that, to unlock the savings behind the continued use of big data to inform demand management, there also needs to be a collaborative effort on the part of finance professionals, children’s services managers and technology providers.

This blog is part of the CIPFA Thinks series

Did you enjoy this article?