Following Audretsch and Acs (1991), industry dynamism was operationalized as the
intensity of industry-level R&D. Industry R&D levels are measured at the North American
Industrial Classification (NAICS) three-digit level, based on a 3-year average (1999–2001)
of data from the Statistics Canada Annual Survey of R&D in Canadian Industry. This
measure of R&D intensity was employed in the regression models to locate firms on a
spectrum from high to low dynamism. Industries were also categorized as high- or lowtechnology on the basis of standardized scores for R&D intensity and the percentage of
knowledge workers in each industry (Bolland and Hofer, 1998). This made it possible to
evaluate the hypothesized relationships within the subsets of relatively dynamic and stable
industries. Using technology indicators is consistent with Zahra and Neubaum’s (1998)
assertion that high- and low-technology regimes differ in their levels of dynamism. Those
industries with positive standardized scores on both R&D and knowledge level were
classified as high technology. Low technology industries were those with standardized
scores below zero (see Appendix A, Table A1).