How did they use to forecast health exp.?

Projection Exercises for Us Total Health Exp. in the ‘80s and ‘90s

A premise is necessary: this is only a simple graphical evidence suggesting some naïve economic readings and interpretations. Absolutely nothing more.

A rapid investigation on the web about old works dealing with health care expenditure projections for Us gives the results that the necessity of forecasting was perceived at least since the ‘80s, that works available on-line are not so many though they can be considered scientific because of their publication on scientific journals/reviews, and that normally the longest horizon of projections was ten years although sometimes also longer horizons were considered.

What can we see if we compare the real evolution of the incidence of Total Health Care Exp. over Gdp with projected values hypothesised some years before? It is just what shown in the graph, where the coloured dots represent outcomes of past rounds of projections. The dots are labelled with the years in which projections have been performed.

It is worth underlining the following points:

• Looking at the entire time horizon (1960-2012), THCE shows a huge dynamics, with an incidence on Gdp that has more that tripled in fifty years, without any lasting evidence of saturation;
• There is a clear positive long-term trend, along which it is possible recognize phases of, let us say, “normal” trend (in the sense of trend aligned with the long-term evolution), phases of acceleration, and phases of slowing down with apparent consolidation;
• The predictive capacity was good in periods when the recent trends of expenditure were, in their average, more or less aligned to what would be the future long-term trend (of course analysts did not know, by that time, what future trends would be). Conversely, it proved out to be bad in periods of evident acceleration / slowing down of the rate of growth of expenditure;
• The previous point can be explained on the basis of the structure of projections models/tools. Only recently (last fifteen years) they have been improved with the introduction of profiles of per-capita expenditure by sex and age brackets, and with the development of multiple scenarios supported by a wide range of sensitivity analysis. In the ‘80s and the ‘90s (see economic literature enclosed), projections were based mainly on extrapolations of recent historical trends and on the demographic change. This probably made projections too sensitive to recent trends and for this reason also to conjuncture and short-term policy interventions.

We could perhaps wonder if this evidence could bring any information about the strength of the drivers that are leading the long-term trend of THCE and that are persisting over decades. Historically, when analysts came from periods of rates of growth aligned to what now we can call the trend of the last fifty years, they performed good ten-years projections. The contrary tended to happen when analysts had to consider, in the extrapolation exercise, also periods of rates of growth falling significantly outside what would be the long-term trend (either over or under).

Is it a naïf proof of the relevance of drivers pushing THCE? Of course there is the need to do more "historical investigation" about past projection exercises, but the guess raised here may be perhaps of some interest not only from an historical point of view but also for helping to improve projection tools currently in use.