Originally Published MDDI January 2006
Manufacturers should know and exploit three sources of competitive advantage: proximity, location, and milieu.
By Fritz Yambrach
The medical device industry is a competitive environment with an increasing number of small and medium-sized firms entering the market each year. If manufacturers can draw upon external resources to enhance their competitive position, they can improve their probability of success.
Medical devices range from commodity products to highly technological instruments patented for their unique and specific design features. The medical device industry by its nature is an integration of technical activities: biological sciences, mechanical engineering, materials science, and entrepreneurial behavior, to name a few. The interaction among these activities creates networks of knowledge transfer, and such knowledge sharing has been shown to promote business activities.1
Medical device firms adopt and develop new technologies during the process of product innovation as a means of developing competitive advantage. Michael Porter has described a firm's ability to create a competitive advantage as a combination of internal and external factors.2
These general factors have a proximity component that enhances interaction. The accepted assumption is that close proximity, i.e., the location of firms relatively near external resources, increases the amount of interaction, and that increased interaction facilitates a network of knowledge sharing that benefits the firms. The presence of these resources creates a milieu that may explain some of the differences in the innovative capabilities of device companies.
The purpose of the investigation discussed here was to determine whether certain resources would improve a device firm's competitive position if the firm were to locate near such resources.
The medical device industry is far-reaching, but in the United States, many device firms are centered in discrete regions (see Figure 1). Six states dominate the U.S. device industry: California, New York, Massachusetts, Illinois, Florida, and Texas account for 5681 of the 12,420 medical device manufacturers registered in the United States.3 The European Union is another major producer of medical products with the same type of spatial concentration of firms as found in the United States. Germany has a number of companies in the device industry clustered around the Baden-Wuerttemberg area.4 Table I shows the number of U.S. medical device firms and the number of employees in each sector. The Medical Device Register (MDR) provides a list of North American medical device manufacturers.5 The biannual registry lists more than 10,000 firms engaged in the medical products industry. More than 6700 are device manufacturers. The data include the firms' zip codes, which were used to identify metropolitan areas for this research. The research sample included approximately 1000 firms registered with FDA.
It is important to look at each factor that influences whether a device firm may locate in a particular area.
Universities with Biomedical Engineering Programs. One influence on medical device firm location decisions is the presence of university programs that train individuals in the development and design of medical devices. These engineered devices interact with human biological systems, and universities combine the disciplines of engineering and biology to train students and to act as incubation sites for idea development. The location of these engineering programs in a particular region creates a source of trained individuals who are experts in the field and who serve as potential participants in an information network.
Biomedical engineering departments serve as an example of the relationship between the industry's application of technology and academic technology development. These departments provide multidisciplinary programs that bridge traditional biology and mechanical engineering disciplines, and thus address the needs of the medical product industry. For this study, academic institutions were aggregated at the metropolitan-area level from a list of all U.S.-based biomedical engineering programs in universities and colleges.
Medical Schools. Medical schools are similar to bioengineering programs in their ability to develop and share specific knowledge with the medical product industry. The schools can provide an incubator for product ideas because students are learning medical procedures and the use of equipment. The University of California at Irvine, for example, has set up a society for medical research with the intent of stimulating the transfer of technologies from the university to industry. Research by Zinner in investigating the effects of National Institutes of Health (NIH) grants has shown that 55% of citations within patent applications for medical products refer to work from academic sources.6 For this investigation, a list of medical schools was aggregated at the metropolitan level based on the amount of research dollars provided by NIH.
Suppliers to the Medical Device Industry. The supplier network in the medical product field is an interesting influence to investigate because of the types of technology that must address specific industry requirements. The supplier network in this industry tends to specialize in supplying medical product firms with raw materials, research and development services, and other contracted services. The relationships between these suppliers and medical product companies provide examples of knowledge spillovers or sharing of technologies.
The MDR identifies the supplier groups specific to the medical device industry, which are categorized in product design, manufacturing, product R&D, and process validation. This supplier information was gathered and aggregated at the metropolitan-area level to be compatible with the other data.
A College-Educated Population. To consider the relationship between an educated population and technical firms in a given region, data on the education level attained by local residents was obtained from the U.S. Census Bureau Department of Statistics. This education level is expressed as the percentage of the total metropolitan population having a bachelor's degree or higher. The data were aggregated at the metropolitan-area level.
Each of these external factors can add value to a firm, enabling it to leverage skill sets outside its existing resources. This interaction between external resources and a firm contributes to innovative activity.
Firm Size. For this study, firms were categorized by annual sales revenue in order to evaluate businesses of like size. There is a consensus in the literature that different-sized firms face different issues in their operations.7 The distribution of medical device firms by size is interesting because the bulk of the industry is concentrated in small and medium-sized firms. The number of firms sampled in various sales categories is presented in Table II.
Patents. For this study, the number of patents a particular firm receives was used to determine the company's innovative level. A list of patents was obtained for each firm for the five-year period from 1999 to 2004. Data were obtained from the U.S. Patent Office.
Location. The locations of firms were aggregated at the metropolitan-area level to be compatible with geographic sources of competitive advantage that were also gathered at the metropolitan level. The zip code of each firm was encoded into a database for inclusion in metropolitan areas. A firm's address and zip code information were obtained from the MDR.
The basic relationships between variables were based on a correlation analysis that was performed to identify possible relationships. The analysis measured the extent to which a given two variables were related and the measure of correlation, r, was used to test that relationship.
Preliminary examination of the data did not allow us to assume that the data were normally distributed, so a nonparametric technique, Spearman's rho, was used as the correlation determinant. These relationships shed light on the strength of the relationship between the variables.
A regression analysis was used to identify the strength of various geographic variables in influencing a firm's innovative performance and in influencing its location. The variables' strength was viewed by analysis of the beta coefficient. The regression technique starts with a model containing the pertinent independent variables and analyzes the change in the dependent variable. The square of the correlation coefficient, R2, of the model is the amount of change in the dependent variable attributable to the independent variables used in the model.8 The strength of each independent variable's contribution is determined by the beta coefficient of each variable.
The population of firms comprising the sample was drawn from the MDR. The firms were stratified by sales revenue into categories representing different sales volumes. A random sample was taken from each category. This stratified sampling technique was based on using a proportionate number of firms in each sample group to represent the number of firms in each sales category. For example, firms in the <$1-million category constitute 25.08% of the total number of device firms in the MDR. The sample from this group is equal to 25.08% of the sampled firms, or 251. These 251 firms were randomly sampled from the total number of firms in that particular category.
The total number of firms sampled was approximately 1000, as indicated in Table II. The importance of investigating medium-sized and small firms is supported by the fact that more than 84% of the total firms listed in the MDR have annual sales of less than $25 million a year. Small and medium-sized firms are the dominant groups in this industry.
Education Levels. The anticipated relationship between the level of education and the number of firms locating in a metropolitan area was the first correlation examined. The rationale behind this expectation is that high-technology firms require a highly educated and trained workforce. This relationship has a Spearman's correlation of 0.368 and is statistically significant with a probability of 0.001. The Spearman correlation is a nonparametric statistical technique that is used when a regular distribution cannot be ensured.
The analysis indicated that the relationship, although relatively small and linear, is significant. A correlation between the education level of a region and the location of high-technology industries is not strongly supported by data. However, for the medical device industry, the relationship is linear and statistically significant. Medical device companies do not seem to locate near the highly educated areas of the country. This result may be skewed by the fact that medical device firms are found mostly in the largest metropolitan areas, where the education figure is smoothed out by a large population.
Specialized Suppliers. The relationship between specialized suppliers and the number of device firms in a region was the second to be examined. It was found that the relationship between device manufacturers and specialized suppliers is statistically significant. The correlation coefficient is 0.595 with a probability of 0.001 (see Table III). In analyzing the data from all metropolitan areas, it became clear that a strong positive correlation exists between medical device manufacturers and specialized suppliers collocating in the same region.
A more-detailed review of these particular data posits that firm size affects a firm's location near specialized suppliers. These relationships were examined for manufacturers (based on sales volume) with specialized suppliers. Again, a Spearman correlation was used to describe the relationships that are statistically significant with a probability of 0.001. The correlation is strong for small and medium-sized firms (those with $25 million or less in sales). The relationship weakens significantly for firms with sales of $25 million–$100 million and strengthens again with firms that have more than $100 million in sales.
This relationship is supported by highly significant correlations between specialized suppliers and manufacturers' locations. The high positive correlation between the location of small and medium-sized manufacturers and the location of specialized suppliers confirms the literature.7 The correlations support three statements:
• Medical device firms and specialized suppliers locate near each other, and this is especially true of small and medium-sized firms. It is possible that small and medium-sized firms locate near suppliers or that suppliers tend to locate near small and medium-sized firms, but it is not clear from this investigation whether the device companies or the suppliers move into an area first.
• Larger device firms (with sales of $25 million–$100 million) are less dependent on suppliers, as evidenced by lower correlations.
• The largest firms (with sales of more than $100 million) are more strongly related to the location of specialized suppliers than was expected. These firms may be so large that suppliers locate near them for access to large project accounts. A contract with a large firm contains large volumes of work and possibly a consistent supply of business contracts for a supplier.
The correlations support the position that small and medium-sized manufacturers do have more interaction with specialized suppliers than larger firms.9 The basic concept is that small firms do not normally have the internal resources for many operational functions and, therefore, seek outside partners to engage in these activities with them. Larger firms normally do have the internal resources to perform these functions.
Academic Sources. This relationship was examined by using two different correlations with two different indicators of academic interaction with medical research. The two variables—medical schools and bioengineering programs—were correlated with the number of manufacturers to see whether there was a relationship between academic research activities and the concentration of manufacturers in a metropolitan area. The correlation between the first pair of variables—that is, number of manufacturers and medical schools—had a Spearman coefficient of 0.581. The correlation between number of manufacturers and bioengineering programs had a Spearman coefficient of 0.503. Both correlations have a probability of 0.001 (see Table III).
These results demonstrate a positive, statistically significant relationship between academic sources of knowledge and a device firm's location. The results support validity for the medical device industry of the concepts presented in the literature.
Innovation as Represented by Patent Generation. Several relationships were investigated to gain insight into the extent that external independent variables contribute to patent generation. In an attempt to elucidate the relationships, a linear regression equation was developed. The equation attempts to identify the relative importance of the independent variables in accounting for the number of patents generated within a metropolitan region.
The regression model displayed in Tables IV and V presents the regression analysis of five independent variables contributing to patent generation. The square of the correlation coefficient, R2, indicates the amount of change that is attributable to the independent variables used in the model. The R2 of 0.678 in Table IV is explained by the variables contained in the model capturing 67.8% of the change in the dependent variable.
In the “Standardized Coefficients: Beta” column in Table V, the standardized regression coefficients measure, in standard deviation units, the change in the number of patents as the independent variable is increased by one standard deviation. The positive relationships evident with medical schools (0.844) and the number of manufacturers (0.224) indicate that these variables contribute the most to an increase in the number of patents a company produces. The presence of medical schools in the model indicates a strong relationship between academic research and patent generation. The interpretation of these results supports earlier findings that the number of manufacturers and medical schools in a region are the most important positive factors in influencing a medical device company's ability to develop innovative products.6
These correlation coefficients between resources in a geographic area that may contribute to patent generation are the key data in this investigation. The number of patents created by device firms correlates with the concentration of firms in metropolitan areas. This correlation is one of the strongest relationships identified in the investigation. The Spearman value between patents and number of manufacturers is 0.775 with a probability of 0.001 (see Table III).
This highly significant correlation coefficient supports the hypothesis that medical device firms that locate near other medical device firms are more innovative in their fields, as defined by patent generation (see Tables IV and V). Additionally, medical schools are highly correlated with patent generation, with a significant correlation coefficient of 0.635. The independent variables used to construct a regression equation were chosen because of their high correlation with patent generation.
Agglomeration. The variables that drive the clustering of medical device firms were reviewed using a linear multiple regression equation method. This technique evaluates the resources in a region that contribute to the spatial concentration of medical device firms (see Tables VI and VII). The regression model displayed in these tables presents the regression analysis of the five independent variables contributing to the number of manufacturers. The square of the correlation coefficient, R2, indicates the amount of change in the dependent variable that is attributable to the independent variables used in the model. The R2 of 0.850 demonstrates that the variables contained in the model capture 85% of the change in the dependent variable.
The correlation table of the dependent and independent variables (see Table III) supports the model developed for the number of manufacturers. A strong correlation is evident between the number of manufacturers and number of patents at 0.775. Additional high correlation coefficients were found between the number of manufacturers and suppliers at 0.595, with bioengineering programs at 0.503 and medical schools at 0.581. These results support the regression model that included these three independent variables, which explains 84% of the variability in manufacturer location in a metropolitan area.
The evidence provided by this research supports the suggestion that there are geographic regions in the United States that have particular attributes that lead to greater innovation in medical device firms located there. This research brings to light some of the resources in a metropolitan area that are specifically significant for the medical device industry and that would be useful to medical device firms currently located or considering locating in a particular area. It identifies two specific factors as important contributors to a medical device firm's ability to innovate. These factors are the proximity of other similar firms in the industry and a milieu
featuring academic sources of knowledge. The contribution of similar firms is recognized in the literature as an attribute of a firm's innovation performance. The positive influence of these other firms in the medical device industry is supported by this research. The research also identifies medical schools as a significant contributor to patent generation. This is also supported by the literature, where medical schools are cited as being a source of innovative technologies.6
The unexpected finding from this investigation is that the proximity of suppliers seems to be associated with small and medium-sized firms more than with larger firms. That fact is of significance to the medical device industry because 54% of all device firms have less than $5 million in annual sales. The presence of suppliers is important to small and medium-sized device firms interested in engaging in information networking.
1. J Adams and Z Griliches, Research Productivity in a System of Universities: The Economics and Econometrics of Innovation (Boston: Kluwer Academic Publishers, 2000).
2. M Porter, The Competitive Advantage of Nations (New York: Free Press, 1998).
3. M Szycher, “The Medical Device Industry,” Journal of Biomedical Applications 11 (1996): 76–118.
4. “Surgical and Medical Instruments,” in D&B Principal International Business: 2000 World Marketing Directory (Bethlehem, PA: Dun & Bradstreet, 2000).
5. Medical Device Register [online] (Los Angeles: Canon Communications, 2004 [cited 20 May 2004]); available from Internet: www.mdrweb.com.
6. D Zinner, “Medical R&D at the Turn of the Millennium,” Health Affairs 20, no. 5 (2001): 202–209.
7. Z Acs and D Audratch, “Innovation in Large and Small Firms: An Empirical Analysis,” American Economic Review 78, No. 4 (1988): 678–690.
8. S Landau and B Everitt, A Handbook of Statistical Analyses Using SPSS (Boca Raton, FL: Chapman & Hall/CRC, 2004).
9. M Storper, Regional World: Territorial Development in a Global Economy (New York: Guilford Press, 1997).
Fritz Yambrach is a professor in the Department of Manufacturing and Mechanical Engineering Technology and Packaging Science at Rochester Institute of Technology (Rochester, NY). Contact him at [email protected].
Copyright ©2006 Medical Device & Diagnostic Industry