Raphael Wittenberg explains the role of CPEC (formerly PSSRU) over the past decade in contributing to policy debates on funding long-term care.
I joined CPEC at LSE at the launch of the LSE branch in 1996, when the Department of Health seconded me part-time to set up the long-term care projections model, one of the three main programmes it was funding at the LSE branch of the Unit. Our aim was to project how many older people would have care needs, the types and volumes of services that would be required, and the cost. It was the Department’s first big venture into modelling long-term care, and CPEC, as the leading English research unit in the economics of long-term care, was the obvious academic institution to host the work. I started by looking at what long-term care modelling had already been done, especially in the USA and UK, in order to think through what we wanted to achieve and work out the model design, given that it was important to get it running reasonably quickly. At first I worked alone, with support from Professor Bleddyn Davies (the founder of CPEC), and then gradually recruited colleagues to join the team.
The projections model is an aggregate or “cell-based” model. This means we look at groups of people rather than individuals. The model categorises the older population using a series of variables such as age, gender, household composition, level of disability, and housing tenure. We end up with about 400 different groups of older people, and, using official population projections and other data, we estimate how many people will be in each group in the baseline and future years, e.g. 2010, 2020, 2030, 2040. We conduct analyses of various sources of data to estimate the proportion of people in each group who receive different types of care. We then assume in our baseline projections of future demand and costs that these proportions will remain constant for each group. These are projections not forecasts: we make a number of assumptions – for example about future population growth, annual increases in the costs of care, and disability levels by age and gender – and conditional on these assumptions we can then say expenditure is projected to rise from £X billion to £Y billion over a specified time period.
The first external use of the model was for the Royal Commission on Long Term Care of the Elderly, which had been set up by the Blair government in December 1997 to look at how the cost of care should be split between public funds and individuals. The Department of Health and the Commission itself both wanted modelling of the likely future costs of long-term care under the existing system and various potential reforms. We had considerable interaction with the Commissioners and the executive summary of our report was reproduced as one of the Commission’s research papers in March 1999. This work laid the foundations for the model’s role in the ongoing policy debate on long-term care funding, with the government repeatedly returning to us for further analyses on subsequent occasions.
Another set of projections of long-term care financing was commissioned by the Institute of Public Policy Research think tank in 2002. This work launched our collaboration with Professor Ruth Hancock, then at Leicester University, who had developed a microsimulation model, known as Caresim, which looked in detail at how the means tested system for state funded social care impacted on individuals. Unlike an aggregate model, a microsimulation model incorporates information on hundreds or thousands of individuals, using information from a data set such as the Family Resources Survey on age, gender, marital status, home owner status, assets and income etc. For each person one can then, for example, ask: given their profile, how much would they have to contribute to the weekly cost of a residential care home place under different variants of the long-term care funding system? Using many individuals one can build up a national picture of the impact of changes to the means-testing rules, including an analysis of the ‘winners’ and ‘losers’ of any reforms of the funding system.
The two models worked well together. Our model projected the numbers of people who might need care and the types of care; those results fed into the Caresim model which looked at how much the local authority and individuals might respectively pay; and those outputs in turn fed back into our model for total UK or England projections of health and social care expenditure on long-term care for older people. Most of CPEC’s subsequent projections work on long-term care funding has been in collaboration with Professor Hancock (now at the University of East Anglia). This has included projections on older people for the Joseph Rowntree Foundation and the Wanless Social Care Review. This is separate from the Wanless team’s own modelling, for which they developed a new dynamic microsimulation model.
In the run-up to the 2009 Green Paper on social care we produced for the Cabinet Office the first projections model for long-term care of younger adults. More recently, we carried out modelling work on younger adults for the Commission on Funding of Care and Support, headed by Andrew Dilnot, plus additional work on older people. As well as producing projections of need and costs, we looked at the impact of variants of the Dilnot reforms, for example the cost of different levels of the cap on private contributions to long-term care costs. We have also conducted modelling for individual local authorities and for Wales, and we have advised on building and using projection models for other countries, including in a European Union funded project with colleagues comparing the long-term care funding systems in England, Spain, Germany and Italy.
It’s fair to say that since 1999 no major policy decision in England on long-term care has been made which has not drawn on some results from the CPEC projections models. We have now also developed different versions of the model, including one for projections relating to dementia. This is not to say that our projections are always fulfilled, as assumptions are not always borne out. For instance, in recent years, notwithstanding the increasing number of older people, adult social care expenditure has fallen in real terms because of the austerity measures, which we could not have predicted.
Is there value in doing these projections? Absolutely. Social care policy is about huge, multi-billion pound programmes, which potentially have a big impact on millions of individuals and their families. The more evidence that Government ministers, parliament and the public have before them to inform the debate, the better informed (hopefully) the decisions will be.