A Comparison of Bundling and Sequential Pricing in Competitive Markets: Experimental Evidence

Abstract

Technological advances enable sellers to identify relationships among offered goods. Sellers can leverage this information through pricing strategies such as bundling and sequential pricing. While these strategies have primarily been studied under monopoly assumptions, the strategies are available to competitive firms as well. This paper reports on a series of laboratory experiments comparing bundling and sequential pricing while varying the underlying relationship between the goods in markets where a fraction of buyers comparison shop. The results indicate that sequential pricing is generally as profitable to the seller; however, there is evidence that sequential pricing may be more harmful to consumers than bundling when the goods have complementary values or the buyer’s values are positively correlated.

Key Words:

• Pricing Strategy
• Bundling
• Sequential Pricing
• Experimental Methods

JEL classifications:

• C9
• D4
• L1

Notes

1. Sellers can also connect different baskets to the same customer through “rewards” or “preferred customer” programs or through credit card information. Online retailers have additional methods including the use of cookies to track customers. Cookies also enable sellers to identify comparison shoppers (see Deck and Wilson, 2006, for a discussion of the pricing implications for this type of tracking).

2. The term “pure bundling” refers to the case where the seller only offers the goods in a bundle.

3. McAfee et al. (1989) extend their monopoly results to a duopoly. Chen (1997) analyzes a situation in which firms compete in a duopoly for a single product and the firms also produce other products under conditions of perfect competition. Bakos and Brynjolfsson (2000) show that bundling low marginal cost goods in competitive environments can produce economies of aggregation, even without economies of scale or network externalities.

4. RFID technology is currently being used to improve the efficiency of the supply chain by tracking pallets as they move through the distribution process. Retailers including Sam’s Club and Walmart are pursuing individual item RFID tagging (Bustillo, 2010; Weier, 2008). The technology is touted as a means of avoiding stock outs and reducing theft. This technology, in conjunction with smart shopping carts, will also enable bricks and mortar stores to offer the same types of recommendations and detailed product information that is available online. United States Patent 5729697 is for just such a smart shopping cart. This technology could move prices from the physical shelf to the shopper’s private monitor.

5. There are three main arguments as to why firms would bundle, relying on exploitation of market power. One reason is as a form of price discrimination. Another is as an entry-deterrent strategy (see Nalebuff, 2004). A third is that a firm leverages market power in one market to capture a greater share of a more competitive market (see Caliskan et al., 2007, for a discussion of this literature as well as an experimental test of such models). In this paper, we are focusing exclusively on the first reason.

6. From a marketing perspective, Mulhern and Leone (1991) analyze retail pricing and promotion strategies, while Venkatesh and Mahajan (1997) examine strategies for marketing products using their branded components. From an information systems perspective, Bakos and Brynjolfsson (1999) and Hitt and Chen (2005) study strategies for pricing a large number of information goods. Netessine et al. (2006) present a stochastic dynamic program that analyzes the problem of dynamically selecting complementary products, as well as the problem of pricing products to maximize profits.

7. Buyer behavior must be individually rational so the buyer has the option to buy nothing.

8. Perhaps travel or time costs make visiting multiple stores prohibitive or slight differentiation makes products from different sellers incompatible.

9. The correlations in the baseline and complements conditions are both zero.

10. The distributions used in the laboratory are consistent with those used in the numerical monopoly analysis of Aloysius et al. (2009). Chen and Riordan (2008) use a similar construction for analyzing monopoly and oligopoly behavior over a uniform distribution over two dimensions with correlation. In their problem, each firm sells a single good.

11. Subjects did not know the total number of periods in the experiment nor were the practice periods singled out for the subjects. Subjects did know the total length of the session, but typically the experiments finished well before the allotted time had expired.

12. See Deck and Wilson (2006, 2008) for other studies using nearly continuous posted offer experiments.

13. A similar tool for the comparison shoppers would require that subjects specify price vectors for each of their rivals, a task that was viewed as being overly cumbersome. It is possible that this design tool led subjects to be overly focused on loyal customers; however, the results suggest that this is not the case.

14. Copies of the directions and the handout are available from the authors upon request.

15. 500 periods per subject × 144 subjects = 72,000 market decision periods. The last 500 periods of each session are used in the statistical analysis to control for learning effects. Most sessions experience prices falling over the first 200 periods, which are excluded from the analysis. After this point, prices tend to relatively flat. If one considers only the last 250 periods, the results are qualitatively the same as those reported below.

16. Efficiency is measured as the percentage of the total possible gains from trade that were actually realized. In these markets, a trade is fully efficient is the buyer purchases both goods A and B, since the values are non-negative and the costs of production are 0.

17. The average difference in the single item prices was less than 1 × 10 − 10 in all three conditions and not statistically different from 0 in any condition.

18. See Hess and Gerstner (1987) for a discussion of this strategy.

19. With independent goods, distribution of the marginal values of good B is the same regardless of whether or not the seller attracts comparison shoppers. For this special case, P _B is the same under monopoly and competition.

20. Given that comparison shoppers only observe good B prices if they purchase good A, the distribution of good B values that a seller faces depends on the posted good A price. The reported p-value is for testing P _B = 50.

21. The p-values reported in Table 5 are for the two-sided test on inequality, but here we have a one-sided alternative hypothesis. While the buyers who were identified as having purchased good A did not receive a statistically different price from buyers in the no discrimination treatment, their prices were nominally lower making them similar to those who were identified as not having purchased good A.

22. The comparison between bundling and sequential pricing without discrimination would be significant in the one-sided test that sequential pricing is harmful.

23. In the cases of positively correlated goods and complements, sequential pricing without discrimination under competition results in a different PB than under monopoly. The optimal price depends upon the probability of the seller setting the lowest P _A and thus being visited by the high good A valued comparison shoppers. In both cases, the optimal P _B is weakly higher than under monopoly, since the distribution of VBs stochastically dominates the distribution faced by a monopolist. Therefore, the monopoly P _B serves as lower bound for P _Bs that could be in the mixing distribution, and thus in these two cases, the observations that cannot be considered inconsistent with the symmetric Nash equilibrium mixing distribution form an area in the two-dimensional price space.