![Sample our Economics, Finance,Business & Industry journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5931/TOC-Subject-Sample-2021-Economics-Finance-Business-Industry.jpg"})
ABSTRACT
In this paper, we address the determinants of clean energy inventions by 946 large firms. We use a new set of large firms’ patent portfolios and we broaden and deepen existing literature on this issue in two main ways: first, we conduct our study directly at the firm level and not at the industry or national levels and second, we do not focus on a single industry but encompass all industrial sectors. Drawing on firm (internal and external) knowledge and knowledge accumulation, we show there is a robust positive association between the (past) knowledge accumulated capital related to clean technologies and the number of inventions produced in that field, even after controlling for industry and nation fixed effects and other factors. The same relation works for (past) knowledge-accumulated capital in other (non-clean) technologies. However, the relation’s impact on the number of clean inventions produced is much lower. The magnitudes of our coefficient are in line with that obtained previously on firms in the auto-industry or at the sectoral level.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
* A first draft of the article was set out at the international conference ‘Gouvernance of a Complex World 2014’, 18–20 Jun 2014, Torino (Italy). We warmly thank our discussant Cédric Gossart for his valuable comments. We are extremely grateful for guidance and feedback from anonymous referees.
1. We define clean energy technology later in the paper.
2. During the last ten years, cleantech and greentech are most often used as synonyms. However cleantech has historically been differentiated from greentech by its origins in the venture capital investment community. It used to define a business sector that included significant and high growth industries such as solar, wind, water purification and biofuels. This was in contrast to greentech, an older term representing the highly regulatory driven, ‘end-of-pipe' technology of the past with limited opportunity for attractive returns, cleantech is driven by economic market, therefore offering a greater financial upside.
3. Regulation is a crucial driver of eco-aiming to correct negative externalities due to pollution.
4. It is acknowledged that environmental innovation oriented policies positively affect the rate of technological change in the field of green technologies.
5. We consider the terms ‘green-innovation’ and ‘environmental innovation’ equivalent. However, we explicitly differentiate the term clean innovation, that is, climate-friendly innovation, from all other green innovations.
6. Other sources and drivers of innovation play a role in this process as organisational arrangements (Dosi Citation1997).
7. Here when we speak about dirty technology, we mean in fact not-clean technology. Many authors put the word dirty in brackets in order to show that dirty technology may not augment the negative impact on climate change.
8. Many authors also take R&D expenditure as a proxy of innovation efforts in clean tech (see Jaffe, Newell, and Stavins Citation2002 for a survey) despite the inability to observe the share of R&D budgets dedicated to cleantech fields.
9. As previously done by Patel and Pavitt (Citation1991).
10. More information on the data set is provided by Laurens et al. (Citation2015).
11. When the firm database was built on an earlier version of Patstat (2009), the latest available year with a complete collection of patent applications in the database was 2005. For more recent years, the patent collection was not yet completed.
12. Priority patent count data have many advantages (see Rassenfosse et al. Citation2013).
13. Using the Orbis database edited by Bureau van Dijk Electronic Publishing we defined the global ultimate owner (GUO) for each of the firms and identified all subsidiaries in which one of the GUOs had more than 50.01% of shares.
14. See the paper by Vezzani et al. (Citation2014) for an overview on PATSAT matched with the EU scoreboard firms in other technological fields.
15. A great deal of literature deals with the problem of the delineation of technologies in nanotech (see Foray, Lhuillery, and Raffo Citation2010 and references therein), in biotech (OECD Citation2005) and of course in environmental technologies (see Oltra, Kemp, and De Vries Citation2010; Haščič and Migotto Citation2015). The Y02E classification proposed in 2012 was a major improvement in identifying environmental innovation (Haščič and Migotto Citation2015) and is now used by specialists to mitigate identification problems (Leydesdorff et al. Citation2015).
16. In our opinion, lexical analysis is a powerful tool to retrieve scientific publications in a given field, but is less adapted for patents where lexical terms are often less meaningful.
17. It counts as equal those patents applied for in different patent offices where the rules for patenting can differ to a large extent. For example, the level of inventiveness to apply for a patent is Japan is lower than in patent offices in Western countries. However counting priority patents avoids national bias, encompasses all patents without any selection. Furthermore, using the first application for the first invention means that the filing date is closer to the invention date.
18. The strong commitment of car manufacturers to clean patenting is linked to the boom of patents in batteries and fuel cells since the late 1990s. They represent more than half of the clean patents of our dataset.
19. Considering that dirty and cleantech are distant technologies and that two distinct absorptive capacities are then deployed by firms, alternative solutions would be to compute or
.
20. When the K variable is null, the value of K is set to log (0.001). Certain R&D investors did not file any invention over the 2003–2005 period. In this case, the explained variables are adjusted and set to log (0.001) with dummy variables set to 1 when this adjustment occurs.
21. Testing α = 0 with .
22. The only difficulty for the reader will thus be thus the interpretation of the coefficients of the dummy control variables, when departs from 0. In this case, the magnitude of the coefficient is computed comparing the patent number predicted when the dummy set to 1 and set to 0 (
) and cannot be interpreted directly anymore since γ = ln(1 + r) departs from r.
23. Two variables not included in the final estimation have no impact: the firm level of R&D internationalisation and the degree of firm technological diversification.