In Chapter 2 of The Cost Disease Baumol explains what drives the rising costs of the health care, education, and some other service sectors. (All posts about the book here.) With respect to health care, his is not like typical explanations that decompose cost growth into components, like that due to income growth, demographics, changes in extent of third-party payment, technology, and market power dynamics. No, Baumol is attempting to go a bit deeper, to reveal something more fundamental that underlies the cost growth across many service sectors.
In fact, Baumol explicitly rejects the notion that health care costs are growing due to increased market power of providers. I did not find his argument convincing. It rests on two points: (1) physicians per capita has not declined, which suggests no decrease in competition among doctors and, hence, no increase in their market power, and (2) physicians’ wages have declined in real terms, the opposite of what one might expect if their market power were growing.
I have several concerns about this analysis. To begin, it doesn’t identify the relevant bargaining units. The individual physician may not be the right unit. If, for example, physician organizations (group practices) or hospitals also bargain for prices for physician services, it is possible for the price of those services to increase independent of the number of physicians per capita. To the extent the additional revenue is profit, it might accrue to investors and not show up as health care workers’ wages. There is a trend toward greater organization of physicians into groups and hospital employment of physicians, making this story at least plausible.
On the other side, a key bargaining unit is the insurer or self-insured employer. Since most of us are insured by such an entity, how provider market power has changed relative to that of insurers (PDF) determines most of the relevant health care prices. Also of relevance, is the ability (or inability) of public programs, like Medicare and Medicaid, to control the price and volume of services.
Finally, health care workers’ wages represent part, but not the entirety, of health care expenditures. One cannot conclude that market power forces aren’t playing a significant role in growing health care costs without examining spending in other areas of health care, including that for devices, drugs, non-wage hospital costs, and the like. Baumol has not persuaded me to put aside the large body of evidence that points toward the price increasing effects of provider market power.
Nevertheless, let’s make the reasonable presumption that market power doesn’t explain all of the growth in health care costs. What’s the deeper reason? Baumol explains that some sectors are better able to leverage productivity-enhancing technology than others. For example, technology can help us produce more vehicles or computers per unit of labor, reducing the cost of those products over time (controlling for quality). However, technology has been less able to help doctors see more patients or teachers to teach more students per unit of labor. If wages tend to grow at the same rate across all sectors, the inevitable conclusion is that health care and education labor costs (money input per unit output) must rise faster than those for automobiles and computers. That is, health care, education, and other sectors have relatively slower productivity growth. (I’m ignoring for the moment that the number of patients seen and students taught may not be the right units of output in health and education, respectively.)
We should be careful here, and Baumol is, to differentiate between a lack of productivity growth that is inherent to a sector and lack of productivity growth due to constraints we impose on that sector.
At least some medical activities can be carried out better by computers than by doctors. For instance, a set of symptoms may be compatible with a dozen different illnesses—three of which may be rare and apt to be overlooked by physicians—but computers do not forget. Moreover, a filmed lecture by an extraordinarily talented teacher may provide better instruction than the same material presented live by a mediocre local professor. And recorded music is sometimes even preferable to live performance. [...]
[P]rofessors and medical doctors often have an inflated view of the benefits of their personal attendance in the lecture hall and the operating room. These attitudes are widely shared by medical patients, students, and others who benefit from such person-to-person interactions. This creates yet another obstacle to labor-saving modifications in stagnant-sector activities, even as labor-saving efforts are constantly under way throughout the progressive sector. Psychological resistance to labor-saving change in the personal services increases the lag in productivity growth that characterizes these services.
The divide between these stagnant and other services is evident when one examines job losses and gains during the current recession. Between December 2007 and June 2009, for example, the motor vehicles and parts sector lost 35 percent of its jobs—the largest loss of any U.S. employment sector. The twenty sectors with the largest declines in employment also included textile mills (24 percent job loss), construction (17 percent), and manufacturing (14 percent). Ever increasing efficiency in these nonstagnant sectors— a key component of the cost disease—has allowed them to keep up with demand despite laying off workers. In contrast, the stagnant sectors actually gained jobs during that period. Because these sectors are far less amenable to labor-saving changes, they cannot continue to provide their services with fewer workers.
In the short run and during or just after a deep recession, it’s economically advantageous to have sectors that resist job losses. The economy would be in worse shape right now if workers in health care and education lost jobs over the past several years at the same rate as those in textiles, construction, and manufacturing. However, in the long run, sectors that resist job losses because they have low productivity growth is an economic drag. Long term, we’d all be better off if we removed as many hindrances as possible to productivity growth in health care and other services suffering from the cost disease. Note that even if we did that, there may still be sectors with persistent low productivity growth relative to other sectors. They cannot all be above average.