Technology Foresight in Japan
- The Potential and Implications of DELPHI Approach -

Abstract

  Many government organizations have conducted technology foresight in Japan. A consensus process was important to form national science and technology policy because administrative structure was decentralized and the industry had a large share of investment. This paper outlines Japanese technology foresight activities including Science and Technology Agency's Delphi that has a history of 30 years. Delphi approach's reliability is examined by chronological evaluation and international comparison. As for analysis of political implication, an importance index of Delphi is introduced and compared with a level of national R & D expenditure. Case studies on IT, environment, energy and life science show so significant correlation between two parameters that Delphi results could be utilized for priority settings.

1. The Characteristics of Japanese Science and Technology Activities: The Factors Behind Technology Foresight

  The goal of technology foresight is to provide basic information for use in government-level science and technology policy, corporate R & D management and so on. The implementation of technology foresight is inextricably related to the characteristics of the governmental organizations that are involved in science and technology and to various social and economic factors current at the time of implementation.

  From this perspective, one of the distinctive features of Japanese science and technology administration is its distributed ness in that many government ministries and agencies have their own research institutions and budgets, with government R & D programs as a whole implemented through their coordination. More specifically, while the Ministry of Education, which oversees universities and colleges, has been allocated a budget for the promotion of basic research, ministries such as the Ministry of International Trade and Industry (in charge of the industry as a whole), the Ministry of Health and Welfare (in charge of health/medical care and welfare), the Ministry of Posts and Telecommunications (in charge of telecommunication and broadcasting), and the Ministry of Agriculture, Forestry and Fisheries (in charge of agricultural, forestry and fisheries industries) have their own research budgets and research organizations to achieve their administrative goals. In addition, the Science and Technology Agency undertakes strategic research programs that bridge basic and applied research. The government's 1999 science and technology budget, which totalled 3.16 trillion yen, was distributed among ministries and agencies as follows: the Ministry of Education 42.7%, the Science and Technology Agency 24.5%, the Ministry of International Trade and Industry 16.1%, the Defense Agency 4.6%, the Ministry of Agriculture, Forestry and Fisheries 3.5%, the Ministry of Posts and Telecommunications 2.4%, the Ministry of Construction 1.3% and other ministries and agencies 2.4%. As an organization to deliberate and decide on a general program for R & D activities undertaken by these ministries and agencies, the Council for Science and Technology has been set up, with the Prime Minister as chairman¹.

  Japan's science and technology is also characterized by industry's proactive stance toward investing in R & D and by the fact that government research spending as a percentage of the nation's total research expenditures is low in comparison with the U.S. and Europe. For instance, in 1998 government spending accounted for 21.7% of the nation's total research expenditures of 16.1 trillion yen. This percentage was only 17.9% in 1990, but has been rising ever since.

  Against a background of the fact that many ministries and agencies maintain R & D functions within the government and that the industrial sector weighs heavily in R & D, a variety of long-term technological forecasts or visions have been prepared in Japan, centering around the government sector, to provide a direction for R & D through government policy formulation.

2. The Composition of Technology Foresight Activities and Delphi Survey

  The OECD defines technology foresight as "the process involved in systematically attempting to look into the longer-term future of science and technology, the economy and society with the aim of identifying the areas of strategic technologies likely to yield the greatest economic and social benefits." Japan is engaged in myriad technology foresight activities that are consistent with that definition, and which can be categorized into the following four types.

  Examples of macro-level activities are the various technology forecasts and social visions that are formulated and released by government agencies. An example of meso-level activities is the technology forecasting that industry organizations carry out with the cooperation of related corporations. The micro-level technology forecasting activities carried out by individual corporations are considerable in number and are not normally made public.

  Among these activities, those occupying the position of holistic-level endeavors are the Technology Forecast Surveys of the Science and Technology Agency. These surveys were begun in the 1970s and have been implemented every 5 years thereafter. The most recent survey is the Sixth, the results of which were released in June 1997[3]. These surveys cover a forecasted period of 30 years and have consistently employed the Delphi method. Although this technique was developed by RAND in the early 1950s, it was Japan that devised and established the approach of a large-scale survey covering every field of science and technology and involving thousands of experts.

  For the Sixth Technology Forecast Survey, a steering committee was established to direct the overall activities, and under it were established thirteen sub-committees; the leaders of these sub-committees comprise the membership of the technology forecast committee. Both the number of topics and the number of fields have increased over the years (Table 2). The National Institute of Science and Technology Policy - which implements the actual surveys - consults with various expert groups and others before selecting the fields to survey and appointing the leader of each sub-committee.

  These surveys are funded under a policy-proposal survey budget that is administered by the Council for Science and Technology, to which the survey results are submitted. Because they are so unprecedentedly large in scale and implemented on a regular basis, these surveys provide basic data that are used in the formation of science and technology policy by the Council for Science and Technology and by other government agencies as well.

  The 6th and most recent survey, which examined 1,072 topics in the 14 fields listed in Table 3, was begun in 1995. The Science and Technology Basic Law was passed by the Diet in December of that year, and under it an action program for the ensuing five years was decided by the cabinet in June 1996. This plan, called the Science and Technology Basic Plan, clearly set forth the government's policy of increasing R & D investments. The Sixth Survey was designed in accordance with the Science and Technology Basic Plan, and assessing the technological topics according to contribution to social and economic development, improving the quality of life, and solving environmental problems, for instance, also includes questions about the role government should play in realizing each topic (Appendix 1).

  As for a political application, for instance, a working group established by the Council for Science and Technology in 1999 used the survey results in its preliminary investigations for the next Science and Technology Basic Plan (which covers a five-year period beginning in 2001).

  Delphi survey results are also widely used by industry. The National Institute of Science and Technology Policy followed up on the Sixth Technology Forecast Survey with a questionnaire-based survey primarily of corporations that had been verified as having purchased and used the survey report. The 175 responses that were received (out of 429 questionnaires sent out) indicated that the results of the Technology Forecast Survey are widely used in the areas of research planning, research and development, and business planning. The objectives for which respondents used the survey results (including multiple responses) included the drafting of R&D and technology development plans (63%), the drafting of business plans (33%), and the administration and promotion of R&D and technology development plans (26%). Sixty percent of respondents used the result to determine trends in a specific technology field, while 40% expressed interest in a wide range of technology trends. As for the time axis of the forecasts, 59% of respondents were interested in medium-term technology trends (5 to 10 years into the future), 22% in long-term trends (more than 10 years into the future), and 18% in short-term trends (less than 5 years into the future). Regarding the usefulness of Delphi survey results, 17% found them extremely helpful, 70% somewhat helpful. This assessment of usefulness most likely reflects the fact that although the majority of respondents were interested in trends less than 10 years into the future, 80% of technologies examined by the Delphi survey were those forecasted to be realized between 10 and 20 years into the future. Reflecting the increasing importance of technology forecasting is the percentage of respondents who see technology forecasting information in general as either highly important (64%) or worthwhile (33%), and who believe such information to be more important than before (59%) or roughly as important as before (31%).

3. Assessment of Delphi Surveys

  First, let us examine how the realization of past forecasts was assessed. In 1996, members of the Sixth Technology Forecast Survey subcommittees assessed 588 (of 644) and 549 (of 657) topics, respectively, of the first and second surveys, respectively carried out in 1971 and 1976. Of the assessed topics in the first survey, 26% are "realized," 38% "partially realized. By division, information has the highest realization rate, followed by food and agriculture, industry and resources, health and medical care, and social development. The realization rate including partially realized topics is highest in health and medical care, followed by food and agriculture, information, industry and resources, and social development (Appendix 2).

  Similarly, of the topics assessed in the second survey, 21% were deemed "realized," 42% "partially realized," and 37% "unrealized." Fields with a high realization rate are space development, information, industrial production, family life, and food resources, while those with a low realization rate are water resources, software science, transportation, environment, and forest resources. The realization rate including partially realized topics is high in space development and health and medical care, and low in software science and energy (Appendix 3).

  It should be noted that this assessment considered only whether a technology was deemed "realized" as of 1996; it did not assess whether past forecasted realization times had been accurate. Further, because some technology topics included multiple aspects and/or had ambiguities, the figures for "partially realized" include not only cases in which a technology had been partially realized, but also cases in which the experts disagreed on how to interpret the aspects of a technology. Nonetheless, realization rates including partially realized topics indicate that roughly two-thirds of the technologies are progressing in the manner forecasted - quite impressive for forecasts so long-term in scope (30 years). It should also be noted, however, that there are technologies that had significant impact upon realization but which had not been addressed by the forecasts.

Relationship between degree of importance and realization rate

  An " importance index " was calculated for each technology topic to elucidate the relationship between assessed degree of importance and realization rate. The first through sixth surveys assessed each topic's degree of importance of by asking respondents to select one of four choices: high, medium, low, or unnecessary. The index was worked out from the following equation; the index is 100 when all respondents indicate " high " and 0 when all indicate " unnecessary ".

importance index = (100*Nh + 50*Nm + 25*Nl + 0*Nu) / (Nh + Nm + Nl + Nu)

Nh = number of " high " responses

Nm = number of " medium " responses

Nl = number of " low " responses

Nu = number of " unnecessary " responses

  While there is little difference between topics with a high degree and those with a low degree of importance in the realization rate, there is a significant difference in the realization rate including partially realized topics, with the more important topics showing a much higher rate. Among topics with a degree of importance index of 50 or more, there is no major difference, and for topics with a low degree of importance, the realization rate including partially realized topics is quite low indeed.

Forecasted realization time and realization rate

  Here we classified the topics by forecasted realization time, and calculated the realization rate for each time classification. In both the first and second surveys, the earlier the topic forecasted realization time, the higher the realization rate and the realization rate including partially realized topics. Moreover, topics that recorded a high percentage of "will not be realized" responses have an extremely high unrealized rate.

International Comparisons of Delphi Surveys

  Although the first through fourth surveys of the Science and Technology Agency limited themselves to Japan, the surveys have become more and more international in scope since the fifth survey, which cooperated with Germany to produce Japanese-German comparisons and a joint Japanese-German "Mini-Delphi" survey. France also followed by forecast survey of identical topics with Japanese fifth one.German Delphi I and Japan's fifth survey examine identical topics and so permit a direct comparison of results for both countries. As shown in Fig.1, both countries' forecasts correspond quite well with respect to forecasted realized time. Each point in Fig.1 represents a topic. The difference between Japanese and German forecasted realized times was less than three years for roughly 60% of all topics. This 60% figure can be interpreted as indicating nearly complete agreement given the fact that respondents' choices for realized time were five-year time periods. Furthermore, because this survey asked respondents about "realization in your own country," such excellent agreement in results serves as a reminder of the global scope of technology. In contrast, considerable differences exist between the two countries' assessment of the importance of topics, indicating the strong effect of social, economic, geographic and cultural factors.

  Fig.1 Comparison of the Japanese and German time of realization of all the topics

4. The Political Implications of Technology Foresight

  Because Japan's technology forecast surveys thus involve such a large number of leading experts and employs the Delphi method, which is designed to reach a consensus among a group of experts, how such consensus relates to policy formation and R&D activity, for instance, is a highly intriguing proposition.

  Let us therefore consider political implications in Delphi-based technology foresight in comparison with past trends in actual R&D expenditures. Investment in R&D by Japan's public and private sectors is monitored by the Management and Coordination Agency through surveys that determine R&D expenditures by objective, and so make it possible to trace changes in Japan's overall research spending in energy, information technology and several other categories. Table 7 gives each category's share of Japan's total R&D expenditures in each survey year.

  These data show that environment-related research expenditures, for instance, accounted for 3.5% of the total in the 1970s, when pollution control was an urgent domestic issue. Effective pollution-control measures allowed this percentage to drop to roughly 1.6% in the 1980s, but it rose again in the 1990s with heightened awareness of global environmental problems.

  The importance indices of topics technology forecast survey were used to create an "Delphi index" showing the degree of priority of technological area by selecting the upper 30% of the topics in importance index and by grouping these topics according to their subject matter into the six categories of "information," "life science," "environment," "energy," or "other," and then determining the ratio of each category. This ratio became the Delphi index for its respective category.

  The ratio for the category of "environment" was high in the 1970s, declined in the 1980s, and once again exhibited a rising tendency in the 1990s, thus corresponding closely to the aforementioned trend in environment-related research expenditures. Graphs 1 through 4 visually indicate the correlation of the Delphi index in four categories to Japan's actual total research expenditures. A close correlation is seen between the two for the three categories of the environment, information technology, and energy. Research expenditures for the life sciences, however, have been included in the survey only since 1982, making comparison difficult and resulting in trends that differ someone from those of the other three categories. Furthermore, because this is the first attempt at the systematic analysis of Delphi forecasts over an almost 30-year period, analysis of further depth is required. Nonetheless, these results demonstrate how technology forecasts arrived at with the participation of numerous researchers and engineers from industrial and other sectors are an accurate indicator of national trends in research and development in Japan.

  Fig 4 Energy

  Fig.5 Environment

  Fig.6 Information

  Fig.7 Life science

5. Conclusion

  In Japan, technology forecasts that are based on the participation of and consensus formed among large numbers of experts have followed a unique path of development owing to the unique characteristics of Japan's R&D infrastructure. Historical experience has shown this approach to have considerable reliability, while its connection to the evolution of macro-level R&D activity is also evident from the correlation with actual research expenditures. In an international context, this approach has also been adopted in other countries beginning in the 1990s, and comparative research has revealed that the nations of the world now share a common awareness of the future outlook of technology. At the same time, however, considerable differences - originating in each nation's social and economic circumstances - have also been quantitatively shown to exist with respect to how each nation assesses the importance of various technologies. Such knowledge bespeaks not just the need for efforts to increase the depth and validity of technology foresight, but also the potential benefit of such efforts. NISTEP has begun the Seventh Technology Forecast Survey, and intends to develop a new methodology through it.

Contents|Next