Innovation Policy and Digitalization

Dominique Guellec, OCDE © 2019 EPFL

Dominique Guellec, OCDE © 2019 EPFL

Dominique Guellec is Head of the Science and Technology Policy Division of the Organization for Economic Co-operation and Development (OECD). He was invited by the College of Management of Technology to discuss the major societal implications of digitalization. Professor Gaétan de Rassenfosse took the opportunity to interview him.

1) Most innovations are digital: why and what are the implications?
All innovations have either a digital form, a digital distribution mode or are themselves the result of a digital process. Digitization affects all areas of research (electronics, chemistry, life sciences, etc.), which has multiple and varied implications:

  • First, production and access to data becomes essential. Without data, or without access to data, we cannot innovate anymore. As a result, data access rules become paramount.
  • Second, economic activity becomes more fluid. Digitized data (unlike physical goods) can be reproduced and manipulated at almost zero cost and is not subject to transportation costs either. In addition, once produced, it can be accessed and manipulated all over the world simultaneously by an infinite number of actors. There is no material equivalent to such a property, which has direct implications on competition in particular (see third question below).
  • Third, the higher frequency of digital innovations compared to hardware innovations. If we take the example of a mixed product (an autonomous vehicle for example), its material component will be renewed every couple of years. Renewal involves mobilizing distribution channels, ordering equipment, etc., while the digital part of the product can be updated several times a day by changing a few lines of code.
  • Fourth, digitization has the effect of blurring the border between services and manufactured goods. It is expected that people will no longer buy a car but will purchase a transport service instead, relying on digital management of networks and services. The winners will be both traditional companies such as BMW or new players such as Tesla or even Uber or Google.

2) Innovation traditionally supports productivity. However, the increase in the frequency of innovation has not been matched by an acceleration of productivity. Where does this paradox come from?

This paradox comes from the fact that although the frequency of digital innovation is higher, these many daily changes (in our mobile phones for example) are often marginal.

Historically, increasing productivity has always been a function of innovation and its diffusion, which should not change with digital innovation. It is always a question of increasing the capacity of production, from a given number of factors, or of increasing the diversity of production and its adaptation to needs. But on a more concrete level, based on the official statistics of countries, there has been no acceleration of productivity, despite the acceleration of digital innovation. As measured, productivity has even slowed or stagnated in OECD countries.

There are three types of explanations.

  • We do not measure productivity well. National accounts do not integrate digital products well, in particular because they are free (e.g., research on search engines or participation in social networks). Despite their success, social networks are not counted as an increase in production; they are priceless and their use is not counted as an economic transaction.
  • Investment slowdown. Although investment has recovered well, its level does not reach that of the pre-2009 crisis. Thus the accumulation of productive capital in OECD countries is lower than before.
  • Mechanism of learning by doing. Learning a new technology takes time, whether it involves business, individuals, society, or modes of government. The effects of these innovations on productivity are therefore not immediate.

There is no unanimity among economists concerning the mechanism at play. Depending on which one you choose, you will end up with different diagnoses and predictions about the future of productivity. The third explanation is very optimistic about the future. The second focuses on macroeconomic solutions, such as investment stimulus programs. If we insist on the first, we must give statistical departments the means to better measure productivity.

3) You spoke about the fluidity of the economic activity generated by digitization. Can you tell us more?
When considering the impacts of digitalisation on competition and income distribution (between individuals, companies, regions), two characteristics are essential: the fluidity of data and their complementarity with human capital.
Digitization is democratizing access to resources of knowledge and production that has not been anticipated until recently. It is now possible to produce a good in a secluded corner of the world while having access to a global market through platforms such as Amazon.
Added to this is the fact that the capital cost of a digital enterprise (software, cloud) is generally lower than that of a manufacturing company (machines) and that the conditions of entry to the markets are more open. These developments are in line with an increased competition and represent therefore, a priori, a more equitable distribution of economic activity.
However, there is also a factor that goes in the opposite direction: Information previously only accessible locally is now accessible globally, and the people or entities that are able to exploit them are not necessarily close; they can be very far from the place where the data was generated.
This is where complementarity with human capital comes into play. Human capital is subject to externalities that are created by the co-location of individuals. When two skilled individuals work together, their production is usually greater than it would have been if everyone had worked on their own side. There is therefore interest in individuals getting together. Because of externalities, large, centralized entities (such as large cities or corporations) are generally more productive than smaller entities, and make more efficient use of information. It can therefore be expected that the productivity of large centralized entities will increase relative to that of smaller ones.

So, what initially looked like democratization, can lead to the opposite: equal access to all data does not imply that everyone is equally qualified to create value from the data. And if there is a limited market place, it will be occupied by the larger players, who are the most efficient.
There are two facets to digitalization: both a democratization of market entry conditions and access to data, but also a tendency towards concentration and a "winner-takes-all" attitude. This has, as a consequence, a more unequal distribution of income between individuals (as found in all OECD countries), a concentration of market structures, an increase in inequality between firms, and between cities and rural areas, or between cities that are centers of excellence and those that are not.

However, economic policy can tip the balance on one side (concentration) or on the other (democratization of information). For example, it might encourage education in rural areas to try to compensate for the lack of human capital, or else develop an adequate support policy for the digitalization of SMEs to try to counter inequalities between companies.

4) There is a lot of discussion about taxing the GAFA. Why is our tax system poorly adapted to the challenges of digitalization and what solutions should be put forward?

Equitable taxation of all is one of the foundations of democracy. In a globalized world, things were already complicated (companies playing tax competition between states for example). With digitalization, things have turned to the detriment of states, namely the ability of companies to relocate revenue from where it is generated (and therefore ideally taxable) to places where taxation is more favorable (tax havens).

With the GAFA (Google-Amazon-Facebook-Apple), the foundations of the international tax system, namely that taxation must take place where the income is created, are questioned. Determining where income is generated is not obvious in the case of digital businesses. For example, income can be generated in Switzerland but economic activity is managed from abroad.
An agreement between the OECD countries and beyond is currently being negotiated with the aim of reducing this international tax distortion and allowing a minimum level of taxation to be accepted between the signatory countries.

5. What are the main challenges of economic innovation policy created by digitalization?

First, innovation policiy must evolve to adapt to current realities. Many countries have taken an innovative approach, creating "sandboxes": they develop small-scale and experimental digital and innovation policy (on a regional scale, for certain actors, or on certain conditions). If the procedure works, we generalize it, otherwise we stop it quickly.
Second, data policy is becoming an essential component of innovation policy, focusing on access conditions and data production conditions, as I have already mentioned.

Governments must push to facilitate access to (especially public) data, as this also facilitates innovation. But it must also be considered that the data, before being distributed, must be produced, which implies a cost. And this cost must be recoverable in one way or another, when the entity that generates it is a private company.

In some cases data access policies must recognize a certain conditional, temporary exclusivity. Full transparency and unrestricted sharing of data would pose significant ethical and economic problems. For example, with respect to scientific research data, governments must ensure that publicly funded data is accessible as widely as possible. However, the natural tendency of a researcher is not to share his or her data with other researchers, to maximize their chances of accessing competitive public funds such as scholarships. How can we solve this problem? Different conditions and appropriate incentives can be imagined in which the data are shareable: the data are disseminated within 6 months of their exploitation in a publication; the data are considered a final product of the research (and not just an intermediate product) and this could be written on the researcher’s CV, as a publication of articles. Efforts are being made in some countries, and this requires cultural change among researchers.

6. What about the digitalization of SMEs?
Digitalization is both an obligation and a risk for SMEs. An obligation because the competition is doing it, so to survive, we must also follow this path. But when it comes to adopting radical technologies (Internet of Things, 3D printing, artificial intelligence, etc.), this requires not only changing technologies, but sometimes even the business model, which constitutes a major risk for an SME.
A big company can take this risk, because if one division misses its chance, another one will succeed, which will limit the damage. If an SME misses the boat, it is often the very existence of the company that is at stake. Digitalization is expensive and complicated to set up. Many SMEs try to delay the deadline because their access to capital is difficult or internal skills are lacking. According to OECD surveys, asking a bank for funding to digitalize is an experiment often doomed to failure.

Public authorities may have a role to play in facilitating access to capital by ensuring access to credit or providing companies with direct or indirect expertise that they do not have in-house. In Germany or the Netherlands, there are various technical centers such as the Fraunhofer institutes that advise SMEs in this area.

Author: Alexandra von Schack