Time to hike up my trousers and wade into some intellectual flotsam. In my entry on the politics of Scott Walker's anti-unionism, there's a comment demi-thread about the Baumol "cost-disease" argument started by Glen McPhee that I want to bring into a main entry, because the fallacy embedded in it is similar to Tyler Cowen's argument about the disappointments of the internet and even David Cameron's attempt to map Britain's Gross National Happiness. There–I can get from art economics to Mihaly Czikzhentmihaly's field in a single sentence.
William Baumol's original argument in the 1960s was to point out that while gross measures of productivity in various "widgety" sectors of the economy showed increasing productivity per labor cost, areas such as performing arts did not–the most commonly-used example I've seen is that a quartet playing a Mozart piece takes roughly the same number of person-hours in performance today as in the 18th century, depending on the tempo chosen. Loads of people (including Baumol) have tried to apply this argument to education, suggesting that "Baumol cost-disease" is inherent as long as education is labor-intensive.
For a moment, let's skip the empirical fantasy of the argument in higher ed–adjuncts and graduate assistants would have something to say about the claims that higher ed's labor costs keeping going up and up–and focus on the broader assumptions of the argument. There are a few glaring flaws in this, starting with a selection bias: over any stretch of time, there are going to be some sectors that have higher productivity increases than others. There is no way to avoid this. To me as an historian, what might be interesting is to see how the nature of those low- vs. high- change sectors of the economy change over broad swaths, because (again, I'm biased as an historian) one would expect there to be swings in which sectors have higher or lower productivity increases, and that would tell you interesting things about economic changes. That's not what Baumol chose — instead, he drew a broad generalization from a relatively slim set of data. It's about as sensible as arguing that women are marrying less than historically because a lower proportion of women marry by 25 than did so in 1950… forgetting that the immediate post-WW2 days witnessed a dramatic drop in the average age of civil marriage in the U.S. Key warning flag for the historian: arbitrary choice of beginning and end dates for trend analysis.
But it wasn't just slim data–it was a very narrow framing of the data. I will let any reader who is a firm economist comment on the choice of organizations and the assumption that the only employees in orchestras and other performance "firms" are the artists themselves. Let me look at the whole framing of the issue as one of producing live classical ("high-culture") performances as the unit of analysis. Why is the relevant unit the live performance instead of the number of performances listened to or the time spent listening to classical music (or otherwise experiencing performed art)? In the last century, recorded music has dramatically expanded the audience-hours for performance in any genre. Even before the internet, I could derive hundreds of hours of listening pleasure for a relatively slim investment in vinyl or CD recordings produced by Deutsche-Grammaphon and others. And if you're willing to listen to ads on the radio, you've had access to payment-free music (from your perspective) as long as you have had control of a radio dial. I strongly suspect that the average cost of each audience-hour for classical music dropped dramatically between 1940 and 1995 and has plummeted since. If you expand beyond a specific genre (and we should–which productivity statistic measures the production of anything similarly niche, such as "widgets of Scandinavian design style"?), you'd see an even more dramatic difference. To the extent that Baumol "cost-disease" exists in classical music, it's as much an artifact of a niche market (those people interested in listening to live performances by professionals) as some deep underlying labor-intensive nature of performance.
But there's another deep problem with the Baumol argument, or rather the sloppy application of it by many of its fans, which is the assumption that there is a fixed economic value to an hour of Mozart quartets, or more broadly the assumption of fixed value. All other things being equal (or, if you wish to use the great economist's Latin phrase, ceterus paribus), a live performance of Mozart K.421 in 1932 is worth the same as the identical live performance in 1965 (K.421 is the quartet in D minor, if you're curious). But that's a very broad assumption, the same assumption as Tyler Cowen's argument that we overvalue government-spending's contribution to the economy and undervalue voluntary work. Or that the gross domestic product needs to be revised to capture happiness because standard economic statistics don't capture how people are feeling. All assume that there is some essential value to stuff and experience.
But all of that flies in the face of the neoclassical assumption that a perfect market can set prices (and thus values) in different ways at different times.. and thus that there is no inherently fixed value in a widget, a quartet, or anything else. You can argue that it's hard to discern what the "true" market value would be if you don't have a functioning market, but you can't easily argue that the value of an hour of a certain experience is the same everywhere and at all times, and that is inherent in the Baumol argument. (And that's skipping the question of whether a dollar always has the same value–see Viviana Zelizer's work on that point.)