Concerns about health care expenditure growth and its long-term sustainability have risen to the top of the policy agenda in many OECD countries. As continued growth in spending places pressure on government budgets, health services provision and patients’ personal finances, policy makers have launched forecasting projects to support policy planning. This comparative analysis reviewed 25 health care expenditure forecasting models that were developed for policy analysis by OECD member countries and other international organisations. The study aims to identify good practices, increase transparency and contribute to improving future forecasting work.
We observed that the policy questions that need to be addressed drive the choice of forecasting model and the model’s specification. Policy questions addressed by forecasting models range from estimating the growth in expenditures in the current policy setting; to identifying the sectors where expenditures are rising the most; to testing the impact of slowing growth in public expenditures; to understanding what factors are driving expenditure growth; to testing scenarios about potential policy changes; and to examining the impact of expenditure growth on the rest of the economy.
By considering both the level of aggregation of the units analysed and the level of detail of health expenditure to be projected, we identified three main classes of forecasting models for health spending.
Component-based models are the dominant class, accounting for more than half of all forecasting models surveyed in this study. One reason for their proliferation is a focus on demographic drivers of health expenditure growth. Component-based models are also less data intensive and less complex than micro simulation models. Micro simulation models, on the other hand, are more capable of answering a greater variety of challenging policy questions. Macro models are the least demanding in terms of data requirements and can be fairly straightforward to implement. They are most appropriate for short-term projections, however, as they depend on clear and undisturbed trends. A special category of macro models are computable general equilibrium models. These models are complex and require strong assumptions about the behaviour of individuals, firms and governments.
Models attempt to estimate the relative impact on health expenditure growth of potential drivers. Virtually all models account for demographic shifts in the population and some focus specifically on scenarios about the potential future health status of older people. The models reviewed here, however, point to innovation in health care as the more important driver of health expenditure growth. Innovation can include new medical technologies and treatments, new uses of old technologies, changes in the intensity of treatments, new modes of service delivery and new financing alternatives. Innovation can influence the intensity of care provided to patients, as well as health care prices. Two important influences on health expenditure growth that are the least understood include technological innovation and the role played by changes in health-seeking behaviour and underlying societal norms about health and illness. There is little empirical evidence on these factors upon which models may be developed.
To influence policy decisions, forecasting models must be credible. The assumptions underlying different models often have a strong influence on results. Transparency of methods and assumptions is a prerequisite for model quality, as is a strategy to validate results. The impact of assumptions should also be considered when models are first specified. Short-term forecasts may be valued for their predictive accuracy because, similar to a weather forecast, they aim to predict events where there is little action that may be taken to change the outcome. Medium-to-long term projections, on the other hand, should be valued for their ability to support policy-planning and decision-making. Such models identify where a society may be heading if future trends continue and give policy makers an opportunity to act to modify the course of events. As a result, policy-support models should not be judged against the benchmark of making accurate future projections.
The landscape for health forecasting models is dynamic and evolving. Advances in computing and in detailed health data are opening up new possibilities for the generation of helpful decision-support tools. The review of models highlights emerging systems of models where different modelling approaches are designed to work together coherently. In this way, techniques with different strengths are amalgamated and a broader range of policy questions may be explored. Through the OECD, there is an opportunity for countries to benefit from the lessons learned from comparing forecasting methods across countries to develop and implement an international decision-support platform. Advantages include international comparability of expenditure forecast, through the standardisation of model structure, assumptions, and data; the ability to test and compare the potential results of policy reforms; and the ability to address emerging global issues related to international movement of patients, personnel, services and capital.
Jillian Oderkirk: email@example.com