Introduction

Endogenous growth theory formalizes the role of technical progress in explaining mod­ern economic growth. Although this is a relatively recent development, many of its ideas were already stressed by authors such as Kuznets, Griliches, Schmookler, Rosenberg and Schumpeter.

The first attempts to endogenize the rate of technical change addressed the first, but not the second, problem. Assuming technical progress to be an unintentional by-product of the introduction of new capital goods through a process named “learning-by-doing”, Arrow (1962) was able to generate sustained growth at a rate that depended on invest­ment decisions. Attempts at explicitly modeling investment in innovation faced another difficulty. A replication argument suggests that, for a given state of technology, produc­tionfunctions should exhibit constant returns to scale. If technical progress is considered as an additional input, however, the technology features increasing returns to scale and inputs cannot be paid their marginal product.

A tractable model of imperfect competition under general equilibrium was not avail­able until the analysis of monopolistic competition in consumption goods by Dixit and Stiglitz (1977), later extended to differentiated inputs in production by Ethier (1982). These models also showed how increasing returns could arise from an expansion in the number of varieties of producer and consumer goods, an idea that is at the core of the models studied in this chapter. The first dynamic models of economic growth with mo­nopolistic competition and innovation motivated by profits were built by Judd (1985) and Grossman and Helpman (1989). Yet, these authors were interested in aspects other than endogenous growth and none of their models featured long-run growth. Romer (1987), who formalized an old idea of Young (1928), was the first to show that models of monopolistic competition could generate long-run growth through the increased spe­cialization of labor across an increasing range of activities. The final step was taken in Romer (1990), which assumed that inventing new goods is a deliberate costly activity and that monopoly profits, granted to innovators by patents, motivate discoveries. Since then, the basic model of endogenous growth with an expanding variety of products has been extended in many directions.

The distinctive feature of the models discussed in this chapter is “horizontal inno­vation”: a discovery consists of the technical knowledge required to manufacture a new good that does not displace existing ones. Therefore, innovation takes the form of an expansion in the variety of available products. The underlying assumption is that the availability of more goods, either for final consumption or as intermediate inputs, raises the material well-being of people. This can occur through various channels. Con­sumers may value variety per se. For example, having a TV set and a Hi-Fi yields more utility than having two units of any one of them. Productivity in manufacturing may increase with the availability of a larger set of intermediate tools, such as hammers, trucks, computers and so on. Similarly, specialization of labor across an increasing vari­ety of activities, as in the celebrated Adam Smith example of the pin factory, can make aggregate production more efficient. The main alternative approach is to model inno­vation as quality improvements on a given array of products (“vertical innovation”), so that technical progress makes existing products obsolete. This process of “creative destruction” was emphasized by Schumpeter and has been formalized in Aghion and Howitt (1992), Grossman and Helpman (1991a) and Segerstrom, Anant and Dinopou- los (1990). The two approaches naturally complement each other. The main advantage of models with horizontal innovation lies in their analytical tractability, making them powerful tools for addressing a wide range of questions. However, because of their simplistic view on the interaction between innovators, these models are less suited to studying the effects of competition between “leaders” and “followers” on the growth process.

Section 1 of this chapter describes a simplified version of Romer (1990) and some extensions used in the literature. The model exhibits increasing returns to scale and steady-state endogenous growth in output per capita and the stock of knowledge.

Growth models with an expanding variety of products are a natural dynamic counter­part to trade models based on increasing returns and product differentiation. As such, they offer a simple framework for studying the effects of market integration on growth and other issues in dynamic trade theory. This is the subject of Section 2, which shows how trade integration can produce both static gains, by providing access to foreign varieties, and dynamic gains, by raising the rate at which new goods are introduced. Product-cycle trade and imitation are also considered.

In many instances, technical progress may be non-neutral towards different factors or sectors. This possibility is considered in Section 3, where biased technical change is incorporated in the basic growth model. By introducing several factors and sectors, the economic incentives to develop technologies complementing a specific factor, such as skilled workers, can be studied. These incentives critically depend on the defini­tion of property rights over the production of new ideas. The high variability in the effectiveness of patent laws across countries has important bearings on the form of technical progress. In particular, governments in less developed countries may have an incentive not to enforce intellectual property rights in order to speed up the process of technology adoption.

Section 4 introduces complementarity in innovation. While innovation has no effect on the profitability of existing intermediate firms in the benchmark model, in reality new technologies can substitute or complement existing technologies. Innovation may cause technological obsolescence of previous technologies, as emphasized by Schum­peterian models. In other cases, new technologies complement rather than substitute the old ones. For instance, the market for a particular technology tends to be small at the time of its introduction, but grows as new compatible applications are devel­oped. This complementarity in innovation can lead to multiple equilibria and poverty traps.

Complementarities in the growth process may also arise from financial markets, as suggested in Section 5. The progressive endogenous enrichment of asset markets, associated with the development of new intermediate industries, may improve the diver­sification opportunities available to investors. This, in turn, makes savers more prepared to invest in high-productivity risky industries, thereby fostering further industrial and financial development. As a result, countries at early stages of development go through periods of slow and highly volatile growth, eventually followed by a take-off with fi­nancial deepening and steady growth.

Finally, Section 6 shows how models with technological complementarities can gen­erate rich long-run dynamics, including endogenous fluctuations between periods of high and low growth. Cycles in innovation and growth can either be due to expectational indeterminacy, or the deterministic dynamics of two-sector models with an endogenous market structure.

2.