Today, Apple announced the new incarnation of their so-called hobby, the Apple TV. The new device is focused on streaming, which is a welcome improvement from Apple’s erstwhile expectation that users manage a library of multi-gigabyte movie files. But the more I think about it, the more I doubt that the iTunes Store that is tied to the Apple TV can continue to thrive. I suspect that without a big change in its business model, it will soon begin to show signs of decline.
The iTunes Store has always been characterized by the logic of unbundling. Why buy an album? Just buy the few songs on it you like. Why subscribe to cable? Just download the few shows you follow. But as it turns out, the logic of unbundling is not as compelling as it seems. In fact, when goods have zero marginal cost (as is approximately true for movies and music), bundling is very often superior to unbundling.
Here is a quick example. Let’s suppose there are 1000 consumers and 10 songs. Further suppose that each consumer’s value for each song is distributed uniformly and independently between 0 and $1. What does the demand curve for each song look like?
The optimal pricing scheme for the monopolist rights-holder (assuming no price-discrimination) is to charge 50 cents for each song. It will sell 500 copies of each song. It will earn $250 per song ($0.50 × 500), and since there are 10 songs, it will earn a total of $2,500.
And here is the shocking twist. What does the demand curve for a 10-song bundle look like? Given my assumptions above, it is not a straight line. Rather, it will look something like this.
The optimal pricing scheme for the bundle is not to charge 50 cents per song. It is to charge something more like 40 cents per song. At a price of $4, over 800 copies of the bundle will be sold. This is total revenue of $3,200, which is $700 or 28 percent more than the unbundled revenue.
Bundling not only generates higher profits for the rights-holder, it generates more media consumption. Since there are 10 songs in the bundle, a total of over 8,000 songs are sold; unbundled, only 5,000 songs are sold. As the number of songs in the bundle increases, the demand curve for the bundle gets flatter, a greater proportion of consumers wants to buy the bundle, the rights-holder makes more money.
Suddenly, the iTunes Store’s business model doesn’t look so good. Compare the idea of renting movies on your new Apple TV to having access to Netflix‘s entire streaming library of movies. Netflix is a giant bundle that Apple can compete with, for now, only by having better titles. As Netflix’s licensing division negotiates better content deals (which, due to a more effective pricing scheme, it will inevitably be able to do), Apple’s movie rental business is going to wane.
The music side of the iTunes Store is equally in trouble. Companies like Spotify are offering vast libraries of music for rental for very reasonable prices. iTunes is still the world’s largest legal music retailer, but this might be only because Spotify is not yet publicly available in the US. In fairness, the comparison between iTunes Music and Spotify is inexact because iTunes is selling while Spotify is renting, but the basic logic of bundling remains. As companies are able to negotiate the necessary licenses, the unbundled model will fade away.
However, the logic of bundling also means that Apple has plenty of room to improve its offerings. If Apple could beat Netflix and Spotify to the punch and negotiate the requisite licensing first, they could just as easily offer users an enormous bundle. But another revenue generator that Apple hasn’t touched yet is app bundles. Apple should make it easy for App Store developers to band together to offer their apps in a bundle. You’ve heard of MacHeist? It’s the logic of bundling that makes MacHeist a good deal for both developers and consumers. There is no reason that hundreds or even thousands of app developers would not want to participate in bundles on the App store. Greater app sales would mean that Apple would rake in a huge commission.
A lot of people praise the Internet’s ability to allow us to unbundle. We can use RSS feeds to unbundle a column from a newspaper. We can subscribe to a podcast instead of a TV channel. But this is confusing the unbundling of consumption with the unbundling of purchasing. Newspaper columns and podcasts are usually distributed for free; if they are not, then they are almost always part of a bundle that is designed to be more appealing to most people than the standalone product. Unbundling advocates such as the agitators for à la carte cable don’t see this. They think that if a 100-channel cable package costs $50/month, then a 1-channel subscription will cost $0.50/month. But it is only because the 100-channel package contains channels you are not interested in that the cable company can offer you such a low per-channel price.
But whether you think it is a good thing or not, bundling is going to be a pervasive feature of commerce in the digital age. Both you, dear reader, and Apple should get used to it.
14 Comments
This blog post compares the profitability of pay-per-item versus bundling. But as Adams and Yellen showed in their 1976 paper, offering consumers their CHOICE of either pay per item or a bundle is even more profitable than either pure bundling or pure unbundling.
Further, the analysis you present (based on the research of Bakos and Brynjolfsson) assumes that consumers assign a non-zero value to every song in the bundle. However, whether it is songs (iTunes) or movies (Netflix) many users will assign a zero value to much of the corpus (there are certain misogynistic rap songs to which I personally would assign a negative value!). The Bakos and Brynjolfsson model breaks down under this more realistic assumption.
Marvin, you are exactly right. I am aware of Adams and Yellen. My claim is that a pure unbundling model is not going to succeed in the long run. It’s also true that I assumed that consumers have positive non-zero value for every song (I further assume that values are independent). These assumptions are for tractability. Given a sufficiently large bundle of songs and movies, I think the logic still holds, although at some point it gets to be an empirical claim, not a theoretical one as I presented it.
Marvin: I agree that mixed bundling is typically even better than pure bundling or pure a la carte (at the expense of more complexity and menu costs). However, I don’t agree that the logic of bundling breaks down if consumers have zero value for some songs. It does break down if they have negative value for a significant number of songs and their is no free disposal, or if marginal costs are significant.
For those who are interested in the original papers that Yannis Bakos and I wrote they can be found here: http://bit.ly/apMdgX
Bakos, Yannis and Brynjolfsson, Erik (January 2000) Bundling and Competition on the Internet: Aggregation Strategies for Information Goods, Marketing science, Vol. 19, No. 1 pp. 63-82.
Bakos, Yannis, Brynjolfsson, Erik and Lichtman, Douglas (April 1999) Shared Information Goods, Journal of Law and Economics, Vol. 42, No. 117.
Bakos, Yannis and Brynjolfsson, Erik (December 1999) Bundling Information Goods: Pricing, Profits and Efficiency, Management Science, Vol. 45, No. 12 pp. 1613-1630.
Bakos, Yannis and Brynjolfsson, Erik Aggregation and Disaggregation of Information Goods: Implications for Bundling, Site Licensing and Micropayment Systems,” in Proceedings of Internet Publishing and Beyond: The Economics of Digital Information and Intellectual Property.
The iTunes store has always offered some bundled options–you can buy albums cheaper than buying each track individually, and TV season passes cheaper than buying episodes individually. So it has never had a pure unbundling model.
Doesn’t bundling break down in the face of competition from non-bundling competitors? Customers who like all the songs in the bundle will buy the bundle at a discount and those who simply want individual songs will find a store that sells that. In monopolistic competition, your analysis holds. I am not knowledgeable as to how exclusive Apple is with their music sales (I pirate all the music I want).
azmyth, it’s the rights-holders who have a monopoly. My intuition is that it does not matter if Apple has exclusivity or not.
At the bottom of this analysis is a hollow construct, a monopoly without leverage over the market in which it operates. Imagine if television networks were designed to break programming into discrete pieces and then to enable (even rely on) downstream redistribution and recombination of those pieces by TV owners. How would demand conditions favouring absurd hundred-channel bundles have obtained in the first place, let alone held firm?
The design of the Internet is unbundling logic made manifest. Spotify is an Internet business. Spotify’s bulk streaming model, if it succeeds, must continually contend not only with other models of selling the same music, but with its subscribers’ collective will. Recall how Napster induced à la carte mp3 sales, or how iPhone jailbreaking induced the App Store, or how Youtube induced Hulu. On Soviet Internet, consumer out-leverages monopolist.
Jason, I find your comment confusing to the point that I’m not sure what your objection is. Do you mean to imply that iTunes is a monopoly? That’s not my contention. Rights-holders are monopolists over their own content. What do you mean by “leverage” (you clearly don’t mean leverage in its most common—financial—sense)? Do you mean market power? If so, how am I improperly addressing market power? Also, do you mean to suggest that consumers can’t demand hundred-channel bundles until the hundred-channel bundles exist? I stand by the original post, but thanks anyway for your comment.
“Jason, I find your comment confusing to the point that I’m not sure what your objection is.”
Sorry. I write bad.
“Do you mean to imply that iTunes is a monopoly?”
Not at all. iTunes is a middle player, vulnerable as any. What I’m saying is much more basic: there are no monopolists in sight. To ground an argument in rightsholder monopoly is to wax nostalgic for a media universe where scarcity regulates distribution and where choke points keep out unwanted competitors. Ours is not that universe.
The iTunes Store’s relative success brings with it the slow demise of cable television, the closed mobile feature set, the album, and other pure bundling gambits. Same story with AdWords decimating newspaper revenues. Same again with Hulu. These trend lines are not coincidental. They mark early milestones in a radical renegotiation between the content industry and the networked public. Napster was the opening volley of this renegotiation.
“What do you mean by ‘leverage’ (you clearly don’t mean leverage in its most common—financial—sense)? Do you mean market power?”
I was (mis)using the dictionary sense of leverage as remote influence. Market power captures it best. Thanks.
“If so, how am I improperly addressing market power?”
As above, rightsholders lack monopolistic influence over distribution. I suppose they could withhold original works from the public out of spite, but that’s not market power. No entity on the web can fabricate conditions of exclusivity or scarcity that once held pure bundling models like the album and the TV channel intact.
“Also, do you mean to suggest that consumers can’t demand hundred-channel bundles until the hundred-channel bundles exist?”
Not can’t – won’t. Any more than demand will naturally rise to grocery stores forcing foods of mixed popularity into hundred-item bundles. It’s not that Spotify’s bulk streaming can’t succeed on merit. If it offers pricing advantages, selection advantages, and ease-of-use advantages over the iTunes Store for most consumers, it might well. But if Spotify measures up poorly on any criterion, the logic of bundling won’t save it or any middle player from the disruptive pressures of a wide-open market.
I think it is worth considering that Apple is not operating exclusively in the sale of music or movies. Digital music and movies are obviously complements to ipods and ipads. In fact since monopoly rights to music and movies are held by other entities I suspect that Apple makes little off of music or movie sales, but as we all know they sell a huge number of portable electronic devices. Unbundling of music likely brings more people into the market for ipods. Perhaps the record companies should have told Apple that they could not unbundle their product, but I think Apple may be the savvy one in this case.
Andrew, that is indeed something worth considering. I don’t have a formal model, but my intuition is that bundling would increase demand for iPods more than would unbundling. Much more music is sold under the bundling model. That should make iPods more useful, no?
With bundling I agree that more music is sold however fewer customers buy that music. Take your example with 1000 customers with iid uniform preferences. Under bundling 800 of those customers will purchase music. Without bundling in expectation almost everyone will purchase some music( 1 in 1024 will not purchase any music). I think the impact of this on the demand for ipods is not completely certain as under bundling the customers who do purchase the music bundle will have a higher value of ipods(as your response to me noted) than they would in the unbundled regime, but the customers who purchase music in the unbundled regime who would not have purchased under bundling would have a higher value for ipods without the bundling. I think this would end up resting on preferences, but I do not think it is hard to imagine preferences such that unbundling will lead to more demand for ipods.
Andrew, I agree with your comment completely. It does rest on preferences, and as you say, it is not hard to imagine preferences that would lead to higher demand for iPods under unbundling. In the real world, remember that the bundle would be far greater than 10 songs, which suggests that far more than 80% of customers would buy the bundle, and that values are not iid uniform, which probably means that far fewer than 80% of customers would buy it. As other people have said, there are opportunities for mixed bundling. I don’t have empirical data on music preferences, but I still suspect that tying to iPods doesn’t save the pure unbundling model.
I’m late to this party, but I was just being introduced to formal models of bundling in Tabarrok’s IO class. Eli, did you use Stata to run a market simulation? I’d be interested in looking at the code that generated the data and the resulting charts, if you have it to share.