Technology network focusing on “small” firms

The “technology space”, or “technology network”, is a network in which each technological field is a node and the weights of the edges connecting the nodes reflect the proximity of the fields. The technological network has been recognized for its usefulness in understanding the processes of technological development and convergence, and for predicting future developments. The proximity between any given technological field is measured by the co-occurrence of IPC codes within a firm's patent portfolio. By mapping a firm's patent activities on the technology network, it is possible to predict the fields in which the company should newly advance according to its technology portfolio [1]. Ref. [2] compared different methodologies for the prediction of firm’s future submission of patents in new sectors. While these studies provide us with very useful insights into technology development, they do not provide sufficient analysis or strategic planning suggestions at a firm level. They focused only on major firms, but the strategies for technology portfolio building of small firms are not necessarily the same as those of major firms. (Note that here, “small” means that a firm has a small technology portfolio – i.e., with only a small number of patents.) Nowadays, small firms such as start-ups with severely limited resources are the key to major technological innovation. In this study, we compare two technology networks: one constructed by capturing the proximity between technological field in the usual way (hereafter, RCA-based network), and one in which technology portfolios of small firms are also reflected in the proximity (hereafter, non-RCA-based network). The proximity between a given pair of technology fields in the conventional measure takes into account RCA (Revealed Comparative Advantage), which means that those firms with small portfolios or evenly diversified portfolios are neglected. Therefore, we used another major to construct the non-RCA-based network, which is simply not using RCA. The data we used were on patents filed between 2000 and 2020, extracted from Orbis IP, one of the largest patent databases. Panel A in the figure shows the annual change in the total number of firms whose portfolio information is included in the proximity calculations. It indicates that the portfolios of the majority of firms are actually not reflected in the conventional RCA-based network. That is appropriate when one wants to look at technology developments, but not when one wants to see where in the network new technologies may emerge (often driven by small firms) and how small firms should expand their portfolios. By comparing the two networks, we could understand which combinations of technological fields are underrepresented in the conventional RCA-based network (see Panel B). Furthermore, we classified patterns of temporal changes in proximities between technological fields. Our analysis captured multiple pattens such as small firms achieving a combination of rare technological fields and then major firms beginning to focus on that combination, or small firms expanding their portfolios into fields that major firms never even looked at in the first place.