IP and the economy: Key roles and impacts
1. Collaboration, commercialisation and patenting in Australia
2. IP and competition: How broad are Australian patents?
A key aspect of patents is their scope, the extent of legal protection that they provide. Broader patents, such as those for more general discoveries or ideas, can spur cumulative investment within patenting firms, as pioneers seek to build on their own discoveries. Conversely, narrower patents increase the likelihood that others – such as a patentee’s rivals – can design around the patent without infringing it and contribute to improving a technology. Machine learning is opening new opportunities to evaluate the characteristics and quality of patents. In 2022-23, IP Australia partnered with the Centre for Transformative Innovation (CTI) at Swinburne University of Technology to assess how Australian patents compare to patents granted for the same inventions in the United States and at the European Patent Office.1
At a glance
-
Australian patents narrowed in the 2010s, relative to patents for the same inventions granted in the European Union and United States, reversing a trend toward Australian patents broadening in relative scope.
-
Australia’s past patent reforms – the Intellectual Property Laws Amendments (Raising the Bar) Act 2012 – contributed to reducing the likelihood of Australian patents being broader than their international equivalents by between 1.0 and 4.0 percentage points.
-
The reforms’ estimated impact was greatest in ICT/software, where open innovation communities and principles are an important force for technological progress.
-
As of 2019, around 51% of Australian patents are estimated to be similar in scope to their international equivalents, around 35% are estimated to be broader, and around 14% are estimated as narrower.
Why patent scope is important
In 1880 Thomas Edison received a basic patent for his “electric lamp… giving light by incandescence.” The invention paved the way for widespread use of electric light. The patent’s first claim was broad – a lamp using a carbonised filament as the source of light. However, the patent was narrowed because the filament needed to be of “high resistance.” Subsequent claims also had a narrowing effect: the invention relied on a bulb made of a single piece of glass that creates a vacuum, and technical principles for connecting the filament to electrical conductors.
The patent provided Edison’s company – now General Electric – a dominant position in the nascent electric lighting industry.2 However, it was sufficiently narrow that others designed around the patent in developing substitute technologies, advancing electric lighting and contributing to the industry’s development.3
A patent’s scope determines its economic impact. Broader patents can spur cumulative investment within pioneer firms.4 However, the broader a patent, the more rival products and processes are likely to infringe it. As a result, broader patents can also limit rival innovation and affect market entry decisions by potential competitors.5
Evaluating patent scope is difficult, since a patent’s scope is defined by the text in its claims. New research methods such as machine learning are opening new opportunities for evaluation and benchmarking. In 2023, IP Australia partnered with the Centre for Transformative Innovation (CTI) to assess Australia’s relative patent scope – the scope of Australian patents compared to patents granted for the same inventions in other jurisdictions – and the impact of past legislative reforms.
Measuring patent scope
The CTI project developed in several stages. First, the team tasked IP Australia’s patent examiners with manually assessing 500 patents for their scope, relative to their equivalent patents granted in the United States and, separately, at the European Patent Office.
Second, the CTI team used these assessments to validate certain metrics and as inputs to train and test supervised machine learning models. A key metric is the difference in word length between the patents’ first claims: longer claims (e.g., “an incandescent lamp using carbonised paper as filament” vs “the incandescent lamp”) tend to include more detail and modifying statements which will narrow a patent’s scope. The CTI team validated this metric against the examiners’ manual assessments.
The machine learning models predict scope based on a range of textual features of the patent’s claims. Applying the models resulted in a sample of over 80,000 “patent pairs” for analysis. The data covered close to the full population of patents for which applicants requested examination in Australia between 2000 and 2019 and for which equivalent patents were granted by the European Patent Office (EPO) and the United States Patent and Trademark Office (USPTO).
Trends in the relative scope of Australian patents
In 2013, Australia significantly overhauled its patent system, when the Intellectual Property Laws Amendments (Raising the Bar) Act 2012 came into effect. The ‘Raising the Bar’ reforms aimed in part to narrow the scope of patents granted in Australia to bring them in line with international standards.
Figure 2.1 charts the proportion of Australian patents predicted to be either narrower, similar or broader than their international equivalents using the Machine Learning model. The figure presents results based on both the full set of claims in the patents and the patents’ first claims. Patents are charted over time based on their examination request date. In Australia, applicants must request examination within 5 years of an application’s filing date or the application will lapse.6
The results suggest the proportion of Australian patents that are broader has declined after 2010, reversing a trend toward Australian patents getting broader. The downward trend in broader patents appears to have corresponded with an increase in the share of Australian patents with narrower equivalents, and a trend toward Australian patents becoming more similar in scope continuing to 2019.
While the chart suggests a small uptick since 2015 in the proportion of Australian patents that are broader, partial data for 2020 (not charted) suggests that the trend is flat. Consistent with this finding, the median difference in word counts between Australian patents and their equivalents has remained flat over the same period. Continued monitoring over coming years will confirm whether the trend remains stable.
Any changes in relative scope over time may reflect changes in Australian patents but may also partially reflect changes in patent scope within the United States or European Union. For example, previous research has found that US patents were narrowing between 2004 and 2014, after broadening from the early 1980s to 2004.7
Figure 2.1 Proportion of Australian patents that are broader, narrower and similar in scope to their international equivalents, by examination request year (averages based on comparisons to US and European equivalents)
Source: IP Australia; USPTO, PatentsView, 2022; EPO, 2022. Note: Values represent averages calculated from relative scope measures based on first claims and full set of claims.
As of 2019, most (around 51%) of Australian patents are estimated to be similar in scope than equivalents granted overseas, a minority (around 35%) are broader, and a smaller share (around 14%) are narrower, based on the Machine Learning models. These findings are consistent with results from the manual assessment of 500 patents pairs by examiners. Based on a small sample of patents, they assessed that around 38% of Australian patents had first claims that were broader than their international equivalents after the Raising the Bar reforms.
The impact of the Raising the Bar reforms
Using the same data, the CTI team estimated the impact of the Raising the Bar reforms in 2013 on the scope of Australian patents. Based on economic modelling, the reforms are estimated to have reduced the likelihood of Australian patents being broader than their international equivalents by between 1.0 and 4.0 percentage points. The effect is significant considering that an estimated 38% of Australian patents were broader prior to the reforms.
Given that the narrowing of Australian patents appears to have begun prior to the reforms, caution is warranted in ascribing causality to the reforms. That said, the results were robust to shortening the study’s timeframe to focus on the period immediately before and after the reforms.
The estimated impact of the reforms varied slightly across technology areas. The strongest impact was in ICT/software, while the weakest impact was in more traditional technology fields like mechanical and process engineering. Like aircraft and semiconductors, computer technologies are complex and cumulative technologies: many inventors contribute to technical advances, developing different components that build on existing technologies. Narrower patents, alongside open innovation, can be important in these contexts. Semiconductors have undergone rapid advancement in part because the pioneer patents were freely licensed and proprietary knowledge was shared.8 Advances in artificial intelligence and quantum computing also rely on open innovation, open source and knowledge sharing, according to market leaders like IBM.9
New opportunities for evaluation and benchmarking
New research methods like machine learning are creating new opportunities to evaluate and benchmark IP rights. This complements other forms of benchmarking conducted by IP Australia and activities under the agency’s ISO-certified Quality Management System. Techniques for measuring patent scope remain experimental in nature and the focus of continuous improvement by researchers. As they are further improved, they will serve as valuable tools for monitoring trends in patents, and evaluating policy, to ensure the IP system remains fit for purpose.
View our economic research papers.
- Kollmann, T., Palangkaraya, P, Sarwar, A., Webster, E., Anglim, C. and Falk, M. 2024. Raising the Bar reforms: Measuring the impact on relative patent scope. IP Australia Economics Research Paper Series 14.
- Merges, R. P. & Nelson, R. R. (1990). On the complex economics of patent scope. Columbia Law Review, 90(4), 836-916.
- Katznelson, R. D. and Howells, R. (2021). Exclusive rights stimulate design around: How circumventing Edison’s lamp patent promoted competition and new technology development. Journal of Competition Law & Economics, 17(4), 1007-1052.
- Arora, A., Belenzon, S., Marx, M. & Shvadron, D. (2021). (When) does patent protection spur cumulative research within firms? NBER Working Paper 28880.
- Merges, R. P. & Nelson, R. R. (1990). On the complex economics of patent scope. Columbia Law Review, 90(4), 836-916.
- Under Australian legislation, a patent is examined only once the applicant has requested examination. The request can be voluntary or result from the Commissioner of Patents directing an applicant to request examination.
- Marco, A. C., Sarnoff, J D. & DeGrazia, C. A.W. (2019). Patent claims and patent scope. Research Policy, 48(9), 103790.
- Lim, K. (2009). The many faces of absorptive capacity. Industrial and Corporate Change, 18(6), 1249-1284. See also Von Hippel, E. and von Krogh, G. (2006). Free revealing and the private collective model for innovation incentives. R&D Management, 36(3): 295-306.
- Gil, D. (2023, 7 January). Why IBM is no longer interested in breaking patent records – and how it plans to measure innovation in the age of open source and quantum computing. Fortune. https://fortune.com/2023/01/06/ibm-patent-record-how-to-measure-innovation-open-source-quantum-computing-tech/.