Social preferences have been increasingly studied in theoretical and empirical research. We develop a social preferences-based model to analyse the music industry which struggles to adjust its conventional business model to the challenges of P2P file sharing networks. Conventional online music stores attempt to implement Digital Rights Management (DRM) systems in order to stop illicit copying. However, effective copy protection appears to be hard to achieve as P2P file sharing still thrives. Moreover, common DRM systems restrict consumers in their consumption in various ways (limitations on sampling, burning or transferring content).

The niche label Magnatune (cf. sources) goes another way. It lets consumers choose from a given price range ($ 5 to $ 18). They can pay what they want for music. Moreover, Magnatune offers a free and comprehensive music discovery tool. An online radio service with unlimited streaming lets consumers try out songs they are interested in. We collected a data set of all the label’s transactions over 18 months and analysed the payments that consumers made.

The results of our investigation of this alternative business model – the variable pricing concept of Magnatune – are presented in the following. First we describe Magnatune in detail; next we introduce our theoretical model and highlight the findings of our data analysis. Finally we draw conclusions from the research performed. The complete theoretical model and the full regression analysis of the data can be found in an extended paper (Regner and Barria 2005).

The music label Magnatune
The label was founded in October 2003 and it has around 200 artists on contract. Magnatune prides itself of having a very strict selection process to guarantee high quality. The revenue is evenly split between artist and Magnatune and its slogan is: "We're a record label. But we're not evil." File quality and format is up to the consumer. Even CD-quality files can be downloaded and the formats on offer give a good choice: WAV, MP3, OGG, FLAC and AAC. The payment is variable as consumers can set the price themselves. The given price range for an artist’s album is $ 5 to $ 18 and Magnatune recommends $ 8. The actual price is selected by the consumer in a pop-up menu where $ 8 is the default setting.

Payment is processed by credit card or PayPal. As it is not compulsory to leave an e-mail, consumers can remain anonymous at Magnatune. Albums can be downloaded online or bought as a CD. A fee ($ 4.97) for the physical costs of material and shipping is due for CD purchases. Magnatune is based in the USA, but as an online store it has consumers around the world.

Magnatune’s artists are categorised in various different genres. There is a wide range of music available from classical music to Electronica, Jazz and Blues, Metal&Punk, New Age, Rock and Pop, World and several more. Magnatune can be seen as a niche label that offers music of relatively unknown artists. Mainstream music of famous artists is not sold. Therefore, the focus of Magnatune – and the article’s – is music of less-known artists and subsequently uncertain quality.

Experience good aspects are well taken into account at Magnatune as music discovery is greatly facilitated. Full streaming access to all songs is provided in low or high quality. An online radio service can be used to listen to genre selections or artists’ albums. Visitors of the site are allowed to test the available music as often as they want. Essentially, consumers have all possible means available to sample music and find out how much a song/album is worth to them before having to make a decision on the payment. This stands in stark contrast to the usual practice of conventional online music stores where merely 30 seconds snippets of songs are available for sampling if at all.

Summary of the model
We studied the relationship between labels and consumers using a moral hazard model for our analysis. It takes social preferences (see Camerer 2003) into account and it considers the importance of free sampling of experience goods (e.g. music). Magnatune’s comprehensive pre-purchase access allows consumers to make an informed buying decision. They can experience the information good long enough to determine how much it is worth to them and decide whether they really want to buy it. This full pre-purchase access can also be regarded by consumers as kind behaviour of labels (as it allows them to make an informed choice). Consumers are willing to reciprocate by making a high voluntary payment, if they are socially-minded. Selfish consumers would free ride and would only pay the minimum.

In the model the label offers music online on its web site and consumers purchase albums. The payment of consumers is not enforceable as substitutes are available for free in P2P file-sharing networks. It is therefore subject to moral hazard. Moreover, the value of consumption depends on the amount of pre-purchase access to music. Limited access and restricted sampling mean a lower value for consumers than comprehensive pre-purchase access.

In contrast to a conventional label the set up of Magnatune allows for an open contracts design. The agents have the opportunity to respond to the action of the other. Thus, both sides are encouraged to reciprocate. Fairness and reciprocity can also be regarded as the enforcement device of this contract (see Fehr et al. 1997 for a similar model in a labour market context). The fact that the consumer in our model is free to choose the payment from a given range adds this feature to the contract design. Hence, our model consists of two stages. First, the label decides whether it allows free comprehensive pre-purchase access to the music or not. Then, the consumers make their purchase and payment decision (being kind, e.g. a voluntary payment, or not).

The theoretical model explains when consumers make a voluntary payment. Social preferences are incorporated into the utility function with a reciprocity payoff. This applies the psychological game theory framework developed in Geanakoplos et al. (1989) and is based on the seminal work of Rabin (1993) and Dufwenberg and Kirchsteiger (2004). The utility function of socially minded individuals increases not only in their material payoffs but also in the psychological payoffs which depend on the individuals’ kindness to others and beliefs about that. Essentially, utility increases when the signs of kindness and beliefs about the other’s kindness match. Two equilibria – a negative reciprocity equilibrium (both playing nasty) and a positive reciprocity equilibrium (both playing kind) – are possible. Consumers will prefer to make a voluntary payment once reciprocity gains outweigh the material loss of making the higher payment. However, consumers and label have to be sufficiently motivated by reciprocity for this to happen.

As concerns for reputation do not play a role in this context we conclude that the premium exceeding the minimum price of $ 5 should be motivated by social preferences, e.g. reciprocity.

Data analysis
Our data set goes back to the actual start of Magnatune’s service in September 2003 and contains all 14,367 album purchases from then until January 2005. Upon initial contact Magnatune was interested in research collaboration and hence provided the data. Apart from the payment information we also collected the purchase date, an encrypted identifier of the consumer, his/her gender and country of residence, the artist, the music genre, the means of payment, the type (download or CD) and whether an e-mail address was left or not. In addition to these variables we computed the total amount of purchases and the number of a respective purchase of a consumer. Moreover, we created a dummy variable for female consumers, if no e-mail was left, if a CD was purchased, if PayPal was used and also various country and genre dummies. The number of purchases has been fairly stable over time and there is only minor fluctuation since October 2003.

The average payment for an album is $ 8.197, the median and mode of the distribution are both $ 8. Only 14.5% of all purchases were at the required minimum of $ 5. No time trend can be seen during the observation period. The data has been generated by 7,620 different consumers; most of them (4,986) purchased only one album. On average consumers bought 1.86 albums. The most albums a consumer purchased were 49. Payments made vary between the minimum of $ 5 and the highest price possible.

Further interesting findings are: Consumers who leave an e-mail tend to spend more on a purchase. The average payment is $ 8.23 when consumers left their e-mail, while it is $ 7.82 when consumers preferred to remain anonymous. The two payment options credit card ($ 8.21) and PayPal ($ 8.16) average very similar payments. CD buyers pay a fee ($ 4.97) for the physical costs of material and shipping. Still, the sale of CDs ($ 8.93) generates a higher payment than the sale of downloadable files ($ 8.17). About two thirds of sales come from the U.S. After correcting for currency and GDP differences between countries only minor variations in the size of payments can be observed.

The average payment decreases with the total of purchases a consumer has made. While the average payment for one-time purchases is $ 8.29, the average payment with more than four purchases is only $ 8.06. There seems to be a decreasing individual trend line for frequent consumers. However, the average payment of first-time buyers is stable around $ 8.26 and they are also "joining" Magnatune at a stable rate over the months.

Our model explains the behaviour of Magnatune consumers who consistently pay more than the requested minimum price and even pay more on average than the recommended/default price. We conclude that reputation effects cannot play a role in this environment. Therefore, social preferences are the likely motivation of the consumers that make voluntary payments.

Reciprocity is the source of social preferences in the model. The comprehensive and free pre-purchase access of Magnatune allows consumers to make an informed buying decision. This is regarded as kind behaviour by sufficiently socially-minded consumers and it triggers a kind reaction. They make a voluntary payment, while self-interested consumers only pay the minimum. All consumers do maximise their utility.

Our empirical analysis shows that the average payment is $ 8.20, far more than the minimum of $ 5 and even higher than the recommended price of $ 8. A regression analysis shows that several factors have an impact on the size of the payment. The purchase of a CD (instead of the mere download) has a positive effect as well as some genres, e.g. "rock". The number of purchase and the anonymity of the consumer affect the payment negatively. The dummy for female consumers is not significant.

Compared to a conventional online music store that charges a fixed price of – for instance – $ 8 an album (and offers only limited sampling possibilities if at all) Magnatune makes more visitors acquainted with its songs and thus turns more visitors of the site into consumers; and they still pay more than the recommended price of $ 8.

Bottom line
Still, despite the positive results of voluntary contributions and variable pricing for music it is important to stress that a niche of the market has been analysed and the results for rather unknown artists cannot be easily applied to the mass market. Nevertheless, the open contracts design of Magnatune should be regarded as a promising alternative to strictly DRM-based stores. In an artist life-cycle model it suits artists in an early stage where they are not (yet) well-known. Then, the experience good aspect of pre-purchase access and the higher exposure it allows is relatively more important.

  • Camerer, C. (2003): Behavioral game theory: Experiments in strategic interaction. Princeton University Press, Princeton
  • Dufwenberg, M. and G. Kirchsteiger, G. (2004): A theory of sequential reciprocity. Games and Economic Behavior, 47, pp. 268-298
  • Fehr, E.; Gächter, S.; Kirchsteiger, G. (1997): Reciprocity as a contract enforcement device: Experimental evidence. Econometrica, 65, 833-860.
  • Geanakopolos, J.; Pearce, D.; Staccetti, E. (1989): Psychological games and sequential rationality. Games and Economic Behavior, 1, 60-79
  • Magnatune:
  • Rabin, M. (1993): Incorporating fairness into game-theory and economics. American Economic Review, 83, 1281-1302.
  • Regner, T. and Barria, J. A. (2005): Magnatune: Variable pricing for music", mimeo, online available at:

About the authors: Tobias Regner is a research associate and Javier Barria a senior lecturer at Imperial College London. They have extensive research background in the economics and business aspects of digital media industries. Javier Barria has also been work package leader in several EU-funded projects. The current projects TIRAMISU and DANAE partially supported this work. Both authors are based in the Intelligent Systems & Networks group of the Department of Electrical and Electronic Engineering, Imperial College London, Exhibition Road, SW7 2BT, London, United Kingdom. E-mail: and

Status: first posted 20/10/05; licensed under Creative Commons