In the very first few weeks of Econ 101 students are introduced to the “demand curve”, relating how changes to the price of a product affect the quantity demanded of the product, all other things held equal. I’ve spent many years drawing these on blackboards, but they are a lot more easily drawn than estimated, and firms faced with the real-world problem of setting prices have to meld a bit of know-how, awareness of their market environment, and the occasional experiment with (slightly) changing a price to find an optimum. I even wrote a short book about it, though caveat lector: it’s a collection of rules of thumb, without magical shortcuts.
So it caught my eye when blog-neighbor and fellow cultural policy scholar Sunil Iyengar noted the following, in his review of SMU / Data Arts report on “When We Re-Open, Whom Will We Gather: A Study to Help Advance Audience Diversity, Equity, and Inclusion as Performing Arts Organizations Re-open Following the Pandemic”:
Changes in ticket pricing also can make a difference: “Our analyses indicate that a 10 percent targeted price discount can increase income representativeness [among audiences] by nearly 3 percentage points,” the authors write. (Lest this finding be received as a no-brainer, it is worth sharing that a previous literature review by the [National] Arts Endowment found only “mixed evidence” that ticket price reductions boost arts attendance in general.)
So, What do we know about this? The NEA review is in line with what I have found in my own surveys, that demand for the arts is pretty price inelastic – I cover some of this ground in a paper I wrote on museum pricing. But all those demand curves I draw in class slope downward: why don’t arts ticket prices work this way? There are a few factors. One is that ticket price is only a part of the cost of attending an arts event: the hours of time involved for someone to attend an arts event are a larger share of the total cost of taking part, and that’s before factoring in any necessary child care and transportation costs. So a 10% decrease in ticket price is a smaller-than 10% decrease in the cost of attending the show. And arts organizations are working with a limited pool of potential audience members: only a small fraction of the population will attend a live opera performance even with free tickets in hand. A price decrease might get that small fraction to attend more often, but it might not do a lot to attract new audience members.
So what can we make of the “nearly 3 percentage points” claim made above? What I think is going on is that real care has to be taken with comparing data on prices and audience characteristics, because the prices were set in the first place as a result of local audience characteristics.
Let me illustrate with an example. Imagine two theatre companies, located in Alphaville and Betatown. Alphaville theatre, after carefully considering its local market, and experimenting a bit with different price options, settles on an average ticket price of $20. That price yields a particular audience demand, and, embedded in that, some level of “income representativeness.” Betatown theatre also does some market research, and a bit of experimentation, and arrives at an optimal average ticket price of $18. It turns out that in Betatown the audience will have 3 percentage-points more income representativeness – it might be a town with more diversity in income to begin with, and indeed that was a factor in choosing the $18 average revenue.
Now if I just took that data, I might be tempted to say “it looks like cutting average prices by 10 percent leads to an increase in income representativeness of 3 percentage points.” But that’s not true: from this data we have no idea whether cutting the average price of tickets in Alphaville from $20 to $18 would have any effect on income diversity in the audience, because we don’t have any data on Alphaville’s very specific market. And it might well be the case that in Alphaville, what we have is what most econometric studies, reviewed in the NEA analysis linked above, have shown – that it’s hard to change audience characteristics through price changes.
There is no universal “demand curve” for the arts, or even for theatre specifically. Each company has a unique situation based on where it is, and has to find an optimal menu of prices to offer given their unique situation. We cannot compare theatre prices in Tulsa and Tacoma, and suggest that Tacoma could achieve Tulsa-like audience demographics simply by looking to Tulsa prices.
It’s why “program revenue per attendee” is not at all a good performance metric. In the first place, the metric can always be raised simply by increasing prices (indeed it can be maximized by setting a price so high that you get exactly one audience member), and can always be lowered simply by lowering prices. But in the second place it neglects that the optimal program revenue per attendee will always be a function on local demand conditions, such that a low number might be optimal for the Boise theatre but not for the Boston theatre.
Smart arts organizations, pricing strategically, know that their optimal prices are completely dependent on local conditions. When looking at price data, we need to know that prices are endogenous, and are already set with market conditions in mind.