The Cost Disease: Chapters 5-6

Baumol’s point in Chapter 5 of The Cost Disease is simple. Things in the relatively more productive sectors become relatively cheaper. It just so happens that there are some big problems — perhaps the biggest problems — associated with some of those things becoming cheaper. In large part, that’s because the productive sectors tend to be those that product objects and consume real, non-labor resources.

When weapons are cheap, terrorists can afford more powerful ones. Terrorism gets more terrible. When products that require or consume a lot of carbon are cheap our carbon footprints grow. Climate change accelerates. Relative to rising health care costs, these are potentially larger concerns.

That’s all I have to say about Chapter 5. Moving on…

Baumol devotes much of Chapter 6 arguing that we need not take account of qualitative change to appreciate the cost disease.

Quality-adjusted productivity is a measure of how much benefit consumers are getting from their purchases of a product. Quality-unadjusted data tell us how much money must be raised to purchase the product.

Since you have to spend the money to get the product independent of its quality, he argues that we need not take account of quality to observe the cost disease. Quality is just not relevant to cost, he argues.

His conclusion is certainly convenient. It is very hard to quality-adjust health care. Apart from how much it costs, how much better is health care today than 10, 20, 50 years ago? Being able to discern a cost disease without addressing this thorny problem would be very powerful.

But, I’m not convinced we can put aside quality-adjustment so easily. I look at it this way: The entire enterprise is concerned with cost (or price) per unit of output. What is the unit? Baumol defines his units in such a way that makes quality seem like another dimension. There’s the (asserted) indivisible unit and then there’s the quality of it. But one can define the unit differently so that quality is the same thing as the unit. For example, instead of the heart transplant or the cholesterol drug as the unit — each of which obviously has a time-varying quality dimension — what if one defines the unit as the life-year gained?

The life-year as a unit has not changed. Not ever. There’s absolutely no issue of quality adjustment. [Update: I now disagree with this. See footnote and the link therein.*] To be sure, we don’t price health care by the life-year, not explicitly anyway. But implicitly we do. When more life-years become available in a different form, say with a new pill as opposed to an older operation, many more opt for the new pill, substituting away from the operation. Insurers may provide incentives — with dollars via cost sharing — to do so. Somebody really is buying life-years or, perhaps, quality-adjusted life-years.

From this perspective, productivity in health care goes up when it can deliver more units (life-years) per dollar. If the pill cost less than the operation, productivity enhancement has occurred, even if the operation’s cost has not changed.

It still may be true that productivity growth in this sense is low in health care, that there is still a cost disease. Showing that, or arguing it, would be more powerful than what Baumol does, which is to argue away the gain in life-years. To him, productivity does not go up just because the pill delivers more life-years than the operation it substitutes for. If the cost of the operation in terms of labor input doesn’t change, then there is no productivity gain. If the cost of producing a pill doesn’t change, again no gain. This is all due to the choice of units, and I am not convinced Baumol has made the right choice or that he need make the choice he does.

As a unit of health care, what’s wrong with (quality-adjusted) life-years?

All posts about the book are under the Cost Disease tag.

* OK, OK, maybe I need to stay write “quality-adjusted life-year” here. Still, what I’ve written in this paragraph is not right, which I explain here.


Hidden information below


Email Address*