Pricing Experience Goods in Information Good Markets: The Case of eBusiness Service Providers

Abstract

We study the pricing strategies of firms providing a service in experience good markets with switching costs. Using data on vendors providing “hosting and related services” at an early stage of the market, we test for pricing distortions that follow from oligopolistic competition with quality uncertainty and switching costs. We find that firms with a brand name charge a premium for their product – leveraging the reputation accumulated in closely related markets. As the theoretical literature suggests, we also find that the type of pricing distortions along the product line depends on consumers’ expectations about quality. If consumers underestimate the quality of the product, firms behave as if they discount introductory contracts in order to build trust, and later on markup upgraded contract. In contrast, firms that offer a quality level that is lower than consumers’ expectations markup initial contracts while discounting upgraded ones.

Key Words:

• Information Goods
• Pricing Strategies
• Electronic Business

JEL classifications:

• L81
• L15
• L11

Notes

1. The intuition behind their results is as follows. In an experience goods market, the seller is facing two different submarkets simultaneously: demand of those who already consumed the product and thus learned their preferences and demand from those who are uninformed. Since some consumers are more informed than others regarding the quality of the product, in mass markets the monopolist first skims the more attractive part of the market. This is in contrast to niche markets where the monopolist must offer low initial prices to capture a larger share of the uninformed consumers at the expense of targeting the more attractive informed segment of the market.

2. See, e.g., Table 6, U.S. Electronic Shopping and Mail Order Houses (NAICS 454110) for total sales in 2002 in E-Stats, http://www.census.gov/eos/www/papers/2002/2002finaltables.pdf.

3. As Greenstein (2000) shows, ISPs typically offer one or more of the following services: ISP services, Hosting services, Web-design services and Maintenance and support services. We focus on the group of providers that offer many of all four services, focusing specifically on whether they do offer maintenance and support. We capture this fourth service with the variable phonesupport, which we define below. While the group of Relation firms we study may still differ in their geographical locations, which might affect the type of consumers they face, we have studied this effect in a previous version of the paper and found that these differences do not affect Relation firms’ pricing behavior.

4. Note that there was almost no new entry into this market after the dot-com crash. As a result, we do not observe any firms who entered around the time of our survey. Most firms expended the vast majority of their entry costs before we observe them.

5. While the coefficient tells us about which designs generated higher or lower margins for these providers, it will not tell us whether the total incremental improvement in revenues from increasing quality over the next highest level exceeded the cost of designing it.

6. As illustrated by Intel’s high profile exit prior to our data collection.

7. This site, maintained by Meckler Media, provides ISPs the opportunity to advertise their services. The ISPs fill out a questionnaire where the answers are partially formatted; answers are then displayed in a way that allows users to compare different ISP services.

8. Since, in some cases, the pricing quotations advertised on thelist were inaccurate, we disregarded these quotes and used only the quotations advertised on the providers’ Web site.

9. These constitute 90% of all observations. The other 10% involve four other uncommon third-party carts.

10. One interpretation suggested to us was that this represented deliberate attempts at obfuscation by vendors – see, e.g., Ellison and Ellison (2005). Another was that this represented a simple marketing strategy to “frame” middle choices, making them appear comparatively more attractive by making the end choices appear to be less attractive. We are agnostic between these and other explanations. As elsewhere, our approach is to characterize this behavior and identify whether it facilitates higher or lower prices, then we discuss the range of interpretations the estimates allow for.

11. See, e.g., Lerner and Merges’ (1998) study of contracts between venture capitalists and biotech firms or Elfenbein and Lerner’s (2003) study of contracts between Internet portals and their online partners.

12. Much of this dates to Griliches (1961). For recent work, see, e.g., White et al. (2004) on prices for operating systems, Berndt et al. (1995) on prices for personal computer hardware, or Berndt and Rappaport (2001) on pricing of mobile computers.

13. We also estimated a random effect regression, which was superior to a fixed effect regression by standard tests; however, it does not add much over the OLS regression with clustered standard errors. Sometimes the coefficients or standard errors change slightly, but not by much or not in qualitatively important ways. For the sake of parsimony and space, we show only the OLS with clustered standard errors results.

14. Box-cox tests strongly favor the log price specification.

15. However, the cart vendors were reluctant to share information about their historical licensing practices with us, so we could not verify what fraction of this premium stayed with hosting firms who resold it.

16. The third column in Table 4 presents the discount/premium a Relation firm can charge due to quality.

17. Unfortunately, the small number of observations of Brand firms in our data set does not provide enough variation for us to study Brand firms’ pricing strategies along the product line.

18. The competitive segments – boxes – are based on the product–storage space introduced in Table 1.