Introduction

This chapter investigates some statistical properties of the growth patterns of a large sample of pharmaceutical companies representing the top incumbents worldwide. We articulate our analysis along three complementary directions, which, with an evident lack of precision but, we hope, with a certain degree of suggestiveness, we denote as “temporal,” “cross-sectional,” and “disaggregated.”

With the “temporal” direction of investigation we refer to statistical analyses of the size of the firm in its time evolution – i.e. the study of the time series of firm sizes and growth rates. The point of departure of this kind of analysis is usually a question à la Gibrat, addressing issues concerning the relationships between firm size and its dynamics. The Gibrat benchmark (Gibrat, 1931), also known as the “law of proportionate effect,” postulates that the growth of business firms is a random walk that ultimately yields an asymptotic log-normal size distribution. The evidence concerning these properties, as explored by a rich and growing literature (see, for instance, Dunne, Roberts, and Samuelson, 1988; Evans, 1987; Hall, 1987; Hart and Prais, 1956; and Mansfield, 1962; and, more recently, Amaral et al., 1997; Lotti, Santarelli, and Vivarelli, 2001; and De Fabritiis, Pammolli, and Riccaboni, 2003), roughly supports two main findings: Gibrat's hypothesis is confirmed by the lack of any relationship between the (log) size of firms and their average growth rates, but is violated by a clear negative dependence of the growth rates' variance on size (see Sutton, 1997, for a review).