James Markels has a piece up at the America’s Future Foundation’s Brainwash webzine puzzling over why Apple insists on charging the same price, 99 cents, for every song on iTunes, whether it’s the top single on the Billboard charts or one of Marky Mark and the Funky Bunch’s lesser known tunes. (I just had a moment of serious cultural vertigo, by the way, when I realized I have no idea who half of the top ten artists on the Billboard pop chart are.) He makes the basic arguments for the virtues of market pricing you’d expect, and normally I’d find those pretty persuasive. The only problem is, it’s basically impossible for me to believe that Steve Jobs (and the army of fresh-faced genii in his employ) don’t understand those arguments perfectly well. They must’ve devoted a lot of expensive brain-hours to the question and decided monopricing was the way to go. So why might that be?

One reason is the flip side of an idea Markels broaches: Prices signal quality. People might conclude that an expensive song is likely to be better (because more in demand), and conversely, might be willing to take a chance on an unknown at a lower price. Yet there’s a pheonomenon familiar to marketers where you can sometimes sell more of a product by raising the price, precisely because peoople do sometimes take prices as a proxy for quality. Maybe not for big-ticket items like cars, where you’re going to do a lot of investigating before dropping tens of thousands of dollars, or for items where the quality can be easily measured by casual observation. But I’m willing to bet that, say, for a pair of headphones, a lot of people who want to get good sound but aren’t devoted audiophiles will follow a heuristic like: “Walk into Best Buy and grab the second most expensive pair.” Now music, like a lot of other cultural goods, is a long tail product: Sure, there are those megahits at the top that sell disproportionate numbers, but most of the action and the sales, especially online, where you’ve got a bottomless inventory, is going to be in the aggregate sales of large numbers of smaller niche artists. (This is especially the case as a kind of online feedback effect kicks in, where record industry economics no longer tend to push convergence for most consumers on a relatively narrow mainstream.) If that’s the case, you might not want to signal that a huge portion of your inventory, which when added up actually accounts for most of your sales, is in the digital equivalent of the remainder bin, especially when you’ve got other sophisticated means of suggesting songs tailored to a person’s specific tastes.

That long tail logic also points to another consideration: People are actually going to be a lot less price elastic than you might think, especially for the niche items. That is, suppose Quasi is selling a lot fewer albums than Kanye West. The normal market conclusion would be that Quasi should be priced lower to move more. But that’s not necessarily the case, because most consumers aren’t actually sitting there making the decision at the margin between Quasi and Kanye. Rather, the people who like Quasi are going to buy it whether it’s at 99 cents or 50, and even if it were 10 cents, Quasi just ain’t going to be most people’s cup of tea. Conversely—and this is more speculative—the items at the top are likely to be stuff for which people have thinner preferences, and are therefore more price elastic. That is, a lot of people are going to download Eminem (or whatever) precisely because it’s the hot track everyone else is listening to, but might be dissuaded by an extra 50 cents.

Note also that one of the major reasons for market pricing doesn’t apply to information goods like downloaded music. If I’ve got a hundred (physical) widgets in stock, I’m going to be able to sell exactly a hundred of them. (Of course, if they’re popular, I may order more, though I as the retailer also have to pay for another batch.) So I’m primarily interested in the hundred consumers who’ll pay the most for it. The finite number of widgets get allocated to the people who place the highest value on them, and I maximize my profits. But the marginal cost of a downloaded song is zero: There’s actually no allocation problem, because I can download the same song you just did. The supply, once the song is created, is effectively infinite. That means it’s going to be especially profitable to sell more units at a lower price.

A final point has to do with that magic 99 cent barrier and what I’m going to call “internal transaction costs.” You see this all the time, of course: Lots of items are sold at $19.99, and many fewer at $20.01, because a lot of people are apparently acting on a heuristic that sees the jump from “less than a dollar” to “more than a dollar” as a kind of special inflection point, whereas they’ll regard the difference between 96 and 98 cents or $1.05 and $1.07 as negligible. Now, you might say this is arbitrary or irrational. But while it may be the former, I’m not sure it’s the latter.

Imagine someone offers you a good you’re somewhat interested in for $10.50. At that price, you think, sure, it’s worth it. Well, how about $10.51? Surely if it was worth it at $10.50 (you might think), it’s just as good a deal at one penny more. But of course, iterate that reasoning and you’ve got an economic version of the sorites paradox: If you really believed that one penny couldn’t be the difference between “worth it” and “too expensive,” the price could be raised indefinitely. The economists’ solution is to suppose that there really is some one-cent margin at which you’d be paying exactly a penny more than the utility you’d get from the good. And as an assumption for formal economic modeling, maybe that works out. But I think what’s more likely is that if we distinguish our actual, psychological preferences from our “revealed preferences” or behavior (something the economist may want to resist, but I think is pretty intuitive to everyone else), is that those internal preferences are actually fuzzy. It’s not like there’s some Platonic form of my real valuation of the good, down to the penny, buried in my head somewhere waiting to be “revealed.” I probably just know I want a certain album roughly so-and-so much—I’ll snatch it up at $10, but not $20.

The way we actually translate those fuzzy preferences into behavior is by using rough heuristics like the “less than a dollar” rule, because it’s probably just not worth spending the amount of time to figure out the one-cent margin at which I want the damn thing or not. And that’s where “internal transaction costs” come in. I’m thinking of things like the “choice paralysis” that Barry Schwartz talks about in The Paradox of Choice, where he suggests that if you present someone with more choices for a particular kind of product, they may just not buy anything rather than going through the bother of figuring out which of a hundred sorts of toothpaste they want. (Or, if they do buy, they’ll probably use some heuristic: “Grab the first one I see, or the name brand, or the one with the prettiest package.”) Markels seems to consider this possibility, writing:

obs also expressed concern that a variable pricing model might confuse shoppers. If he means shoppers that never go to supermarkets, restaurants, car dealerships, or practically any other store in America, he might have a point. [….] Otherwise, shoppers are more than able to weigh a variety of prices to make their purchasing decisions.

But things are likely to be somewhat different for very low priced goods purchased online. Because there, all the other transaction costs are drastically reduced. The time and energy required to acquire the songs is near zero, and you’ve got a fair amount of information readily available to suggest whether you’re going to like a particular song or album (collaborative filtering, reviews). That means the relative influence of those “internal” transaction costs, like translating a fuzzy internal preference into a “buy” decision at a particular price point, rises dramatically. [Cross-posted at Hit and Run]