Medical Care Spending: How Big a Bite?

It should come as no surprise to anyone that, partly as a result of a persistently higher-than-average rate of inflation in medical care, consumers have been spending a larger and larger fraction of their income on medical care.

So how much of household spending actually goes to health care? Well, that depends on whom you ask.

One good estimate is provided by the thorough surveys done by the BLS to try to estimate the consumer price index. The BLS provides a measure of the relative importance of various sorts of expenditures in the average household budget. The following chart shows how the portion of household spending devoted to medical care has steadily grown over the past two decades. The chart breaks spending down in to two parts: direct out-of-pocket spending on health care, and individual premiums for health insurance.

It’s interesting to note that out-of-pocket health insurance premia actually remained quite constant through much of the 1990s, only to resume their rapid rise since the year 2000. Most of the explanation for this is the massive shift of workers into HMOs during the 1990s, which were generally cheaper than traditional health insurance plans, in part because they covered less medical treatment.

Another measure of the importance of health care spending can be gleaned from national income accounting statistics from the BEA. The next chart shows the BEA’s measure of spending on medical care as a percent of total personal spending, and in real (inflation-adjusted) terms.

A couple of comments. First, the BEA figures show a much greater proportion of spending going toward medical care, largely because the BEA includes payments by health insurance plans for medical care while the BLS figures shown in the first chart only include out-of-pocket payments by households.

Second, it is very interesting to notice that spending on medical care as a percent of total spending shows the same period of stability between 1992-2000 as the BLS figures on health insurance spending showed above. The most compelling explanation for this period of stable health care spending of all types is therefore also the widespread shift of workers into HMOs during the 1990s.

Finally, notice what’s happened in the past few years: after that long period of stability, spending on medical care has risen dramatically since 2000. Presumably, everyone who could be shifted into an HMO had made the switch by 2000, so that source of cost savings had been exhausted.

Most interestingly, the BEA data attributes much of this increase in health care’s budget share to an increase in real medical care spending, not to inflation. In other words, the rapid advances in health care spending during the 2000s are largely due to the fact that individuals are consuming more health care, according to the BEA.

This seems rather at odds with the experiences of most people that I know. What are some possible explanations? I can think of a few. It may be the case that HMOs are simply covering more medical services than they used to, and so individuals are indeed consuming more. Or, it may be the case that the overall quality of medical care has improved, so that (for example) delivering a baby or getting a knee replaced is safer and more effective than it used to be. In that case, 100 procedures performed today will be measured as greater consumption of real medical services than those same 100 procedures performed five years ago.

A third possible explanation involves mis-measurement of inflation in medical care. It is extremely difficult to differentiate between pure price changes and changes in quality when it comes to medical care. The BLS document that I cited yesterday notes the following:

There is a growing consensus that the most appropriate way to measure hospital services is by tracking treatment outcomes rather than medical inputs consumed. From this vantage point, a day spent occupying a hospital room and the time in an operating room are not separate consumer services, but individual components of one hospital visit which may be all or part of a medical treatment. The current CPI method follows medical treatments, a method that lies between the old medical-inputs method and the ideal medical-outcomes method. Measuring the value of different treatment outcomes is the subject of research in the industry, but not yet a feasible methodology for the CPI medical care indexes.

If it is the case that receiving any particular outcome (e.g. getting that knee replacement mentioned above) involves more elaborate and costly procedures but yields substantially the same outcome, then current statistical methods will show an increase in real medical services although they yield the same outcome as before.

I don’t have the answers as to what is underlying dramatic change in medical care spending of the past few years. But clearly its an important question, because it plays a crucial role in understanding whether the rapid rise in people’s spending on health care in recent years is a good thing or bad.

Kash