Innovation, imitation, and growth with cumulative technology

Abstract

This paper analyzes the interaction of innovation and imitation in the growth process. Technology is assumed to be cumulative: only leaders can conduct next-round innovation. Outsiders can become leaders by imitation. Our results show that subsidizing imitation may increase the economy-wide rate of technological progress. There are cases where competition and growth exhibit positive correlation. In these cases, promoting imitation enhances not only the static efficiency but also the dynamic performance of the economy.

Introduction

It has long been known that a wave of imitative activity follows a creative innovation. In The Theory of Economic Development, Schumpeter wrote:

…if anyone has in him all that pertains to success … then he (the innovator) can … make a profit which remains in his pocket. But he has also triumphed for others, blazed the trail and created a model for them which they can copy. They can and will follow him, first individuals and then whole crowds. (J.A. Schumpeter, The Theory of Economic Development, Harvard University Press, Cambridge, 1934, p. 133.)

In their influential empirical study, Mansfield et al. (1981) found that 34 new products out of 48 samples had been imitated during the sample period.¹

Although the effect of imitation is known to be substantial in the determination of the level of technology and economic growth, it has not been analyzed enough in the growth literature.² This paper explicitly discusses the implication of imitative activity in growth theory.

In standard models of R&D-based growth, the existence of imitation and resulting competition are always detrimental to an economy's growth performance. In the growth literature, it has been argued that anticipated competition reduces the incentive to engage in innovative activity by reducing the reward from innovation.³ We call this the negative-incentive prediction from imitation-induced competition to growth.

This negative-incentive prediction imposes a difficulty on policymakers. While the dynamic performance of an economy is enhanced by protecting the monopoly power of innovators, static efficiency is sacrificed.⁴

The negative-incentive prediction also presents an empirical puzzle to researchers. There are no strong empirical support for this prediction in the large literature which studies innovation and competition.⁵ Typically the results are mixed: competition seems to enhance innovation in some cases, while it discourages innovation in other cases. Among these studies, recent evidence by Nickell (1996) and Blundell et al. (1995) shows a positive correlation between innovation and competition, which led Aghion and Howitt (1997) to construct a series of growth models that have this feature.

In this paper, we propose an alternative explanation to solve this puzzle and show that it is the structure of knowledge spillover that produces the negative-incentive prediction. In the model, a more realistic process of knowledge diffusion is proposed. The features of the resulting equilibrium are examined.

Another empirically puzzling prediction of standard theories is the leapfrogging prediction. Due to the well-known Arrow effect (Arrow, 1962), incumbents always have less incentive to conduct innovative activities than outsiders do. Then innovations will always be performed by outsiders and the current industry leaders will be replaced.

Nevertheless, it is also difficult to reconcile this prediction with real-life fact. In reality, innovative activities are often carried out by incumbents in industries.⁶ To fill this gap between theory and reality, Barro and Sala-i-Martin (1995, pp. 254–259) constructed a model where leaders have cost advantages in innovation. The result, however, leads to another counterfactual prediction. Leaders conduct all the research and stay as leaders forever.⁷

In our model, leaders have advantages in knowledge, but there is always a possibility that outsiders learn and enter the industry. In equilibrium, the previous monopolist conducts an innovation race with a newcomer who has successfully learned about the state-of-the-art technology. Both have an equal chance to succeed in innovation and become a future monopolist.⁸

In the growth literature, the fact that imitation has an aspect of knowledge diffusion has long been ignored. In the standard models of Romer (1990), Grossman and Helpman (1991), and Aghion and Howitt (1992), knowledge is assumed to be perfectly shared by all people in the economy, even though each good is produced by only one firm. This assumption is counterfactual. As Jovanovic (1997) estimates, the learning cost of new knowledge is huge.

Imitative activity has long been identified as a learning activity. Many firms start up by imitating, and they often develop new technology based on the knowledge they learned from others. Toyota, who entered the automobile industry in 1930s, first learned the Ford production system, and then improved upon it to create a more efficient production system called the “lean production system”.⁹ Some of its main features were later imitated by American firms (Freeman and Soete, 1997, Chapter 6). The first business computer was called UNIVAC, built in 1951 by John Eckert and John Mauchly. It is known that in its early years IBM tended to follow a strategy of rapid imitation, learning the most important technological knowledge from universities and competitors (Freeman and Soete, 1997, Chapter 7). Although IBM's first data-processing computer (the model 702) was troublesome and unreliable, an improved version (the model 705) was soon introduced in 1954. By 1955, IBM was selling more computers than UNIVAC (Campbell-Kelly and Aspray, 1996). The first graphical operating system was introduced by Apple in 1984. Microsoft entered the graphical operating system market in 1985 with Windows 1.0. Windows 1.0 and the subsequent Windows 2.0 did not sell well. Microsoft finally gained a huge success with the introduction of Windows 3.0 in late 1990 (Liebowitz and Margolis, 1999).

There are also many episodes of firm “spin-outs”: entrant firms are often started by former employees of incumbent firms.¹⁰ Intel was founded in 1968 by two engineers who left Fairchild, one of the largest semiconductor corporations in the Silicon Valley.¹¹ Intel discovered the microprocessor in 1971, and has since grown up to be one of the highest-valued companies today. Franco and Filson (2001) present some examples in the rigid disk drive industry where spin-outs imitate the technical know-how of their former employers.¹² Christensen (1993) shows that in the rigid disk drive industry, spin-outs were very important source of entry. There were 39 spin-out entries during 1976–1989, and spin-outs accounted for all but four of the start-ups that actually generated revenue. Bhide (2000) interviews 100 founders of the 1989 Inc. 500 companies,¹³ and concludes “The Inc. founders we interviewed typically imitated someone else's ideas that they often encountered in the course of a previous job.” These studies suggest that knowledge diffusion by imitation is important in a wide variety of industries.

In an influential paper, Scotchmer (1991) argues that when technology is cumulative (second-round innovations are based on first-round innovations), the risk of patent infringement affects the incentives of second-round innovators. Her focus is mainly on patent policy; from this perspective, she claims that a broad patent protection may not lead to efficient innovation.

Merges and Nelson (1994) also emphasize the cumulative nature of technology. They state: “By a cumulative technology, we mean one in which today's advances lay the basis for tomorrow's, which in turn lay the basis for a next round, and so on, with the sequence often progressing very far from the original invention starting place.” They argue that this cumulative aspect may be more important than the conventional trade-off between the incentive to innovate and monopoly distortion. Strong patent protection may discourage next round innovators. After describing some historical cases, they conclude:

… we believe that the granting and enforcing of broad pioneer patents is dangerous social policy. It can, and has, hurt in a number of ways. It has made the entry of creative and energetic newcomers difficult. … there are many cases in which technical advance has been very rapid under a regime where intellectual property rights were weak or not strongly enforced.

In this paper, following the spirit of Scotchmer (1991) and Merges and Nelson (1994), we introduce an assumption that technology is cumulative in the sense that only leaders (who are “on the shoulders of giants”) can conduct innovative activities. An outsider has to learn the already accumulated technology by imitation before inventing a new technology. We believe that this is a more realistic setup than the standard one.

Our model is mainly based on Segerstrom (1991), who developed a tractable model with innovation and imitation. Although many of his results are qualitatively similar to ours, the current model differs from his in some nontrivial ways. First, his model is silent about the competition issue, since monopoly pricing always prevails in his model: a successful imitator colludes with the incumbent and they enjoy the monopoly profit together.¹⁴ In contrast, we assume Bertrand competition between a newcomer and an incumbent, so that the competitive pricing prevails. Firms cannot obtain any monopoly profits after the entry of an imitator. Second, Segerstrom's model shares the same counterfactual knowledge structure with the standard theories. In his model, outsiders can innovate a new quality of product at the same cost as incumbents. The consequence of this assumption is the leapfrogging result. A new innovator is always a firm other than two leaders (the previous monopolist and the successful imitator) who are already in the industry. In contrast, in our model, there is a possibility that the previous monopolist succeeds in innovation and remains in the industry.

Independent work by Aghion et al. (2001) shares a similar view with ours: imitation and competition may enhance growth. They assume duopoly with no entry, and imitation is conducted by a lagged incumbent. Many of their results parallel ours, but their setting is more complicated. They do not conduct policy analyses.

This paper is organized as follows. In Section 2, we construct a dynamic general equilibrium model with innovation and imitation. In Section 3, we discuss the implications of the model for market structure and growth. Section 4 concludes.