International Links of Innovation Patterns

The traditional theory of economic growth has concentrated on capital accumulation, exogenous population growth and exogenous technical progress. Indeed, capital accumulation and labor growth proved to be important components in the explanation of growth patterns, but capital accumulation and labor growth alone explain only part of observed growth rates. The rest has been typically attributed to technical progress (see Solow (1957) for the original contribution and Maddison (1977) for a recent discussion of growth accounting).

It has been clear, however, from the very beginning that exogenous technical progress cannot be a satisfactory working hypothesis. Major productivity gains require a deliberate effort of invention an innovation, especially in modern times (see, for example, Freeman (1982)). The latter implies that resources need to be devoted to these activities. In order for these resources to be forthcoming, however, the economic system has to properly reward inventors and innovators, thereby providing incentives to engage in these activities. All this is rather self evident, but direct measurement of the contribution of inventions and innovations to economic growth proved to be very difficult (see Griliches (1979)). Part of the difficulty results from the lack of satisfactory data for the problem at hand. But another part of the difficulty results from the lack of a satisfactory theory.

Following the general slowdown of productivity growth in the 1970s the interest in economic growth has been renewed. This has been manifested in the publication of numerous empirical studies that attempt to explain the events of the seventies. More recently, however, the theory of economic growth has also been modified in order to deal with the new concerns. The new approaches emphasize factors that lead to sustained long run growth at rates that are endogenously determined. This constitutes an important departure from traditional theory in which long run growth was exogenously determined. The new studies have proceeded along several lines:

(i) Endogenous population growth (Becker and Murphy(1988).
(ii) The role of public services (Barro (1989)).
(iii) Accumulation of human capital (Ohyama (1989).
(iv) Learning by doing and related spillovers (Romer(1986, 1988) and Lucas(1988)).

Although learning by doing has been explored in the earlier literature (see Arrow (1962) for the original contribution as well as Uzawa (1965), Levhari (1966) and Sheshinski (1967)), its recent combination with an explicit treatment of product innovation -- in the form of development of new products or the improvement of existing products -- has yielded important new insights on endogenous technical progress (see, for example, Romer(1988) and Aghion and Howitt (1989)).

This study is concerned with international links of innovation patterns, where innovation drives economic growth. Such links have been explored in a series of papers by Grossman and Helpman (1989a-e) for long run steady states. (Grossman (1989) has also used it to explain Japan's recent performance.) They showed in a variety of models that build on endogenous product innovation how the long run growth rate of a country depends on its own features, features of its trading partners, its own trade policy, trade policy of its trading partners, its own policy towards innovation and imitation and its trading partners' policies towards innovation and imitation. These links proved to be rather involved. For example, a less developed country that encourages imitation of products that have been originally developed in an advanced industrial country may thereby speed up growth around the globe.

In what follows I study the time pattern of endogenous innovation, including out of steady state dynamics. For this purpose I employ a simplified version of a model developed in Grossman and Helpman (1989a). In that model the growth rate is closely associated with the rate of innovation. Two major questions will be addressed:

(i) How do the time patterns of innovation and growth respond to a change in the available resources?
(ii) How do the time patterns of innovation and growth respond to R&D subsidies?

The first question addresses a structural issue while the second addresses a policy problem.

Some interesting results emerge from this analysis. For example, the short run response of the rate of innovation to an R&D subsidy can differ from the long run response, and short run rates of innovation can overshoot or undershoot long run innovation rates. The answers to the policy questions depend on structural parameters as well as on whether the policy active country has a comparative advantage in research and development. A summery of the results is provided in the closing section.

In order to make the paper self contained I present in the next section a simple one country model in which profit seeking entrepreneurs develop new products. An expansion of the menu of available products raises productivity in manufacturing via the refinement of specialization (as in Ethier (1982)) or it raises productivity in consumption via an increase in the available product choice. For this reason a larger product choice is desirable per se. In addition current product development reduces costs of future product innovation. This feature captures the idea that even when targeted at particular products R&D also generates broader knowledge that can be applied to other products. It thereby leads to the accumulation of non-appropriable knowledge capital. Knowledge capital is particularly useful for R&D. Thus, current innovative activities reduce costs of future innovations. This specification leads to endogenous innovation and growth, with their rates depending on available resources and on the government's industrial policy (as in Romer (1988)) . In the one-country case the rates of innovation and growth settle down immediately on steady state levels. Therefore in this case there exist no out of steady state dynamics.

In Section III the model is extended to represent a world of two countries. With two countries in place there exist non-trivial out of steady state dynamics. In a stationary environment the rate of innovation may rise or decline over time (depending on initial conditions) until it converges to a steady state. In that section I also analyze the response of its time pattern to changes in available resources.

The model is further extended in Section IV in order to deal With industrial policy. I show how the patterns of innovation responds to R&D subsidies. The resulting changes in these time patterns depend on whether the policy is enacted by a country with comparative advantage or disadvantage in innovation. The paper closes with a summary of the main findings and some concluding comments.