International Choice Modelling Conference, International Choice Modelling Conference 2017

Font Size: 
Don’t judge a book by its cover: A discrete choice model of cultural experience good consumption
Paul Crosby

Last modified: 28 March 2017


Don’t judge a book by its cover: A discrete choice model of cultural experience good consumption


Paul Crosby, Ph.D. Fellow,
Department of Economics, Macquarie University, Sydney, Australia

Paper proposed for the International Choice Modelling Conference 2017
Cape Town, South Africa, 3-5 April 2017



Technological change is transforming the book industry. In particular, the digitisation of books has given rise to a suite of new content delivery formats for publishers, such as the ebook and audiobook. While the rise in popularity of the ebook has been well documented (Wischenbart, 2013),[1] the format’s influence on the future of the book industry is a matter of much conjecture and uncertainty (Carreiro, 2010).[2] This conjecture and uncertainty is perpetuated by the fact that books are a cultural experience good, that is to say books can be said to possess varying degrees of cultural content, which may give rise to the presence of a cultural value that exists above and beyond a books value in exchange. Furthermore, the experience good qualities of a book render them a product for which quality and value is indeterminate prior to consumption. As is well documented, the neoclassical framework struggles to account for such factors, therefore declarations on how consumers value particular attributes of a book (such as format, genre, critical consensus and the like) remain anecdotal and, importantly, the impact such attributes have on consumer choice remains a matter of speculation.

This paper attempts to allay such speculation by conducting a stated preference discrete choice survey that examines consumers book consumption choices. In this online survey a representative sample of 250 Australian book readers were tasked with choosing from a selection of books, divided between four discrete format alternatives: paperback books, hardback books, ebooks and audiobooks. Interacting with each of these alternatives are a suite of attributes, with each attribute corresponding to a particular characteristic that makes cultural experience goods, such as books, so unique.

Specifically, the inclusion of a genre attribute[3] in the choice model permits an examination into how the amount of cultural content inherent to a book impacts format choice. The inclusion (or perhaps more importantly, the exclusion) of an attribute that indicates a measure of crowd sourced opinion (for example, consumer book reviews) permits inferences on how imperfect information can impact a consumers purchasing decision. Furthermore, in line with Lévy-Garboua and Montmarquette (1996)[4] the inclusion of attributes that measure a consumers prior experience with an author (degree of knowledge) and intrinsic taste for reading allow for the investigation of ‘learning by consuming’ effects. That is to say, one can investigate in detail the possibility that a consumers prior consumption experience with similar types of cultural experience goods (be they positive or negative) may influence current consumption choices. As is standard in such experiments the inclusion of a monetary attribute (i.e. price) allows for the derivation of the marginal rates of substitution of one attribute relative to another, along with permitting a variety of elasticity and willingness to pay calculations.

In order to undertake the stated choice survey a Bayesian efficient experimental design was generated. Each respondent was presented with 12 choice tasks, resulting in a total of 3,000 observations, and each choice was made between the four book format alternatives that are available on the market. A selection of socio-demographic information was collected from each respondent, along with questions relating to their actual reading habits, to be utilised in the final empirical estimation. Analysis of the choice experiment is still underway, therefore results and policy implications are yet to be determined at this stage. The data will be modelled under the multinomial logit framework. A range of model specifications will be examined in order to identify non-linearity and interaction effects.

All told, the results of the stated choice survey should yield a myriad of insights into how consumers arrive at their book consumption decisions. In light of the debate on the future book industry outlined briefly above, of particular interest are questions such as: Is the impact of ebooks on the existing market for books bounded by consumer preferences in any way? That is to say, can one expect the sales of ebooks to completely cannibalise those of physical books, or are there particular attributes of physical books that consumer’s value, which will prevent this from happening? This research will also help to shed light on our understanding of consumer choice for other cultural experience goods, such as music, movies, theatre and the like.

Keywords: Discrete choice models, Cultural consumption, Willingness to pay, Book Industry

[1] Wischenbart, R. (2013). The Global eBook Market: Current Conditions & Future Projections, Sebastopol CA: O’Reilly Media Inc.

[2] Carreiro, E. (2010). “Electronic Books: How Digital Devices and Supplementary New Technologies are Changing the Face of the Publishing Industry”, Publishing Research Quarterly, 26(4), 219-235.

[3] The genre attribute is developed in conjunction with earlier work conducted by the author that examines social consensus concerning genre-level evaluations of cultural content.

[4] Lévy-Garboua, L. and C. Montmarquette (1996). “A Microeconometric Study of Theatre Demand”, Journal of Cultural Economics, 20, 25-50.

Conference registration is required in order to view papers.