Why we won't die of Baumol's Disease

I was at work one night a few weeks ago, waiting for the 3rd movement of the Shostakovich 1st cello concerto to end (no slap at our soloist, Johannes Moser, who played wonderfully, but it’s a long sit with no need to count), when, for some unaccountable reason, I began to think about Baumol’s cost disease.

The best explanation of Baumol’s disease is likely the one given by William Baumol himself in an interview he did with Paul Judy for Harmony:

Any economic activity affected by it will tend to rise in cost persistently and at a rate faster than the economy’s rate of inflation, obviously leading to financial pressures for anyone who supplies the product. Orchestral music is a prime example. The reason the problem arises is that orchestras experience little or no labor-saving innovation-no growth in productivity. A Haydn symphony written to be performed by 30 musicians and lasting one-half an hour will require 15 person-hours of human labor for an “authentic” performance, no less than it did a the end of the 18th century. But elsewhere in the economy it takes less and less labor every year to produce a product. The amount of labor needed to produce an automobile declines more than three percent each year, on the average.
This means that if wages rise at more or less the same rate in car production and in orchestras, then clearly, the cost per performance must rise faster than the cost per car, because in car production rising wages are offset by the reduced use of labor per car, while in orchestras there are few such offsets. Thus, orchestra costs are condemned to rise every year, cumulatively, at a rate faster than the average of the economy’s prices; in other words, faster than the rate of inflation.

Baumol’s cost disease is often cited by those who believe that the orchestra business is in bad shape and likely to be in worse shape in the future (although it does have other applications). I’ve been thinking about Baumol’s disease (BCD) for a while. I even blogged about it in March 2008. I got some blowback for that; Greg Sandow wrote:

There’s also a notable music blogger whom I won’t name, someone I otherwise greatly respect, who – though he probably wouldn’t tell physicists what’s wrong with the theory of relativity – runs out to tell the world why Baumol was wrong, why Baumol’s dilemma shouldn’t be taken seriously.

But my point was not that Baumol was wrong in his observation; as a theoretical matter, he clearly wasn’t. The real question was whether BCD has anything useful to tell us about our business. I don’t think it does.

The fundamental problem with using BCD as an analytical tool regarding orchestras is that it assumes that musician costs are an economic input that behaves like a labor cost. The reality for full-time orchestras, though, is that musician compensation is functionally much more like a capital cost.

Most observers of the orchestra world will concede that there is a relationship between what musicians are paid and how good their orchestras are. It’s hardly a linear relationship, and there are both orchestras that play better than higher-paid competitors and orchestras that play worse than lower-paid groups. But most orchestra insiders would concede that it would be virtually impossible to duplicate, on a long-term basis, the quality of Cleveland or Chicago or Boston (or Berlin or Vienna, for that matter) without paying musicians full-time salaries comparable to those orchestras.

So, if a community wants an orchestra of the caliber of Cleveland or Berlin, a necessary (if not sufficient) condition for doing so is paying 100+ musicians salaries that are competitive with Cleveland, Berlin, et al. That means that, regardless of how many concerts the orchestra does, it needs to make a minimum annual payroll of around $15 million. Whether the orchestra does one concert a week for 20 weeks or 4 concerts a week for 50 weeks, to attract the necessary level of talent, that orchestra will have to have, on staff, 100+ musicians making a minimum aggregate compensation package in the very low 8 digits.

When an enterprise has to make a big-ticket expenditure before it even starts production, that expenditure functions like a capital expense. If, for example, a small community in North Dakota decides it wants major-city quality air service, it could decide to buy (or lease) a Boeing 737 and then put bids out for a management to run said airplane. There would be lots of expenses management would have to pay out to run the new airline; management payroll, flight crews, maintenance, landing fees, and fuel, and it would need to bring in sufficient ticket income to pay those expenses, unless the community is willing to underwrite operating expenses as well as capital costs.

But one expense it wouldn’t have to cover is the cost of the airplane; the small community had to provide one entire plane as a minimum condition for going into business in the first place. Renting half a plane, or renting the plane on a per-flight basis, is not going to work. The new airline needs to have unhampered access to at least one airplane, whether it runs one round trip a week to the nearest big city or manages to create a route system that keeps the plane flying 20 hours a day.

In other words, there are two kinds of costs associated with creating this service from scratch: the cost of being able to have the service at all (the purchase or lease of a 737) and the cost of actually running the service (salaries, fuel, etc.).

An analogy closer to home would be from the museum world. Art museums generally don’t appear to pay for the acquisition of their collections from what we would consider “earned revenue” – paid admissions and the like. The artwork comes either from direct donation or out of what we would call a capital campaign – ie, a fund drive devoted specifically to funding a permanent addition to the artistic capital of the institution. Again, there are two different kinds of costs associated with the existence of an art museum: the cost of being able to have the art in the first place (acquisition costs or their donated equivalents), and the cost of being able to display the art to the public (utility bills, guard and curator salaries, and the like).

The key point missed when BCD is cited as a problem for orchestras is that musician salaries function far more like capital costs (the lease of an airplane or the acquisition of paintings) than like operating expenses. This is because, as I stated above, having an orchestra of a certain quality requires a substantial base salary expense, which functions as a necessary but not sufficient condition for actually being in business as an orchestra.

So why does this mean we won’t die of BCD? Stay tuned.