The effects of one-way compatibility on technology adoption in systems markets

Abstract

This paper analyzes the effects of one-way compatibility on technology adoption in a market that is characterized by a free-entry condition on the software side. We show that hardware-backward compatibility increases a new-generation hardware firm’s profit; the effects of software-backward compatibility on the hardware firm’s profits depend on the distribution of consumer types.

Introduction

When a new system consisting of hardware and software is introduced into a market,¹ one important decision is whether the new system will be compatible with an existing system. The compatibility choice determines whether the new hardware technology can facilitate old-generation software (hardware-backward compatible), and whether a new-generation software program can be used with the old-generation hardware technology (software-backward compatibility). The compatibility choice decisively determines the adoption of the new system.

There are several papers on the effects of compatibility. Many of them analyze a firm’s choice between two-way compatibility and non-compatibility (see Farrell and Saloner, 1992, Choi, 1996, among others.) However, the effects of one-way compatibility have been relatively ignored because the conventional wisdom is that having only hardware-backward compatibility, not software-backward compatibility, is beneficial for a firm with new-generation hardware. More specifically, an entity introducing a new hardware platform can make it more attractive by allowing its hardware to use software products designed for an old-generation platform. Also, if new-generation software products cannot be played on an old-generation hardware platform, it makes the old-generation hardware obsolete, which is good for the new-generation hardware firm. For instance, in Farrell and Saloner, 1992, Choi, 1996, Ellison and Fudenberg, 2000, the benefits from a new technology are assumed to increase monotonically with the size of the consumer group owning compatible technologies. In their settings, it is always a weakly dominant strategy for a firm to offer a one-way converter with its product, as doing so derives some positive network effects from the other system, but prevents the other system from deriving any network effects.

If the number of software programs is fixed, the validity of this argument is obvious. However, as Chou and Shy, 1990, Church and Gandal, 1992 show, when the software side is characterized by a free-entry condition, the variety and prices of software programs are endogenously determined by consumers’ spending on software products. Then, the effects of one-way compatibility could be quite different from the conventional wisdom. For motivation, we offer one example.

In 1950, the FCC voted to adopt CBS’s color system as the industry standard. In the CBS system, color TV signals could not be received by Black and White (B/W) TV.² If the number of programs had been fixed, it would have been optimal in terms of color TV adoption. However, a majority of households had B/W TV sets, which could not receive the color TV signal. Since broadcasting companies wanted to send their programs to a majority of households, they kept airing major programs in B/W. This resulted in a limited number of programs in color. Thus, consumers did not have an incentive to upgrade to color TV sets. Finally, in 1953, the FCC decided to change the color TV signal so that color signal could be received by the existing B/W TV sets without any additional equipment.³ After the change, broadcasting companies could send their color programs to both households with B/W TV sets and with color TV sets, giving them much better incentives to develop color TV programming. As a result, in contrast with the conventional wisdom, software-backward compatibility increased the number of color TV programs and accelerated the adoption of the new-generation hardware platform.

As the case above shows, compatibility decisions have two effects on the benefits of hardware products. First, compatibility choices directly affect the benefits of hardware products by determining what kind of software products hardware technologies can use. Second, since software prices and varieties depend on consumers’ spending on software, compatibility choices, by changing consumers’ spending on software products, bring in a different set of software prices and variety. For instance, under hardware-backward compatibility, new-generation hardware can use both old-generation and new-generation software, which directly increases the benefits of the new-generation hardware. However, since consumers with new-generation hardware can use old-generation software products, it might result in fewer varieties of new-generation software products, which, in turn, might reduce the value of new-generation hardware.

In order to analyze the overall effect, we need to combine these two effects. This paper builds a vertically-differentiated products model with heterogeneous consumer preferences and analyzes how consumers and software companies react to these compatibility choices when software prices and varieties are endogenously determined.

There is a commonly held belief that hardware-backward compatibility increases consumer surplus from new-generation hardware. We partially confirm the validity of this belief by showing that hardware-backward compatibility always increases the marginal type’s surplus from new-generation hardware.⁴ Thus, hardware-backward compatibility increases the new-generation hardware firm’s profitability. However, we can have a case in which hardware-backward compatibility may reduce higher types’ consumer surplus. Thus, hardware-backward compatibility does not benefit all consumers with new-generation hardware and involves a welfare trade-off among consumers with the new-generation hardware. In addition, we show that the effect of software-backward compatibility on new-generation hardware depends on the distribution of consumer types. Software-backward compatibility increases the new-generation hardware firm’s profits only if the ratio of high-type consumers who differentiate the quality difference between new and old systems a lot is low.

This article relates to several lines of research.⁵ In the hardware/software structure, Church and Gandal (2000) analyze compatibility issues in the context of foreclosure by fixing the number of software programs. In an interesting article, Chou and Shy (1993) investigate how (exogenously given) partial compatibility affects software provision and hardware market shares. However, there are important differences between their models and our paper. In their Hotelling-type models, there is no quality difference between the two hardware platforms. When there is no quality difference, if any platform can use software products designed for the other platform fully, then software products are supplied at the potentially lowest price, and the highest consumer surplus is realized no matter what the hardware market shares are.⁶ However, by allowing for a quality difference between old- and new-generation systems and consumers’ heterogeneity in the value placed on this quality difference, we provide different results. For instance, under hardware-backward compatibility, old- and new-generation software products coexist, and hardware-backward compatibility does not benefit all consumers with new-generation hardware.

Ellison and Fudenberg (2000) analyze a monopolist’s incentive to introduce a new technology when that technology is characterized by hardware-backward compatibility. In Ellison and Fudenberg (2000), benefits from a technology increase monotonically with the size of the consumer group owning compatible technologies. Thus, in their setting, the optimal compatibility choice in terms of new technology adoption is always to support hardware-backward compatibility, not software-backward compatibility.

The paper is organized as follows. Section 2 presents the formal model. In Section 3, we explore the effects of one-way compatibility on consumer welfare and monopoly profits. Concluding remarks follow these analyses.