Portal de Periódicos da CAPES

The Evolution of Firm Growth Dynamics in the US Pharmaceutical Industry

2010; Routledge; Volume: 44; Issue: 8 Linguagem: Inglês

10.1080/00343400903241469

1360-0591

Pelin Demirel, Mariana Mazzucato,

Innovation and Knowledge Management

Abstract Demirel P. and Mazzucato M. The evolution of firm growth dynamics in the US pharmaceutical industry, Regional Studies. This paper studies the dynamics of firm growth and firm size distribution in the pharmaceutical industry from 1950 to 2003. Growth dynamics are studied in the context of how the size composition of firms changes, how innovation patterns change, and how location leads to growth differentials among US firms. It is found that the growth advantage of small pharmaceutical firms increases after the 1980s as small firms become more active in patenting and their patenting activities become more ‘persistent’. Location is found to affect growth differences only for the most innovative firms (that is, for non-innovative firms, location does not matter). Demirel P. et Mazzucato M. L'évolution de la dynamique de la croissance des entreprises dans l'industrie pharmaceutique aux Etats-Unis, Regional Studies. Cet article cherche à étudier la dynamique de la croissance des entreprises et de la distribution des entreprises par taille dans l'industrie pharmaceutique de 1950 jusqu'à 2003. On étudie la dynamique de la croissance pour déterminer comment la distribution des entreprises par taille évolue, comment la structure de l'innovation change, et comment la localisation entraîne des écarts de la croissance des entreprises aux Etats-Unis. Il s'avère que l'avantage des petites entreprises pharmaceutiques en termes de croissance augmente après les années 1980 au fur et à mesure que les petites entreprises obtiennent davantage de brevets et que le brevetage ‘persiste’. Il s'avère que la localisation n'influe sur les écarts de croissance que pour les entreprises les plus innovatrices (c'est à dire que la localisation n'importe pas pour les entreprises qui ne sont pas innovatrices). Croissance des entreprises Innovation Industrie pharmaceutique Demirel P. und Mazzucato M. Die Evolution der Wachstumsdynamik von Firmen in der pharmazeutischen Industrie der USA, Regional Studies. In diesem Beitrag untersuchen wir die Dynamik von Firmenwachstum und Firmengrößenverteilung in der pharmazeutischen Industrie im Zeitraum von 1950 bis 2003. Die Wachstumsdynamik wird im Kontext der Frage untersucht, wie sich die Größenzusammensetzung der Firmen ändert, wie sich die Innovationsabläufe ändern und wie der Standort zu Wachstumsdifferentialen zwischen Firmen in den USA führt. Wir stellen fest, dass der Wachstumsvorteil kleiner pharmazeutischer Firmen nach den achtziger Jahren zunimmt, da die kleineren Firmen bei der Patentierung aktiver und ihre Patentierungsaktivitäten beständiger werden. Der Standort wirkt sich hingegen nur bei den innovativsten Firmen auf Wachstumsdifferenzen aus (mit anderen Worten, bei nicht innovativen Firmen kommt es nicht auf den Standort an). Firmenwachstum Innovation Pharmaindustrie Demirel P. y Mazzucato M. La evolución de las dinámicas de crecimiento de empresas en la industria farmacéutica de los Estados Unidos, Regional Studies. En este artículo analizamos las dinámicas del crecimiento de empresas y la distribución del tamaño de las empresas en la industria farmacéutica de 1950 a 2003. Estudiamos las dinámicas de crecimiento en cuanto a cómo cambia la composición del tamaño de las empresas, cómo cambian los modelos de innovación y cómo lleva la ubicación a diferenciales de crecimiento entre empresas estadounidenses. Observamos que la ventaja de crecimiento de pequeños laboratorios farmacéuticos aumenta después de la década de los ochenta cuando las pequeñas empresas se vuelven más activas con las patentes y las actividades relacionadas con patentes son más ‘persistentes’. Vemos que la ubicación afecta a las diferencias de crecimiento solamente en empresas más innovadoras (es decir, para las empresas no tan innovadoras la ubicación no importa). Crecimiento de empresas Innovación Industria farmacéutica Keywords: Firm growthInnovationPharmaceutical industryJEL classifications: L100L110 Acknowledgements The authors are grateful to the Economic and Social Research Council (ESRC) Innogen Centre for support (http://www.innogen.ac.uk). The research leading to these results received funding from the European Community's Seventh Framework Programme (Grant Number FP7/2007-2013) under Grant Agreement Number 217466. Notes The original data set which includes 323 pharmaceutical firms was cleaned to exclude the firms that report fewer than seven years of consecutive data. This ensures that there are at least five years of data available in the equations where a two-year lag of the firm size is used. Some firms have been subject to merger and acquisitions (M&A) during the period under observation. This creates a potential bias as it introduces an artificial ‘growth’ for those firms that merged with, or acquired, a small firm. It also overestimates the exit rates as the acquired and merged firms are counted as exits even though their economic activity has not completely ceased. Given this issue, there are two options for treating the data. The first is simply to leave the data in their raw form and ignore these events. In this case, the analysis is likely to suffer from the bias discussed above. The alternative method is Bottazzi et al.'s Citation(2001) methodology of forming ‘super-firms’ by restructuring the data set to adjust for M&As that have taken place between firms while under observation. The present authors decided to utilize the first of these solutions and interpret the results with the limitations in mind. However, an alternative data set of super-firms was also formed so as to check the robustness of the findings. The results proved robust irrespective of the data set used. The number of employees is averaged throughout the life span of a firm to determine its classification as a small or a large firm. This way, the problem of missing values in the employment data is tackled when firms are classified in one or the other size category. Patent data are taken from the National Bureau of Economic Research (NBER) patent database (Jaffe and Trajtenberg, Citation2002). The use of patent data in the paper is limited and mainly aims to describe the innovative activities of firms in a general way. Average length of the patent spells is 2.97 years. All firms in the data set are classified under Pharmaceutical Industry by Compustat. The distinction between the pharmaceutical and biotechnology industries is mainly based on the activities of firms in these two industries. While pharmaceutical firms specialize in the development and manufacturing of prescription and over-the-counter drugs, biotechnology firms research and develop biological substances for use in drugs and diagnostic tools. The authors' prior investigations confirm that biotechnology firms are structurally different from pharmaceutical firms in terms of size, age, and activities. The small pharmaceutical firms in the data set appear to be larger and older than biotechnology firms classified under Biotechnology Industry by Compustat. Some typical examples of ‘small persistent patentee firms’ from the database include Cima Labs, Inc., with more than twenty commercialized products and a proven record in successful inter-firm research alliances; Emisphere Technologies, Inc., the owner of an innovative drug-delivery platform called Eligen®, which provides a significant advance in delivery of insulin to improve the life quality of diabetics patients; Guilford Pharmaceuticals, Inc., which targets medium-to-large-scale under-serviced markets with innovative products to treat acute and chronic neurodegenerative disorders like Parkinson's disease; Acusphere, Inc., a specialty pharmaceutical firm that concentrates on improved efficacy and reduced costs with its Microsphere Techonology®; and Avanir Pharmaceuticals, Inc., whose lead product Abreva® made it to number one pharmacist-recommended cold sore treatment in the over-the-counter market within a short period of its launch. The non-United States category includes firms from Canada, the UK, Germany, France, Ireland, Austria, Switzerland, Israel, Sweden, Bermuda, Chile, Denmark, India, and Japan. Readers may find broader definitions of the NYC Tri-State Region ‘area’ that replace Pennsylvania or Long Island with Connecticut. Excluding the Pennsylvania from the NYC Tri-State Region definition does not significantly alter the results. Alternative results for Tables 1, 3, and 4 that include the Pennsylvania firms in the NYC Tri-State Region definition can be obtained from the authors upon request. Note that the coefficient in front of is now formulated as (β – 1) such that the left-hand-side variable is formulated as growth at time t. The limitation of using Model 1 is that it does not allow for individual firm-specific effects even though the (β 1 – 1) has a standard distribution and it is easy to test whether β 1 = 1. On the other hand, while Model 2 allows for the firm heterogeneity to be accounted with fixed firm effects, the β 2 coefficient has a downward bias in limited samples where T is small (Greene, Citation2003). The data set used for the regressions has a fifty-four-year span and the downward bias is especially significant when one looks at ten-year sub-periods of the data. However, as pointed out by Goddard et al. Citation(2002), this is a common issue for the panel data unit root tests (such as Augmented Dickey–Fuller (ADF)) and does ‘not present insurmountable obstacles’ to testing for Gibrat's law. The Kernel density estimate of a series X at a point x is: where K(.) is the Kernel density function; and N is the number of data points in the empirical distribution. The Kernel density function K(.) determines the shape of the bumps and can be chosen to be a function such as Epanechnikov, Gaussian (normal), and Uniform, etc. ‘h’ is called the ‘bandwidth’, which is the smoothing parameter. A larger bandwidth leads to a smoother curve. Note that the β-coefficient in Model 2 is also closer to 1 in the 1950–1960 period compared with other sub-periods. The firms that have persisted in BIMODE 1 are as follows: Abbott Laboratories, Astra Zeneca, Aventis SA-ADR, Bristol Myers Squibb, GlaxoSmithKline, Johnson & Johnson, Lilly (Eli) & Co., Merck & Co., Novartis, Novo-Nordisk, Pfizer, Inc., Pharmacia & Upjohn, Inc., Roche, SmithKline Beecham, Warner-Lambert Co., and Wyeth. The market share instability index introduced by Hymer and Pashigian Citation(1962) is calculated as: where sit is the market share of firm i at time t. Dosi Citation(2005) notes that structure in firm growth is indeed good news for evolutionary economists because it provides evidence that the persistent heterogeneity of firm characteristics and the competitive market selection among them is translated differentials in the growth performance of firms. Yet, it is difficult to speak about ‘structure’ without better understanding the underlying changes in production and technology and how these have impacted the dynamics between small and large firms.