Since the dawn of file sharing with Napster’s appearance in 1999, the global recorded music industry has seen its revenue in a tailspin, and most observers agree that unpaid sharing of music files is responsible (Oberholzer-Gee and Strumpf, 2007; Zentner, 2006). But whatever the cause, circumstances have put music labels under pressure to develop methods for generating revenue from digital music products. Apple’s iTunes Music Store appeared in 2003 and has become the dominant digital music retailer in the world. Until recently, Apple employed uniform pricing, that is, the same price for each song ($0.99 in the US, £0.79 in the UK, €0.99 in the Euro zone).
Alternative pricing schemes in theory
Microeconomic pricing theory provides numerous alternatives to uniform pricing with the promise of generating more revenue without necessarily reducing the overall benefit experienced by consumers. These alternatives include song-specific pricing (a potentially different price for each song), bundle pricing (a price for all songs together), nonlinear pricing (different prices depending on the number of songs purchased), two-part tariffs (an initial “hookup fee” to purchase any songs plus a per-song price), as well as discriminatory schemes such as person-specific pricing (an extreme form of “third-degree price discrimination.”) These pricing approaches have been developed in a literature that includes Stigler (1963), Adams and Yellen (1976), Schmalensee (1984), and Wilson (1993), among others.
The share of the area under the demand curve appropriable with alternative pricing schemes has implications for welfare economics as well as managerial practice. With uniform pricing, it is possible for revenue from a product to fall short of cost even if the full benefit – the whole area under the demand curve – exceeds cost. The possibility of the resulting inefficient under-provision is generally dismissed with the suggestion that price discrimination could make adequate revenues appropriable. But little is known about the share of the area under the demand curve available under alternative pricing schemes.
Alternative pricing schemes in practice
Implementing alternative pricing schemes requires a great deal of information about the distribution of valuations across consumers and songs. Economists typically infer these distributions from econometric estimates, i.e. from information on the number of persons purchasing each song in the face of varying price configurations (e.g. Chu, Leslie, and Sorenson, forthcoming). We lack such data, and even if we had them, the absence of price variation would make it hard to econometrically map out the distributions. Instead, we employ “direct elicitation,” asking 500 students to report their maximum willingness to pay for each of 50 songs. We then fit a flexible distribution to these data (multivariate lognormal for the positive observations, along with a multivariate probit for the probability of reporting a positive valuation for each song). Using data simulated from these fitted distributions, we can directly implement the various pricing schemes.
We find that uniform pricing captures 27% of the area under the demand curve as revenue. Song-specific pricing – a different price for each of the 50 songs – raises revenue by 3%, to 28% of total surplus. Pure bundling – all 50 songs available as a take-it-or-leave-it proposition – raises revenue by 16% (to 32% of total surplus) in one sample and 28% (to 37%) in another. Two-part tariffs and nonlinear pricing yield revenues nearly identical to pure bundling.
Apple generates far more revenue from selling hardware (iPods) than music. Low music prices stimulate consumers’ demand for iPods. While the revenue-maximising pure bundling scheme raises revenue by 28% in one sample, it reduces consumer surplus slightly. By reducing the price of the bundle from its profit maximising level, consumers can be made better off. We find, for example, a pure bundling scheme that raises the well being of 60% of consumers while raising revenue by nearly 20% relative to uniform pricing.
While some of the non-discriminatory schemes raise revenue by almost 30%, revenue’s share of surplus does not exceed 37% under these schemes.
We then turn to discriminatory schemes. Under person-specific pricing, the most extreme form of third-degree price discrimination, revenue could be raised by 50% to 75%, delivering over half of surplus as revenue. Third-degree price discrimination based on the limited observable characteristics we have in our college-student data, on the other hand, raises revenue by no more than a few percent.
Our conclusion is therefore somewhat mixed. On the one hand, we identify pricing schemes that can raise revenue substantially without hurting consumers. These seem like promising managerial strategies, and indeed, a number of music retailers are experimenting. Apple now has song-specific pricing in the form a 3-tier pricing structure, and Nokia offers all-you-can-eat music bundled with its “Comes with Music” phone. On the other hand, none of the non-discriminatory schemes we examine is able to raise revenue above 37% of surplus, suggesting a limited capacity for sophisticated pricing to capture surplus as revenue.
Adams, W.J. and J.L. Yellen (1976), “Commodity Bundling and the Burden of Monopoly,” Quarterly Journal of Economics, 90, 475-98.
Chu, Chenghuan Sean, Phillip Leslie, and Alan Sorenson (forthcoming), “Nearly Optimal Pricing for Multiproduct Firms.” American Economic Review.
Oberholzer-Gee, Felix and Koleman Strumpf (2007), ‘The Effect of File Sharing on Record Sales: An Empirical Analysis’, Journal of Political Economy.
Schmalensee, R. (1984), “Gaussian Demand and Commodity Bundling,” Journal of Business, 57(1), 211-231.
Stigler, G.J. (1963), “United States v. Loew's Inc.: A Note on Block Booking,” Supreme Court Review, 152-7.
Wilson, Robert B. (1993), Nonlinear Pricing. New York: Oxford University Press.
Zentner, Alejandro (2006), ‘Measuring the Effect of File Sharing on Music Purchases’ Journal of Law and Economics, 49, pp. 63-90.