Page - (000307) - in Knowledge and Networks

304 Results Table 14.4 reports the results of regression estimates. The first columns present the model with only the control variables. The following columns progressively include the variables accounting for external linkages and gatekeepers. All mod- els report estimated coefficients transformed to incidence rate ratios (IRR), defined as exp b        . Concerning the control variables, the coefficient of the number of internal pat- ents shows, as expected, that the scale of inventive inputs and experimentation mat- ters. More specifically, estimates indicate that, all else being equal, a unit increase in the (log) number of internal patents is associated with a doubling in the number of new combinations of technology groups introduced in a city. Concentration of inventive activities among relatively few firms is negatively associated with technological recombination, suggesting that competitive market structures are more conducive to a renewal of the local knowledge base. As far as the specialization variables are concerned, absolute technological specialization in a narrow set of technology fields has a negative impact on the ability to expand the knowledge base, whereas relative technological specialization (i.e., a dissimilarity of the technological profile to that of the other cities with which it exchanges knowl- edge) is positively related to the number of new combinations introduced in a city. A plausible interpretation of this result is that a lower cognitive overlap enhances complementarities and opportunities for learning. Table 14.2 Summary statistics of dependent and independent variables Variable Mean SD Min Max Number of new combinations 30.29 26.64 0 217 INTPAT 136.24 309.37 1 2916 HFIRMS 0.13 0.15 0 0.91 HPATENTS 0.07 0.06 0.01 0.49 NPLCIT 1.28 1.07 0 6.26 KI 1.40 0.35 0.37 1.99 LARGE 12.22 10.35 1 67.26 CLUST 0.25 0.23 0 1 ADWR 7.30 10.10 0.07 162.99 NUMGK 91.27 259.60 0 2876 SHREACH_GK 0.38 0.25 0 1 SHINDIR_GK 0.24 0.20 0 0.86 SHDIR_GK 0.14 0.09 0 0.67 GKREMOV 0.45 0.24 0 1 CONCGK 0.78 0.28 0 1 Note. 2940 observations (196 cities × 15 years). Source: Elaboration by authors S. Breschi and C. Lenzi