Home Page   Research  Teaching  Economics of IT   For Startups  Calculus

Index   Pricing   Network Effects   Digital Piracy   Telecommunications   Digital Convergence  
Pricing in information technology markets often seems strange. Products are actively given away free, the functionality and power of hardware devices grows rapidly as their prices remain constant or fall, and sellers of digitized products offer their customers unlimited usage, access or content. Some simple (yet unique) properties of digital products are responsible for the unusual pricing strategies we observe in IT industries; and while the properties are simple, the task of designing pricing can be extremely challenging.

The cost structure of information goods
Many of the unique pricing and market structure questions in IT industries stems from a specific unusual cost structure: high fixed costs of production, but near-zero or zero variable costs of production. This cost structure characterizes a class of technology products which are collectively termed information goods. Put differently, the cost of producing the first unit of an information good is very high, and yet the cost of producing each additional unit is virtually nothing. For instance, Microsoft spends hundreds of millions of dollars on developing each version of its Windows operating system. Once this first copy of the OS has been developed, however, it can be replicated costlessly.

Early examples of information goods were computer-based information services and software; currently, a wide variety of diverse products -- video, music, textbooks, research, digital art, long-distance telephone service, network bandwidth, to name a few -- share this unique cost structure. Since this cost structure is a fundamental property of digital information, as more and more previously physical or analog products become digital, this class of products is expanding; in addition, a number of other technology products comprise a bundle of a physical device and an information good.

Pricing strategies for information goods
A sound (and creative) pricing/price-discrimination strategy is critical for sellers of information goods, since they have to recover their high fixed costs while ensuring that competition does not force their prices close to variable costs. Moreover, the pricing strategies available to each seller can be a critical determinant of equilibrium market structure. A unique advantage that sellers in technology markets often have is that these variable costs are in fact almost zero, and this feature underlies many of the popular and successful pricing strategies:

  • Versioning: This is a form of second-degree price discrimination based on product quality, and is especially useful if degrading one's information good to create one or more lower quality versions is not expensive. Shapiro and Varian's 1998 HBR article provides an excellent description of the concepts and implementation of this strategy, highlighting a number of bases for differentiation. Jones and Mendelson (1998) have noted that versioning may not be optimal for certain kinds of utility functions; building on this, conditions under which versioning is optimal are analyzed in Bhargava and Choudhary 2002, and suggested indirectly in Sundararajan 2003.

  • Bundling: Many 'small' information goods (like individual items of web content, individual digital music streams) are sold in extremely large bundles. This is feasible largely because their variable costs of production are zero. In a pair of influential papers, Bakos and Brynjolfsson (1999, 2000) justify this strategy by showing that bundling information goods increases the revenue potential of each individual good. Bundling also increases the market power of a firm; it can serve as an entry-deterrent to high-quality providers of individual information goods, facilitate entry into adjoining markets, give a seller higher procurement power, and overall, further increase the likelihood of monopoly. Subsequent analyses have highlighted some cases where pure bundling may not be optimal: for certain categories of information goods (such as research articles), where customers place zero value on a large subset of the bundle, Chuang and Sirbu 2000 show that mixed bundling maybe better; the effects of customization of content (via categorization) on bundling strategy are described in Kephart and Fay 2000. Some other interesting strategic effects of bundling in general are provided by Barry Nalebuff's recent papers.

  • Fixed-fee pricing: Sellers of information goods often give customers unlimited usage for a flat (periodic) fee. Again, this pricing strategy is feasible because of their zero variable costs of production. This pricing strategy is always profit-improving in the presence of transaction costs, as established by Sundararajan 2003. Morevover, fixed-fee pricing should often be offered conjunction with second-degree price discrimination based on usage (though not always -- Varian 2000 provides a simple counter-example), and this pricing strategy complements both the versioning and bundling strategies described above. Some insight into the dynamics of competition between fixed-fee pricing and usage-based pricing can be obtained from Fishburn, Odlyzko and Siders 1997, who study conditions under which sequences of alternating profit-maximizing price changes (best-responses) converge to positive profit levels, and when they lead to prices of zero.

  • Other strategies: Electronic commerce increases the viability of using first-degree price discrimination for information goods, using dynamic pricing software and agent-based technologies. Some issues related to using this strategy are analyzed by Vulcan 1998, Ulph and Vulcan 2000, and Aron et al. 2001. Shapiro and Varian also recommend the use of third-degree price discrimination, a popular strategy for desktop software manufacturers. When there is either uncertainty or asymmetric information about the value of an information good (such as Internet-based advertising), Sundararajan 2003 shows that performance-based pricing can be profitable, and can also signal product quality.

    Rapid technological progress
    The technology that semiconductor devices (like computers and networking equipment) are based on evolves very rapidly. For microprocessors, one aspect of this progress is summarized by the well-known Moore's Law. Moreover, the fixed manufacturing costs of creating the 'fab' for each successive generation of microprocessors increases exponentially; however, the long-run variable costs per microprocessor do not change significantly. This leads to cycles of high initial prices for 'new generation' microprocessors, followed by steep price declines (especially just before a new generation is announced), and overall, relatively stable prices over time. According to this McKinsey Global report (see exhibit 5), average Intel microprocessor prices fluctuated between $210 and $230 between 1995 and 1999, even as their processing power increased several-fold. Similarly, the average price of a PC has fallen over the last 10 years, despite a tremendous increase in performance.

    It is likely that an essential first step towards explaining these pricing trends is to understand the underlying economics of general purpose technologies like the semiconductor. Helpman's book and Breshanan and Trajtenberg 1995 are good places to start; on a somewhat related note, Bresnahan and Greenstein 1999 provide a great analysis of the evolution of the computer industry. On the demand side, Mantena and Sundararajan 2002 suggest that continuously expanding product scope in conjunction with underlying technological progress can also lead to similar cycles.

    Some related links

  • Copyright © 2003 Arun Sundararajan.