Why some ideas are worth more than others?

Data from the London School of Economics suggests that the more ‘knowledge spillover’ is associated with an idea, the greater its value

Some ideas create new forms of knowledge that extend far beyond their direct commercial applications and spill over into other areas. Such ‘knowledge spillovers’ arise if an innovation generated by one inventor helps others come up with further inventions of their own. These spillovers are often what create an idea’s biggest impact, and therefore, its greatest value.

Spillovers are the follow-on benefits resulting from an initial investment. These are frequently produced in geographic areas where industries have formed “clusters.” Silicon Valley is the classic example for the tech industry. Strategy expert Michael Porter has also identified industries where clusters of companies in the same field created knowledge spillovers that sent productivity gains rippling throughout the local economy.

If we could more accurately measure the spillover effects associated with particular types of technology, we could deliberately channel financial resources and human capital toward the ideas most likely to produce the broadest benefits. Research conducted by economists at the London School of Economics (LSE),together with Huawei, provides some ideas of where to start.

Measuring innovation’s impact

Traditionally, researchers attempting to measure the impact of knowledge spillovers in industries or technologies have been confined to legacy techniques. For example, when studying the subject at the enterprise level, they have focused on measuring R&D capital stocks. Researchers have looked at how the quantity of money spent on R&D caused the innovation profiles of companies to rise and fall.

At a national level, economists have used patent citations to measure the influence and value of an invention. As a proxy for influence and value, citation analysis does an adequate job, but, it has limitations. Not all innovations are patented, and not all forms of knowledge spillover are captured by patent citations. Particularly, when the goal is to work across different technologies and different countries, there have been limited alternatives for empirical work in this area.

Patent rank: New addition to measurement tool box

To get around these limitations, LSE researchers introduced a new analysis method called Patent Rank. An adaption of the well-known algorithm used by Google to assess the relative importance of web pages, Patent Rank analyzes patent citations, instead of hyperlinks. This analysis identifies the patterns in which knowledge tends to flow. Unlike traditional citation counts, Patent Rank captures the indirect connections between ideas. Researchers have used Patent Rank to make some interesting discoveries.

First, the Patent Rank method of analysis strongly suggests that ICT generates knowledge spillovers far more than those produced by other technologies. Within the ICT category, wireless technologies generate more than 50% higher knowledge spillovers than the other comparable fields, with an average of around US$900,000.

A second finding is that ICT knowledge spillovers have fallen, on an average, across all technology areas worldwide following a sharp spike in the early 1990s. A pattern of falling research productivity is evident in areas including semiconductors, drugs, and general listed firms’ performance. This pattern suggests that we are living in an era where increased inputs (hiring more scientists, increasing R&D expenditures) are needed just to maintain a constant level of innovation output. In other words, we are experiencing decreasing returns to scale. Despite the decline, ICT spillovers are still 25% to 50% higher than the pooled set of all other technology areas at the end of the period considered.

The value of knowledge spillovers from ICT grew rapidly from the mid-1990s

The evidence indicates that knowledge spillovers evolve in cycles. A particular scientific breakthrough might trigger a wave of innovations whose effects work their way through the economy over a period of years or even decades.

The cyclical nature of spillovers could be one reason for the decline observed during this period. Lags in the patent citation process have prevented researchers from capturing the benefits of knowledge spillovers – of the latest wave of robotics research in the late 2010s, for example, or the surge in AI-related research since 2009. The recent surge in AI-related innovations suggests that the average knowledge spillovers are likely to rise in the near future, although it is hard to predict the size of this effect.

How should policymakers take advantage of these new insights?

Firstly, national governments should use the LSE findings to optimize their national innovation strategies by sharpening their focus on ICT.

Some countries are already racing to adopt digital technology. In China, huge investments in 4G telecommunications have helped expand coverage from 7% of the population in 2014 to 69% last year. The government has also launched a three-year plan to spur development of AI. Nearly half the world’s smart cities initiatives are in China, backed by large investments in data centers, cloud services, data analytics, and the Internet of Things.

Governments in Europe, by contrast, could strengthen their position further. McKinsey estimates that Europe has captured just 12% of its potential gains from digital technologies, while PwC says that just 5% of manufacturers in Europe, the Middle East and Africa are “digital champions,” compared to 11% in the Americas and 19% in Asia-Pacific. All other things being equal, China is more likely to benefit from the outsized knowledge spillovers, resulting from investment in ICT, than will the European governments. Secondly, policymakers can use Patent Rank to develop or refine their national innovation strategies. As the World Bank notes,such strategies can serve as engines of GDP growth.

Innovation strategies require measurement. IMF research shows that the existing methods of measuring the indirect benefits of the digital economy are inadequate. It creates the risk that governments won’t set the right strategy, or, if they don’t clearly understand the benefits the digital economy can deliver. Capturing and quantifying the social value of technologies and ideas, as measured by the Patent Rank algorithm, can help keep them on track.

Third, national governments can use Patent Rank to prevent under-investment in ICT by the private sector. As the broad public benefits of ICT-led knowledge spillovers aren’t solely captured as corporate profits, some companies may under-invest in ICT to the detriment of the wider economy and society. To counter this tendency, policymakers should consider creating incentives, such as tax credits and subsidies, designed to encourage ICT production and diffusion activities. Many already do this in other areas where there is a clear “social multiplier effect”.

Fourth, governments often express a desire to support areas of national comparative advantage. In fact, the LSE research found that knowledge spillovers at a national level closely match a country’s innovations per capita in a particular technology field. Basically, countries usually develop a comparative advantage in one or two kinds of technological innovation, which then breed knowledge spillovers in a network effect.

Governments may want to consider divesting control or oversight of ICT policy similar to how central banks behave over national monetary policy. Such independent control would allow enhanced long-term planning, while insulating policymakers from political pressures. Targeted and enlightened support for the most powerful ideas and inventions will be of enormous benefit to developed societies around the world as we approach the fourth industrial revolution.

Andrew Williamson, VP, Market Insights, Huawei Technologies Co. Inc