Going Digital -- Insights and policy implications from OECD work on digital transformation

Date March 7, 2018
Speaker Dirk PILAT (Deputy Director, Directorate for Science, Technology and Innovation, OECD)
Moderator IKEUCHI Kenta (Fellow, RIETI)

This presentation will provide an overview, key insights and policy implications from the OECD Going Digital project, which is a horizontal initiative across the OECD (involving all key policy areas), mandated by Ministers, to: a) Better understand the digital transformation and its impacts on economy and society; b) Provide policy makers with the tools needed to develop a pro-active, whole-of-government policy response; c) Help overcome the gap between technological change and policy development.



Dirk PILAT's Photo


The Organisation for Economic Co-operation and Development (OECD) is engaged in a project titled, "Going Digital." First, I want to briefly talk about the digital transformation itself, what it means, and why it is special. I then want to focus mainly on productivity, jobs, and skills. These are two of the main policy issues being discussed, but many more issues exist.

First, we are now in a new phase of the digital economy. Over a very short period of time, we have seen the emergence and rapid diffusion of smartphones of enormous power. A smartphone has the processing power equivalent of a mid-1990s supercomputer. This is only one element of why we are in a different phase. New platforms and business models such as Uber are coming into existence, enabled by new technologies such as smartphones.

Second, a very wide range of technologies are currently emerging. Artificial intelligence (AI) has been talked about for a very long time, but now it actually exists and is being implemented and used by firms. We see that with other digital technologies as well. Many changes are occurring simultaneously.

The third factor is that every part of the economy is being affected. In the 1980s and 1990s, we saw technologies such as ATMs and barcode scanning, which had a widespread impact on financial and business services. Today, previously unaffected areas are also being affected by digital technologies. A massive potential for change exists across the entire economy, even in sectors such as government, health, agriculture, science, and education. This means that not only do we have great advancements in the power of computing, but also a wide range of technologies emerging very quickly that are affecting a wide range of economic sectors.

The OECD has had policy committees looking at many of the issues related to the digital economy for about 30 years. However, a broader approach is now required because we need to think more strategically about how we can make this digital transformation work for everyone. Some critical thresholds have been crossed in terms of the speed and scope of the changes that are happening. We are also moving to more of a socioeconomic focus in the sense that every sector of the economy is being affected. This transformation offers enormous potential improvements in productivity, employment, and service quality, among other things, but we need to make it work. This is transformative, disruptive change.

Gap between technology and policy

We see a gap in many countries between technology and policy. The technology has been moving so fast in recent years that it has been very difficult for policymakers to keep up. What can we do to reduce that gap? Can we provide better tools for policymakers to make sure that we can deliver on making this work for everyone? The OECD is engaged in a broad, organization-wide collaborative effort across all of the different policy areas to reach a better understanding of what this digital transformation means for the economy and society, and also provide better tools to policymakers to enable them to be more proactive and forward-looking, etc.

In our work, we have been trying to determine what policy areas we need to look at to make this work. Framework policies, trade policies, tax policies, and structural reform policies related to regulation of economies are all being affected by the digital transformation. We need to determine how they need to change to make the transformation work.

A number of issues emerge from our work. The first is access: can we make sure everyone has access to these technologies and can seize their benefits? Many countries have problems with this, particularly countries with remote geographical regions or certain groups that find themselves on the wrong side of the digital divide. Small and medium-sized enterprises (SMEs) are also not yet making sufficient use of digital technologies. This raises many questions about telecommunications policy; about whether countries have broadband policies in place that reach everyone, etc. The second category of issues involves use of the technologies: the set of policies that need to be put in place so that once we have provided the technology to people and firms, we can make it work to their best advantage. This includes policies to create trust in the technologies (i.e., with regard to privacy and security), ensure that firms can effectively use them, and ensure that people within firms have the necessary skills and management abilities, etc. The third issue is digital government: applying technology to the delivery of policies by the government, etc. The fourth and final issue is that of the overall strategy for digital transformation by which we make all of this work together.

The OECD is doing a great deal of work on specific issues in a wide range of projects. For example, we issued a report on road freight transport and its potential for automation. Other reports have been issued on energy, competition, taxes and the digital economy, etc. We are focusing on four different issues in particular, two of which I will talk about in greater detail. One is jobs and skills. Many countries have concerns about the employment impacts of the digital transformation: the possibility that we are moving to a "world without work," as some think, and the types of skills that will be needed in the future. The second issue is productivity. The third is wellbeing, partly because the impacts of this digital transformation go beyond gross domestic product (GDP) and affect many elements of wellbeing. The fourth is measurement. We have little data on a great many areas related to the digital transformation. We lack evidence of what is really happening, and we need to develop better data and indicators to better understand what is going on.

Productivity growth

Over the past two decades, productivity growth has been slowing down in many countries in most regions around the world. A huge debate is underway about this, with pessimists and optimists. Robert Gordon, a professor at Northwestern University, holds that many things that can be invented already have been invented, and that nothing as transformative as electricity or the steam power revolution is happening today. In this more pessimistic view, we will only see further incremental change and not the same rate of productivity growth seen in those other revolutions. More optimistic views, such as that from Andrew McAfee believe that we may be in a slower period, but emerging technologies, etc. offer a massive potential for future productivity growth. So what does the evidence tell us about where we are in this debate?

One of the contributions we have made at the OECD is to try to break down what we are observing in terms of productivity growth. I know RIETI is working on this as well in the Japanese context. What we have found is that if we look at productivity growth at the firm level in different countries across the OECD and in different industries, we find that the frontier firms—the most productive firms—were still seeing productivity growth even after the crisis, but the rest of the economy was having much greater difficulty achieving the same rates of productivity growth. A group of firms is doing extremely well, but many firms at the bottom are not really keeping pace in terms of productivity growth. We think that this may be linked to digital transformation. Some firms see the potential for the technology and have the management, skills, etc. to make it work by reorganizing work processes, implementing new business models, and others. Many others lack some of the key capabilities needed.

A few issues are relevant to Japan. One is the way in which technologies are being implemented in various countries. Most firms in OECD countries now have access to broadband networks. However, if we look more closely, they are not yet using very advanced technologies that much. Looking at big data, for example, very few firms are using it right now. The same is true of AI. Huge differences exist from country to country. The diffusion of these digital technologies across firms is still in progress. Japan is making progress in some areas, such as cloud computing, where it is among the most advanced countries in the G7 in its use by firms compared to countries such as Germany and France. SMEs are lagging everywhere. Cloud computing technology in principle would be useful to SMEs by enabling outsourcing to other firms via the cloud. However, large differences between the largest and smaller firms exist. This is particularly the case in some European countries. Germany has very little use, but Finland has a high level of use.

We are also looking at a number of indicators to determine the intensity of IT use by industry to understand how digitalization is progressing in different sectors. Digital intensity undoubtedly is very high in sectors such as telecommunications and finance. Other sectors still have low intensities: agriculture, mining, transport, hotels and restaurants, education, and health. These large sectors also need to go through a digital transformation, after which we will probably see the impacts much more clearly. It is therefore important to realize that the diffusion of technology is still ongoing. It will take more time, particularly for SMEs and in certain sectors of the economy.

It's also not just about technology. Technology is advancing rapidly, but within firms, the organizational changes, introduction of new business models, and the training of workers and management in needed skills, etc. are not. When new technology is introduced and new businesses emerge, some firms will make it work and others won't. The biggest impacts emerge from firms that learn how to use technology, which then grow and gain market share. Other firms don't grow and lose market share. Because of these shifts between firms, structural economic changes occur along with the digital transformation.

Policy can make a difference in making these changes work. We are optimistic that digital transformation will help strengthen productivity. A recent report by McKinsey and Company looking at these issues also stated that we should expect productivity impacts as a result of technology. A few things can be done, however, that would probably help. For example, we need to diffuse this technology as quickly and widely as possible. We need to strengthen skills and promote structural change. We need to make sure that sufficient competition exists in the economy.

Jobs and skills

On the second issue—jobs and skills—we are currently seeing three megatrends affecting employment in many different countries. The first two are technology and aging. Japan is among the countries most affected by the aging of its population, and fewer younger people will need to take care of many more older people. The third is globalization.

Labor markets are changing. How many jobs in different countries are at risk of automation? We estimate that approximately 15% of jobs in OECD countries are at high risk and could be automated within the next 10 to 15 years. We expect another 30% or so of jobs to undergo significant change over the same period. They might not be fully automated, but they will likely change, because some tasks will become automated in these areas. The second area in which work is changing relates to new forms of work. New platforms enable people to provide their services directly. Uber is an example.

We are also seeing polarization in the labor market. We have seen growth in high-skilled jobs and significant growth in the number of low-skilled jobs (care services, cleaning, etc.) and a drop in the number of middle-skill jobs. We have also seen growth in inequality within many OECD countries; while it is not happening everywhere, it is happening.

In Japan, the number of jobs at high risk of automation is actually quite low—about 10%-12%. The number of jobs at significant risk of change, in the sense that they are being affected by new technologies, is higher. Differences between countries result partly from sectoral differences. Some countries see this as a challenge, while others see it as an opportunity. The risk of automation means that jobs could be replaced which are very unhealthy, very unsafe, and in the context of an aging population, automation may not be such a bad thing. Concerns remain, however.

At the same time, a recent article in the Wall Street Journal looking at the number of jobs that have emerged in U.S. retailing showed that more jobs had been created since the crisis from electronic commerce than had been lost in traditional retailing. We also know that many of the occupations we see now did not exist 10 or 15 years ago. Many new jobs are emerging all the time. New markets, new opportunities, and new areas of demand imply new jobs.

At the moment, we are not overly concerned about all jobs being automated. Many OECD countries are currently close to full employment. The real issue is the polarization in skill demands. The shifting composition of jobs needs to be emphasized. Many people—particularly those with low levels of education—often do not have the right skills to apply for a new job. People with higher levels of education, by contrast, often have more of the right skills. In Japan, many people who actually use software in their jobs say they don't have the right skills to use it. Many don't have the right skills for the future. So far, people in the middle-skill-level have been losing out. In the future, we think more low-skilled people will lose out because of developments such as AI. They will have more trouble moving into new occupations. Risks exist for young people, for men and women, and for older workers which differ by sector.

The skills level of older workers in technology-rich, problem-solving environments is also a concern. Future workplaces will demand that workers have the right skills for jobs that involve the use of technology and require problem solving. We find that younger people in many countries have a reasonably high level of skills. In Japan, about 60% of people have a good level, which means they can compete. However, for older workers in Japan, only 10%-15% of workers have sufficient skills. Some of these people will be affected by automation, e.g., older truck and taxi drivers, and it will be difficult to give them the right skills for the future. That is a concern.

We also see gender differences. This can be partially accounted for by differences in the sectors in which the two genders work. Far more men work in sectors such as transport, manufacturing and construction, where workers are at high risk of automation. Retailing and restaurants are sectors in which more women work, and have even higher risk levels.

One of the areas of concern is also whether people have the foundation for continued learning. Do they have the skills to learn in the future? The skills people have upon completing their education will not be sufficient. They will need to change jobs and learn new skills. The question is whether they have basic skills such as literacy and numeracy. Japan does the best here compared to all other OECD countries. In other countries, as many as 50% of people do not have these basic skills, which makes it much harder to give people new skills. We can also see who is receiving retraining. Unfortunately, the people with the lowest levels of skill also receive the lowest levels of training. The people with the highest levels of skills also receive the most training, which is an important policy challenge because the most likely to be affected by digitalization are people with the lowest level of skills.

We also need to look at how technology and skills complement one another. The question of what skills we will need in the future is a real concern. The more technology we introduce into the economy, the greater the changes in the types of jobs available and the less routine the work become. Countries with a very high intensity in information and communications technology (ICT) and digital use also have more jobs that are not routine and that involve more complex tasks. We need to find ways of ensuring that we have skills that complement technology. We don't want skills that can do the same things that technology can do because the technology is moving so fast. We need programmers, but we also need special skills that only people can do—softer skills. These may be more important in the future than they have been so far.

Most of the employment issues are skill-related. We have been talking about lifelong learning for a long time, but it needs to be implemented urgently now. Also, most countries have tried reforming their education systems for some time and have found it to be very difficult. Regulatory issues exist also in terms of how to deal with the changes in the labor market, with social protection, and social dialogue issues as important challenges. At the OECD, we are working on recommendations to help policymakers with this.


Q1. What are your thoughts on changes in productivity on both the supply and demand sides in relation to digitalization?

I phrased the changes in terms of digitalization. However, other cyclical factors exist as well. Since the economies of many OECD countries are approaching full employment, etc., I think more firms will start to focus more on productivity improvements. The productivity slowdown is also linked to lower capital investment, which is also partly a cyclical process. During an OECD conference in January 2018 attended by the International Monetary Fund (IMF) and others, we discussed the fact that the monetary policies of many OECD countries since the crisis have enabled many firms to survive that might not have had they not been provided with relatively cheap capital. We have seen a growth in the number of "zombie firms," or nonviable firms. This may also be linked to monetary policy. Many factors exist that make it difficult to untangle all of this, however.

Q2. What is the impact of the "Going Digital" project on the OECD? What are the internal difficulties?

The OECD works through different committees that work on different policy areas. We work with different ministries. I think what we are seeing with regard to this project is that we have the largest number of OECD committees working together on this project. Everyone realizes that this is affecting their policy areas. That was probably not the case three or four years ago. In finance, for example, cryptocurrencies have seen increased interest in digital technologies. We need to understand the technologies. There are also some cross-cutting policy issues affecting everyone, such as security, privacy, and the role of data. We are also working to make sure that the OECD speaks with a more unified voice on some of these issues.

Q3. I formerly worked for the OECD. The biggest issue then—from the late 1990s to the early 2000s—was the "New Economy." I find that we currently share some policy issues also faced at that time, like the need to invest in human capital and the diffusion of ICT. What is the biggest policy challenge of this economic era compared with the "New Economy" era?

Some issues that make good policy are omnipresent and enduringly important. Sometimes they appear in different forms or under different names, but things like skills and education are always important. Since technology is changing so quickly and putting new demands on people, I think this will become even more important than it was before, as will the diffusion of ICT. A new paper looks at AI and productivity at firms, and states that the range of investments that firms need to make is increasing, not only in technology but also other things needed to make it work, such as skills, management, organizational capital, branding, innovation, etc. It may be becoming more difficult; perhaps you need to invest more and more and it becomes increasingly complex, making it harder for firms to convert this into productivity. SMEs can now be "born global." However, we see very few SMEs making this work at the moment. How the diffusion works is another question. Does it favor certain firms? Our microdata work suggests that some smaller firms are making this work but these mostly have new business models, new skills, new talent, etc. We see opportunities and potential benefits, but challenges exist here as well.

Q4. I was surprised by the very high utilization rate of cloud computing at Japanese firms, even in small- and medium-sized firms. I cannot imagine that 40% of SMEs in Japan use cloud computing. What kinds of services are included in this?

It comes from your own data, from statistical offices. Some of the differences result from concerns in some countries about data security and so on. We need to do more detailed research on cloud computing, but the country-to-country differences are large. Perhaps Japanese firms are using Japanese servers that they can trust while some European firms worry about non-European providers, etc.

Q5. Most participants likely share the view that the digital economy will have a positive effect on productivity. However, we would like to know the quantitative impact. What types of data collection are important to enable estimation of the contribution of these new technologies to productivity growth?

If we really want to understand the impact of specific technologies on productivity, we must first collect data on those technologies and then find their relationship to productivity. In past studies, we found a positive impact on productivity growth, but it was often combined with other factors. For example, firms investing in technology were also investing in skills and/or engaged in organizational change and investing in innovation. It is often a combination of things. Technology is a tool, but the tool only works if you do a few other things. That makes it hard to respond to your first question. If we give numbers, we are attributing it all to the technology. It's a range of complementary investments. Firm-level studies exist that give numbers, but that is because it is in a specific context. An aggregate estimate depends on individual firms and their various situations.

Q6. On the various industries and the extent to which they use digital technologies, agriculture, mining, and real estate make almost no use of them while telecommunications puts them to intense use. Is that a simple survey or does that have a correlation with productivity?

In a past study, we found that countries with high-ICT using and producing sectors had a greater impact in terms of productivity growth. Here, we wanted to look more broadly at how ICT is used in various sectors with a range of indicators. In Australia, mining is very ICT-intense, with autonomous digging. That's the difficulty of using a number of indicators across different countries. So many differences exist in different sectors and in different countries.

Q7. What do you think about the impact of the progress of digitalization and AI technologies on income distribution?

We are also looking at inclusive growth. Looking at the national level in many countries, we typically see an increase in wages in the more productive firms and no increases in the bottom firms. This is one factor contributing to the growth in inequality. There are other elements. We do not understand competition that well, and we need to look at how digitalization is changing the forms of competition. There is the "winner-takes-most" idea. There are links here, and we have to do more work to elucidate them, including through microdata. We also want to examine firm-level data linked to employee-level data.

*This summary was compiled by RIETI Editorial staff.