Schumpeter's legacy: A new perspective on the relationship between firm size and R&D

Abstract

This paper shows that firm heterogeneity in technological competence, rather than differences in industry-specific characteristics, is the primary condition determining the long-debated relationship between firm size and R&D. Specifically, by utilizing a formal model of firm R&D that shows that profit-maximizing firm R&D intensity is determined jointly by firm-specific technological competence and consumer preference regarding quality and price, this paper suggests that firm size affects firm R&D intensity not directly, but through its influence on firm-specific technological competence. In particular, four predictions are drawn and tested empirically: (1) in general, the size–R&D relationship is less-than-proportional or inverted U-shaped, especially for low-technological-competence firms; (2) however, the common less-than-proportional relationship disappears, and a more-than-proportional relationship becomes increasingly likely, for firms with high levels of technological competence, plausibly due to competence-enhancing, learning economies of scale and/or scope in R&D; (3) firms with larger accumulated R&D experience are, ceteris paribus, less likely to exhibit the common less-than-proportional relationship; (4) among industries, a greater within-industry departure from the proportional size–R&D relationship is expected for industries with seemingly high, rapidly changing technological-opportunity conditions. These predictions, especially pertaining to the conditioning role of technological competence in the size–R&D relationship, are empirically supported by the unique data by the World Bank.

Introduction

The relationship between firm size and R&D expenditure has drawn considerable attention from researchers for over three decades. As a result, a large body of empirical studies has been accumulated with some stylized facts regarding the size–R&D relationship, including a positive, more-or-less proportional relationship (e.g., Cohen, 1995, Cohen and Klepper, 1996).¹ However, as Cohen and Levin (1989), Symeonidis (1996), Van Dijk et al. (1997), and recently Mazzucato (2000, p. 16) pointed out, the most notable feature of this considerable body of empirical research on the relationship between firm size and R&D is its inconclusiveness. That is, the literature provides considerably diverse and often-conflicting results: Many studies found linear, positive relationships, but others found either non-linear or statistically insignificant relationships, or suggested that firm size and R&D are independent of each other (e.g., Klette and Griliches, 2000).² Thus, as Scherer (1992) indicates, even though the bulk of existing studies goes against the so-called Schumpeterian hypothesis of the advantages of large corporations in R&D, questions on diverse results still remain unanswered.

As Cohen and Levin (1989) pointed out, the inconclusive studies on the size–R&D relationship may have been afflicted with some empirical problems such as sample selection biases, the likely collinearity between firm characteristics and firm size, and the effect of innovation on firm growth. However, as emphasized by Arvanitis (1997), the diverse and often conflicting empirical findings seem to be largely attributable to the lack of appropriate formal theories or models of firm R&D that suggest the determinants of firm R&D and their relationship to firm size,³ as this lack leads to arbitrary empirical specifications that are vulnerable to omitted variable biases and prone to diverse and often conflicting results.⁴

The purpose of this paper is to derive a formal model of firm R&D that provides theoretical predictions regarding the size–R&D relationship, and to test them empirically, thereby offering new insight into the long-debated size–R&D relationship. Particularly, in line with a suggestion of Acs and Audretsch (1987, p. 573), this paper asks the question of under which circumstances large or small firms have the relative innovative advantage, rather than the conventional question of which firm size is most conducive to innovation. Specifically, based on the firm R&D model, this paper draws conditions for the likelihood and direction of departure from the benchmark case of the proportional relationship between firm size and R&D expenditure.⁵

The formal model of firm R&D employed in this study shows that firm R&D intensity is determined jointly by firm-specific technological competence and consumer preference regarding quality and price, thereby implying that firm size can affect firm R&D intensity not directly but indirectly through its influence on the two determinants of R&D, especially on firm-specific technological competence.⁶ Hence, firms with high absorptive and learning capabilities are more (less) likely to show a more-than-proportional (less-than-proportional) size–R&D relationship, since they are better able to utilize various advantages of large firm size in enhancing their technological competence as a result of, for example, learning economies of scale and scope in R&D and in creating better technological opportunities.⁷ Since the concept of technological competence encompasses both the absorptive and learning capabilities, it is also predicted that the relationship between firm size and R&D is conditioned by the level of firm-specific technological competence.

The conditioning role of technological competence in the size–R&D relationship is supported by an empirical analysis of a data set, compiled by the World Bank, of about 1400 R&D-performing firms in nine manufacturing industries across six countries. Considering the argument raised by Cohen and Levin (1989) and Symeonidis (1996) that the literature on the size–R&D relationship has paid little attention to the need to control for firm-specific characteristics other than firm size, this study contributes to the existing literature by identifying the conditioning role of firm-specific technological competence in the size–R&D relationship and by suggesting the mechanism through which size confers advantages in R&D.⁸

This paper is organized as follows. Section 2 derives a model of firm R&D that shows the determinants of profit-maximizing firm R&D expenditure and R&D intensity. Section 3 draws from the firm R&D model four theoretical predictions on the relationship between firm size and R&D, including those on the likelihood and direction of departure from the benchmark case of the proportional size–R&D relationship. Section 4 empirically tests the theoretical predictions, especially the conditioning role of technological competence in the size–R&D relationship, and Section 5 concludes this study with some policy implications.