Fig 1.
Example: calculation of total, local, and global crowding.
Total crowding calculates the overlap in citations (of antecedent inventions) between our focal and all technological components. Local crowding calculates the overlap in citations between our focal and biotechnological (local) components only. Global crowding calculates the overlap in citations between our focal and all non-biotechnology (global) components.
Table 1.
Descriptive statistics.
Table 2.
Correlation matrix.
Fig 2.
Yearly entry rate of biotechnology patents.
The yearly entry rate of biotechnology patents shows a steady increase until 1998, after which we can witness a decline in the number of patents. This seems to indicate that biotechnology has entered into a different stage of technological development, in accordance with the S-shaped growth path commonly reported in technological development.
Table 3.
Descriptive statistics of individual component niches.
Fig 3.
Locally weighted regression (lowess) of relative entry per niche.
While a clear decrease in the ‘average’ relative entry of patents in component niches cannot be observed in this figure, a decreasing growth rate in the relative entry rates (or leveling-off) can clearly be established, again in accordance with the S-shaped growth path of technological development.
Fig 4.
Yearly entry in clustered component niches.
As we can learn from Table 3, the major components are in groups 1 to 3, which display a common pattern. Groups 1 to 3 all peak between 1998 and 2000, and the main different between these groups of components is what occurs after this peak. The growth of the components in group 1 declines, whereas the growth of components in groups 2 and 3 first declines but then picks up again. Groups 4 to 6 show a rather erratic entry rate. The reason for the difference in growth patterns of individual components is that they are located differently along the technological lifecycle. Some are in the beginning of their development while others seem to be more fully developed.
Table 4.
Negative binomial random effects panel regression estimates.