Chapter Listing | Return to 1997 Topical Index


METHOD

Sample

    The sample for analysis is drawn from the new biotechnology firms. Biotechnology industry is widely recognized as a high–technology industry with a sizable number of cooperative relationships between entrepreneurial start–ups and pharmaceutical industry incumbents (see for example Powell et al., 1996). At first, biotechnology innovations were foreign to established pharmaceutical firms whose technological tradition was built around organic chemistry (Pisano, 1990). Incumbents thus had to rely on cooperative relationships with start–up companies to access the latest developments in biotechnology R&D, and the startups were able to benefit from the production and marketing resources of incumbents (Mang & Katila, 1996). Only later did both groups start developing their own internal capabilities in these fields. Characteristics of these incumbent?startup technology cooperation relationships are measured in this chapter.

    Cooperative arrangements studied in this paper include technological, production and marketing cooperative relationships, joint ventures, and licensing arrangements. We are particularly interested in the effects of technology cooperation, but other types of relationships were coded as well to be included as control variables.

    The study is restricted to biotechnology firms that managed to avoid the early threats of the liability of newness (Stinchcombe, 1965), and survived at least for seven years since their founding. To be included in the sample, the firm had to be founded between 1980 and 1988. We were unable to include firms that died in the early period, since they lacked necessary data to test our collaborative hypotheses. Furthermore, only biotechnology firms concentrating on human
 therapeutics and in–vivo diagnostics were included to hold the underlying technological setting and expertise required relatively constant. Our final research population included 103 new biotechnology companies. The majority of these companies were based in the US. On average, companies in our sample had three collaborative partners, although some had none.

    The data for this study were collected from several biotechnology–specific databases and directories, 10–Ks and annual reports of these companies. Multiple sources were used wherever possible to validate the data. For example, several independent sources, including PaineWebber and Genguide directories were checked to identify the relevant population of new biotechnology firms for the study.
 

Measures

    Dependent variables. The three dependent variables; total innovative output, the likelihood of radical innovation and organizational growth; are critical measures of performance for new high–technology companies. The first variable, innovative output is an important indicator of the performance and future opportunity of a biotechnology company, and it is measured by number of patents in this study (N Patents). Although the strength of patent protection in biotechnology was under a lot of discussion in the 1980s, there are several reasons supporting the use of the number of patents to measure innovative output. First, patents provide external visibility and legitimacy for a newly established organization (Aldrich & Fiol, 1994). For example Singh, Tucker and House (1986) showed that by acquiring external and formal legitimacy, young service organizations were able to increase their chances of survival. Since new biotechnology organizations have limited  legitimacy, and few relationships in industry networks, obtaining patents early on is likely to promote future growth and survival. Second, biotechnology firms have a high propensity to patent, since pharmaceutical patents are the most effective of all industries (Levin et al., 1987). Third, several authors have successfully used patents to operationalize the innovative output of biotechnology companies (see for example Shan et al., 1994). Patent information for this study was obtained from the U.S. Patent and Trademark Office documents and it includes patents that the firm had applied for and subsequently obtained during the first seven years since its founding.

    To distinguish between companies that concentrate on incremental improvements, and those that focus on more radical innovations, we include an additional measure of the innovative output: patent citations (Cit/pat). Several authors (see for example Jaffe, 1986; Trajtenberg, 1990) have argued that since patents can vary enormously in their importance or value, simple patent counts are unlikely to totally capture the innovative output of companies. Consequently, they propose patent counts weighted by citations as an alternative measure. In other words, patent citations (i.e., references that a particular patent has received in subsequent patents) should indicate the technological, as well as economic value of innovations better than simple patent counts (Trajtenberg, 1990; Albert et al., 1991). The implicit assumption of this study is that companies who receive fewer citations focus more on incremental innovations. To measure radical innovative output, all patent citations that the firm’s patents (applied during the first seven years) had received until February 1997 were included, and divided by the number of patents. The more citations these patents had on average, the more radical the firm’s innovation portfolio was taken to be.

    Although some firms (those established in the early 1980s) had considerably longer periods for their patents to be cited, Jaffe et al. (1993) have empirically shown that most patent citationsoccur during the first three years since the patent has been granted. To be on the safe side, companies founded in 1988 were excluded from the test, since it is likely that some of the patents of these companies had not had a long enough time to be cited.

    Following Eisenhardt and Schoonhoven (1990), the third dependent variable, organizational growth (Ln sales), is operationalized as the difference in sales at year five, relative to sales at founding. The rationale behind this 5–year lag is that new firms require an initial period in which to perfect their product design and manufacturing; differences in financial data are usually not meaningful in the early period (Romanelli, 1989). 10–K reports of the companies were used to collect this data. The sales figures were adjusted to constant dollars (1990) using the annual Producer price index of the Bureau of  Labor Statistics, and logarithmic scale was used. An alternative measure for this variable would have been employee growth (Powell et al., 1996), but controlling for yearly differences would have been more difficult.
 

Top of page | Chapter Listing | Return to 1997 Topical Index


1997 Babson College All Rights Reserved
Last Updated 03/15/98