An alternative approach to technology policy assessment: dynamic simulation analysis of Taiwan’s IC industry

全文

(1)Int. J. Technology, Policy and Management, Vol. 6, No. 2, 2006. An alternative approach to technology policy assessment: dynamic simulation analysis of Taiwan’s IC industry Ting Lin Lee Department of Asia Pacific Industrial and Business Management National University of Kaohsiung 700 Kaohsiung University Road Kaohsiung, 811, Taiwan E-mail: Linda_lee@nuk.edu.tw Abstract: To investigate the progress of the IC industry and its determinants, the present research adopts a multidisciplinary approach, by combining questionnaire/in-depth interviews with System Dynamics simulation based on ten-year (1990–1999) primary and secondary material. In comparing the respective results, one objective is to increase our insight into the dynamics of National Innovation Systems (NIS) by means of computer modelling; another objective is to create scenarios to verify the behaviour of the institutions under investigation by simulation, and to assess possible outcomes in those varying scenarios. This research therefore centres on developing a mathematical model of Taiwan’s NIS, particularly with regard to its IC industry. The paper concludes with some simulations of policy alternatives confronting the industry and the Taiwanese government. Keywords: National Innovation System (NIS); Integrated Circuit (IC); System Dynamics (SD); vertically disintegrated. Reference to this paper should be made as follows: Lee, T.L. (2006) ‘An alternative approach to technology policy assessment: dynamic simulation analysis of Taiwan’s IC industry’, Int. J. Technology, Policy and Management, Vol. 6, No. 2, pp.121–153. Biographical notes: Ting Lin Lee is currently an Assistant Professor in the Department of Asia-Pacific Industrial and Business Management, National University of Kaohsiung, Taiwan. She received her PhD degree in Science and Technology Policy at PREST from the University of Manchester (UK) in 2002 and an MBA from the Sun Yat-Sen University (Taiwan) in 1997. She used to work for the Central and Local Government in Taiwan during the period 1987–2000. Her research interests include the areas of science and technology policy, innovation and R&D management, industrial policy, and the practical application to industry of System Dynamics, Social Network Analysis, and Cluster Analysis. Recently, she published papers in the Research Policy and Technology in Society.. Copyright © 2006 Inderscience Enterprises Ltd.. 121.

(2) 122. 1. T.L. Lee. Introduction. The IC industry is one of the strategic industries or technologies for whose development Taiwan’s government has exerted great effort. Since the foundation of UMC and TSMC in the late 1980s, Taiwan’s semiconductor industry has boomed in prosperity. The unique vertically disintegrated structure of Taiwan differs substantially from that of other countries. This vertical disintegration is undoubtedly advantageous to the development of a dedicated foundry industry, since it effectively lowers the production costs and shortens production cycles. In addition, this structure has successfully rooted its core competence in the semiconductor business with a strong arm in fabrication, and has fostered a prosperous IC design business at the same time. By 2003, Taiwan had become the fourth largest IC producer and the second largest IC designer in the world. Taiwan was the first country to set up chip foundry operations, and its share in the world market amounted to 70.8% in 2003, earning it the reputation of being the world’s No.1. Among the industrial sectors, the IC industry has become one of the most promising opportunities for growth. Mathews (1997) pointed out that Taiwan’s semiconductor industry’s growth and development can be viewed as an institutional framework or dynamic structure. This ecological system of semiconductor activity includes private sector firms, public sector research institutes and infrastructure, government regulatory and coordination agencies, and interorganisational structures. Therefore, the object of this case study is to show how a System Dynamics (SD) model can be constructed from this complex situation, and to see what light can be shed on the issues involved. In particular, we wish to show how actors involved in the NIS interact to produce performance over a substantial period (1990–1999), and how robust that performance will be. System Dynamics is recognised as an approach to studying the behaviour of complex systems: it aims to demonstrate how policies, decisions, structures and delays are interrelated and influence growth and stability. The interrelationships generate a structure, which in turn produces behaviour (indeed, there is an implied sequence of time, with behaviour being preceded by structure). Therefore, the purpose of applying SD is to facilitate understanding of the relationship between the behaviour of a system over time and its underlying structure and strategies/policies/decision rules. The model aids conceptualisation of the causes of the existing behaviour and the testing of new policy designs in an effort to improve behaviour. Furthermore, based on the model, we will go on to study policy design by changing the parameter values, or the structural and aggregation level of the model obtained from the in-depth interviews or questionnaire survey. By using computer simulation, the interactions and driving forces within sectors can be clearly observed. The rapid increase of the R&D budget, nurturing of S&T researchers and adjustment of product and process innovation give rise to policy testing. These simulations show that single policies are relatively ineffective and that innovation performance requires combining a range of policies and capabilities. Although the work could be further developed with enlarged resources and funding, this approach could constitute a basis for policymaking in science and technology policy that would give Taiwan a head start in terms of reliable assessments of alternative policy strategies. This is the major new practical contribution of this research..

(3) An alternative approach to technology policy assessment. 2. 123. The overview of the National Innovation System. The concept of the National Innovation System (NIS) was first introduced by Freeman (1987). However, Bengt-Åke Lundvall was the first person to use the expression ‘NIS’, as the editor of a highly original and thought-provoking book on this subject (Freeman, 1995). Many scholars have proposed different definitions of the NIS, while various models of the NIS have also been developed (such as Freeman, 1987; Lundvall, 1992; Nelson, 1993; Patel and Pavitt, 1994; Metcalfe, 1995; OECD, 1997; Edquist, 1997). There are differences in detail among these models in terms of the elements stressed and the interlinkages drawn (Lee, 2002). Besides this, the model can be different for different industries. With a different history, culture and endowment, the model for the IC industry can be absolutely different too. The NIS is a system of interaction among various actors (Smith, 2000; Mytelka, 2000; 2001; 2003; Fagerberg et al., 2005). The characteristics of the NIS have been reported frequently, which involve cumulative causation and feedback mechanisms of non-linearities and instabilities, due to its complex and multiplicitous status. As Lundvall (1998; 2000) proposed, innovation incorporates ‘interactive processes’ (including linkages and flows) where agents/institutions and organisations communicate, cooperate and establish long-term relationships. These ‘untraded interdependencies’, especially, have played a key role in explaining the rate and direction of innovation. Innovation is thus the result of a complex interaction between various actors and institutions (OECD, 1994). The fundamental mechanism lying behind such complex systems is instability, which in turn moving toward a multiplicity of status, including the possibility of deterministic chaos (Nicolis and Prigogine, 1994). As Lundvall (1992) emphasised, the NIS represents: “… a system of innovations constituted by elements and relationships … that the national system of innovation is a social system… It is also a dynamic system, characterised both by positive feedback and by reproduction… Cumulative causation, and virtuous and vicious circles, are characteristics of systems and sub-systems of innovation. ”. In these particular senses, the processes through which technological innovations emerge are extremely complex. The processes translate the emergence and diffusion of knowledge elements into new products, for which production has to be carried out. Moreover, this translation by no means follows a ‘linear’ path; instead, it is characterised by complicated feedback mechanisms and interactive relations involving science, technology, learning, production, policy and demand. These relations are often characterised by reciprocity, interactivity and feedback mechanisms in several loops. They are clearly not characterised by unilateral and linear causal relationships (Edquist, 1997; OECD, 1994; Mowery and Ziedonis, 1998; Balzat and Hanusch, 2004). Modelling the NIS may therefore need to take into account delays and hold-ups, or sources of inadequate response, in the national system. This will be an important aspect of this present study. The research shows that it is indeed feasible and fruitful to apply the SD approach to analysing National Systems of Innovation..

(4) 124. 3. T.L. Lee. The boundary of the Taiwanese IC innovation model. A fundamental problem confronting the analysis of NIS is the boundary, which might include all aspects of a country’s social, economic, political and cultural activities. As some have pointed out, the whole socio-economic system cannot, of course, be considered for inclusion in the NIS. Which parts then should be included in the innovation system? This focus is in line with recent work by Johnson and Jacobsson (2000, pp.3–4),1 Rickne (2000, p.175)2 and Liu and White (2001, p.1094).3 Edquist (2001, p.13) also presented the ‘functional’ approach to identify three main components/functions, referred to as the spatial/geographical, sectoral and functional in the boundary of innovation systems. In this sense, it is important for the innovation system to identify the boundaries/environments in which lie the components and relations between them. The innovation system adopted here includes five ‘sectors’ (or actors/sub-systems) in the Taiwanese IC innovation system: these are the Financial, Human Resources, Science and Technology Transfer, Innovation Commercialisation, and Product Market sub-systems. Government macro-economic policies and government legal and financial systems are treated as exogenous factors (see Figure 1), because much of them are matters of legislation, politics and externalities. To simplify the model, this study will not take the determination of these factors into account. Figure 1. The boundary of domestic IC industry in Taiwan’s NIS Government Macro -Economic policy Government Legal and Financial Systems Human Resources (HR) Manpower (University; Public, Private, or Quasi-government Research Institute ). Training RD capacity (Education, Experience,. Mental Skill Internalisation. Technology process capacity, Patents, Publications, Annual R&D expenditure per person). Manpower Prediction and Recruitment. Embedded RD Capacity Science and Technology (ST) Labor Cost. Innovation Commercialisation (I) (Product and Process Innovation). Product Market (M). Transfer Pattern (Spin-off, Joint Venture, Alliance, Licensing, Turn key etc ..). Knowledge and Skill. Knowledge Internalisation. New Innovation Capacity Process Innovation Product Innovation .. New Products. Market Share Sales profits. Equipment and New Plant Investment Technology Transferring Investment. Financial (Capital Market) (F) Government R &D Funding Private R&D Budget (Retained Earning) Investors (Venture Capital) Foreign Depository receipts (ADR, GDR, ECB). Profits.

(5) An alternative approach to technology policy assessment. 125. Figure 1 helps us to understand, among other things, the several interrelationships among the mentioned sectors. From the diagram, we can see three types of flows that will be crucial and which need to be included within the model: 1. 2. 3. the interlinkages within the model between the technology segment and the industry segment •. technology push from science and technology to industry. •. R&D in industry. the external sources of inputs (e.g., government funding of R&D, university supply of graduates, the media role of institutes) •. the inflow of skilled researchers/employees into science and technology and the innovation sector. •. the inflow of financial resources into each sector. the external sources of demand for outputs (e.g., structure of industry product markets) •. the outflow into products. •. the outflow into markets.. 3.1 Financial (F) sub-system The IC industry is a highly capital-intensive industry. The royalties also involve a huge amount of expenditure. This sector supports the essential ‘labour cost’ of the human resources sector while it invests in establishing and renewing essential equipment and new plants for the innovation commercialisation sector. In addition, it also provides ‘science capital and technology transfer’ to the science and technology transfer sector. Government funding is scattered over the science and technology transfer sector and the innovation commercialisation sector.. 3.2 Human Resources (HR) sub-system Human resources clearly contributed to the rapid development of Taiwan’s IC industry. Our goal here will be to describe the Human Resources sector’s providing ‘embedded fruitful knowledge and R&D capacity’ to the innovation commercialisation sector, regardless of embodied or disembodied technology know-how. It also provides the ‘mental skill’ to manage the staff of new organisations which spin-off from the research institutes, joint ventures, alliances and turnkey plants.. 3.3 Science and Technology (ST) transfer sub-system Science and technology transfer is achieved in two ways: one is by technology transfer out of universities and institutes (public or semi-public) into commercial use (private), and the other is from overseas to domestic use. There are several ways to absorb alien technologies (spin-off, joint venture, alliance, licensing, purchase of technology, turnkey, etc.). Thus the kind of approach to the transfer and the extent to which content is.

(6) 126. T.L. Lee. transferred are quite important. Nevertheless, how new scientific knowledge and technological skill transferred from outside becomes one’s own R&D capacity by way of internalisation processes is comparatively more important. For example, in the IC industry’s case, Taiwan’s technologies used in production were mainly transferred from industrialised countries such as the USA and Japan through turnkey contracts, licensing and direct investment (Xue, 1997, p.747). Since the IC design and processing technology of Taiwan lagged far behind advanced countries, technology transfer was adopted as an initial strategy for the quick development of the industry base at the very beginning. ITRI, at that time, was chosen to execute the introduction, assimilation, improvement and verification of the feasibility of manufacturing ICs in Taiwan. ITRI, as a medium-sized research institute, was responsible for the work of introducing foreign technologies and then transferring them to industry after assimilation (Chang et al., 1994). Following the acquisition of the technology through licensing, ITRI tested the feasibility of a wholly indigenous semiconductor production facility in Taiwan by operating a demonstration plant (pilot plant) (Chen and Sewell, 1996, p.772). Then the technological R&D capacity of the IC industry gradually increased through imitating, copying, or limited improvements in the existing foreign products (i.e., various reverse engineering tactics), and later the IC industry developed in-house R&D capacity through self-directed effort, direct alliances, mergers and joint ventures with foreign companies. Figure 2 shows the mode of technology transfer since the IC industry’s inception, and the dotted line shows the changing mode of technology transfer in the last ten years. Figure 2. The evolution of technology transfer in Taiwan’s IC industry. Foreign/Companies. Licensing. ITRI /ERSO. Spin-off Turnkey Plant. Alliance Merger Joint Venture. Industry /companies. Reverse Engineering Imitating Copying. In-House R& D. 3.4 Innovation (I) commercialisation sub-system Before the product goes for sale in the market, getting new innovation capability is essential. New innovation capability includes embodied and disembodied technology know-how. The innovation process comprises process and product innovation. R&D must generate new products and new processes to meet market requirements.. 3.5 Market (M) sub-system New products are offered to the market, and more sales revenues may then be achieved through commercialisation processes, then fed back to the financial sector. In this research, the role of government is just to supply the needed financial support and tax incentive policies..

(7) An alternative approach to technology policy assessment. 4. 127. Methodology. 4.1 Scope of the research and the expert questionnaire This research centres on an attempt to develop a mathematical model of the national innovation system of Taiwan, particularly with regard to its IC industry. In this research, an assessment of how technology production or importation feeds into the industrial system with regard to the IC industry in Taiwan is pursued. In order to analyse these, the kinds of flows that take place between the organisations or institutions as built from the expert questionnaire and literature review will be considered. Finally, a System Dynamics theory is developed to install into the system. Most of Taiwan’s IC companies are located and concentrated in the Hsinchu Science Park (HSP). In 2004, the HSIP’s IC industry included 164 companies, made $US18,496 million in sales and experienced a growth rate of 30.9% (Hsinchu Science Park Administration, 2004). Because of this success in technological development, this research investigates the interactive relationship among the actors by taking the IC industry in Taiwan and especially HSIP as a case study. The aims of this questionnaire were the following: Firstly, to justify and provide additional perspectives on the structure and its relationship among actors, as well as the parameters of the model. Secondly, to compare the actual findings from the questionnaire survey with the pattern simulated from the model of the NIS. Thirdly, to provide the main source of weightings which will be used in the qualitative parameters of the model. Fourthly, some of the findings would be adopted for technology policy recommendations. The questionnaire contained four sections: conception of the NIS; extent of scientific and technological capacity; performance and delays in innovation and commercialisation; and future international cooperation. There were 20 questions in all and a total of 127 items. The majority of those questions were Likert-type items based on a scale from ‘a very small extent (1 point)’ to ‘a very great extent (5 points)’. The results obtained from the questionnaire will be served as a base for formatting a model as well as being used for analysis in their own right (refer to Appendix 1). During the period from January to March 2002, 50 expert questionnaires were distributed to the following: the IC authorities of central government (five copies); departments or institutes of universities related to science and technology management and policy (ten copies); research institutes related to IC (six copies); and companies which were related to IC and mostly located in the Hsinchu Science Park in Taiwan (29 copies). Among these last 29 questionnaires, there were nine for Fabless, three for Foundry, five for IDM, five for Package, four for Test and three for others. Forty-one questionnaires, plus four incomplete forms, were returned. Based on the 41 complete answers, the response rate came out at 82%, a high ratio that reflected the care taken with the responses (refer to Table 1)..

(8) 128. T.L. Lee. Table 1. The statistics of questionnaire copies. Respondents. Copies distributed. Copies returned. Response rate (%). Government. 5. 4. 80. 10. 7. 70. 6. 5. 83. University Research Institute Industry. 29 Fabless. 86. 9. Foundry. 3. Foundry. 3. IDM. 5. IDM. 4. 8. Package. 5. Package. 5. Test. 4. Test. 2. 3. Others. Others Total copies. 25. Fabless. 50. 3 41. 82. 4.2 Operation of the interview Proceeding with an expert in-depth and face-to-face interview had a double purpose: the questions for interview were designed to focus on the shortcomings of the model and supplied needed information. Most of the interviews were face-to-face, except for one respondent for whom was used a telephone interview owing to a recent surgical operation. The original plan was to interview 14 people, but finally only 13 respondents were the subjects. There were two from government, two from universities and two from research institutes, and seven were industry representatives. The common characteristic of those representatives was their often raising original, constructive and positive opinions on specific affairs related to IC issues, as they have extensive training and work experience in this field.. 4.3 Retrieving the results of the interviews and questionnaires The use of the questionnaire and in-depth interviews aimed to justify and provide additional perspectives on the structure and parameters of the model to follow. The findings of the questionnaire gave rise to the following insights: •. The germination of Taiwan’s IC industry was helped along by government; however, the real driving force for the whole innovation ‘system’ came from two main forces: one was the individual company facing a shift in business style and the other was external technology opportunity.. •. The priority items in aiding the strength of the NIS were: increasing patenting and protection of IPRs, reinforcing higher education systems (e.g., MA or PhD programme) university system, strengthening private sector R&D capacity, more explicit S&T development policy, developing the capability of technological commercialisation, and encouraging technological spin-offs..

(9) An alternative approach to technology policy assessment. 129. •. In strengthening cooperation objectives, the relationships among industries, institutes and universities needed to be developed further. This is because once the IC industry accumulated a body of capital, it became easier for the industry/firms to get advanced science and technology via purchasing or strategic alliances, while research institutes, such as ITRI, which is a quasi-government cooperation, had limited finance and were unable to engage in state-of-the-art/advanced technology R&D. With regard to the universities, most of them concentrated their attention on basic research instead of applied research.. •. In the policy input aspect for raising ‘S&T’ and ‘innovation’ performance, improving R&D capacity, cultivating technological HR, and strengthening the cooperation with international companies are the three main policy focuses. The fast, direct and effective instruments for achieving these are increasing R&D expenditures of industry, strengthening on-the-job training, and alliances. The USA and Japan are thought to be the best partners for Taiwan’s IC firms. No doubt, the results of using instruments are the production of more new inventions and increasing of sales rates and market shares.. In addition, the insights gained from the in-depth interviews provided some guidelines for modelling: •. The substantial amount of capital invested in the IC industry was originally came from the government and local finical markets, but now is shifting to self-financing and from international markets (e.g., ADR, GDR, and ECB) over time.. •. The main acquisition of scientific knowledge and technology came from cooperation with foreign companies through strategic alliances, mergers and joint ventures in recent years. There were no delay effects arising in this part of the process of transfer except in the needed learning curve time.. •. In the early stages, the foundry as a process innovation (the capability of manufacture) forced design houses to engage in product innovation. Following that, the growth of design houses, which are capable of strong design capability, in turn has driven foundries to develop their technology to match the needs of design houses in recent years.. 4.4 Tests for building confidence in the SD model By testing, we compare the simulation to empirical reality for the purpose of model robustness. System Dynamics is able to present many tests for assessing the realism of model assumptions and behaviours, and for generating insights into the causes of observed phenomena. More often the goal is itself a result of the past history of the system that has been established to guide present action (Forrester, 1978). Here the proposed model was traced back to the past history to see if the behaviour changed when a policy was changed (changing a rate variable). Before any policy change, a robustness test is necessary. Hence, the author proceeded with four basic and essential tests. However, there are difficulties in estimating the general model which would allow parameters to also vary across different products of IC, and even across different value chains of the industry..

(10) 130. T.L. Lee. 4.4.1 Dimensional-consistency test The test involves dimensional analysis of a model’s equations. All the equations of the proposed model had to pass the dimensional-consistency check. This means that the author always specifies the units of measurement for each variable in building this model. The scope of this inspection includes whether parameters have meaningless names, strange combinations of units, or non-dimensionless parameters with values of unity, and so on. This proposed model has easily passed this test.. 4.4.2 Parameter-verification test Getting the data will often be difficult. Many of these quantities will be unknown at first; a few can be measured, some can be closely estimated, others will be guessed at. For corrective parameter assessment, judgemental methods based on interview, expert opinion, archival materials, direct experience and statistical method (for example: AHP) have been used in verifying the parameters. Fortunately, most parameters have their real world counterpart and are consistent with relevant descriptive and numerical knowledge of this system. All variables are required to give references, regardless of whether they were drawn from interview, expert opinion, or the author’s judgement.. 4.4.3 Behaviour-reproduction test (RBP replication) Actually the model is a ‘model’: it cannot completely match the real world point-by-point, but similarity of pattern is crucial and necessary. As Sterman stated (2000, p.874): “a good model should exhibit the same modes of behavior observed in the data”. By means of the comparison between the real mode from the past 10 years’ data (1990–1999) and the simulation modes of the IC R&D budget (financial sub-system), R&D researchers, R&D capacity (HR sub-system), product and process innovation (ST sub-system) and annual revenues (product market sub-system), the proposed model might be accepted as an adequate (though not perfect) representation of the real system relative to the study object.. 4.4.4 Extreme-condition test Giving an extreme policy or parameter, and then experimenting to examine conformity with basic physical flows, is still reasonable. For example, given a zero value of inventory, the undertaking of policy simulation can be indicating the shipments also to be zero, then the formulation of a model is qualified as having validity; if this policy simulation does not produce the zero value of shipments, then the formation of a model is invalid. In the proposed model, human resources absolutely depend on the R&D budget. Once the R&D budget is reduced to zero, the model is able to indicate that the rate of HR recruitment is also zero, and the simulation correctly reports the results..

(11) An alternative approach to technology policy assessment. 5. 131. Cause-effect loops. Forrester defined a ‘complex system’ as ‘a high-order, multiple-loop, nonlinear feedback structure’; thus it is evident that feedback loops may be understood to be a major source of mysterious behaviour and policy difficulties. Graham et al. (1994) also indicated that research in dynamic decision-making and System Dynamics shows that environments characterised by multiple feedback processes, side effects, time delays and nonlinearity are particularly troublesome. The detail of the cause-effect loops is drawn as follows. Regarding the proposed System Dynamics model, please refer to Appendix 2. As illustrated in Figure 3, the higher the sales revenues, the greater the investment in R&D, leading to process innovations that lower costs. Lower unit costs enable lower prices, increasing product attractiveness, in turn increasing both industry demand and market share, and then boosting sales still more (R1: F-HR-I-M-F Loop). The development of entirely new products is a core engine of growth for much of industry. The greater the revenue of an industry, the greater and more effective the new product development effort can be. New products create new demand, boosting revenue and increasing investment in raising R&D capacity, and then new product development even more (R2: F-HR-I-M-F Loop). According to this figure, increases in science and technology transfer, leading to R&D capacity promotion corresponding to high success rates of process/product innovation, result in high innovation performance (product innovation and process innovation) which leads to rising sales and high revenues, in turn making industry invest much larger R&D budgets, then feeding back to S&T transfer. (R3: F-ST-HR-I-M-F; R4 & R5: F-ST-I-M-F Loop). Figure 3. Financial, innovation and product causal loop diagram IC Fin ancia l. + +. Sa les Rev enue. +. + Inve st in R &D B udge t. Marke t Share. +. + R3. S&T Transfer. + + R1. Produ ct At tractiv eness. +. + R4. + +. Proc ess Innov ation. Price. +. +. R&D Capa city. Unit C ost. + R2. + + R5 Product Inn ovatio n. +.

(12) 132. T.L. Lee. As illustrated in Figure 4, increasing inputs of education, work experience, R&D expenditure per person, training, patents and publications lead to the elevation of average R&D levels. The higher the R&D level, the stronger the R&D capacity, which leads to high innovation performance (R6: F-HR-I-M-F Loop). Desired numbers of S&T researchers depend on the predicted revenues to complete the work of hiring: larger predicted revenues in turn lead to the hiring of more S&T researchers, then stronger R&D capacity, promoting advanced product and process capacity, which may enhance the industry’s competitive position, in turn increasing sales revenues (R7: HR-I-M-HR Loop). Following the recruitment of S&T researchers, the level of manpower will improve. Once the level of manpower rises, the capacity of absorption and transmission which is embedded in the workforce will also increase during the process of S&T transfer. Evidently, the more the S&T transfer, the greater the performance of innovation commercialisation; and in turn, much higher revenues, larger budgets, more researchers and so on (R8: HR-ST-I-M-HR Loop). Figure 4. R&D capacity generation and human resources causal loop diagram. Revenues. + R6. +. Work Experience. Education. Innovation Commercialisation. + +. + R&D Budget. + R&D Expenditure per person. + Desired ST Researchers. Training. + R7. Average R& D Level. Patents Publications. + RD Capacity. + ST Researchers Recruitment. +. + + R8. Level of Manpower. + S&T Transfer. In Figure 5, the numbers of S&T researchers are increased by recruitment and decreased by the quitting rate. When more S&T researchers are hired, the positive loop operates as a virtuous cycle – efforts to boost R&D capability, greater innovation performance, and still more predicted revenues. The more profitable the industry, the higher the wages and profit shares it can pay to recruit and retain the best (R9: F-HR-I-M-F Loop). There must however be limits to growth. These limits are created by negative feedback. Therefore, the reinforcement loop will show exponential growth, but, as we know, in the real world, the budgets and the personnel cannot have unlimited growth. The higher the.

(13) An alternative approach to technology policy assessment. 133. R&D level, the easier the transfer of S&T. The more the transfer of S&T, the higher the performance of innovation commercialisation, then the system returns to the R9 feedback loop (R10: F-HR-ST-I-M-F Loop). The higher the R&D level, the greater the R&D capacity to reinforce the R&D level, and then accumulate more capacity to transfer S&T, and in turn re-strengthen the R&D capacity (R11). Figure 5. R&D capacity and profitability causal loop diagram. Innovation Commercialisation. +. +. Sales Revenues. +. +. + R10. Net Profit. New R&D Capacity Transferring Capability of Knowledge and Technology. +. + + R11. + R9. + +. Wages and Profit Sharing. R&D Level. + S&T Researcher. +. Figure 6 mainly portrays the interaction and interdependence between product innovation and process innovation in the system. Loop R12 describes how more process innovation leads to more pressure for desired product innovation; in turn, the more the product innovation, the greater the need for desired process innovation. In Loop R13 (I-M-F-I), the more the product innovation, the more the product becomes attractive, the higher the market share, hence more sales, more revenues and then more R&D budget reinvested in product innovation. Similarly to product innovation, in the positive (self-reinforcing) Loop R14 (I-M-F-I): the more the process innovation, the higher the production capacity, causing prices to be reduced, then raising product attractiveness; the more attractive the product, the greater the market share, sales, revenues, R&D budgets and investment in process innovation. Loop R15 (a simplified loop for I-M-F-ST-I, similar to Loop R4 and Loop R5) states that if the R&D budget increases, the larger the budget that will be invested in raising R&D capacity and in S&T transfer, simultaneously promoting the innovation rate of process and product..

(14) 134. T.L. Lee. Figure 6. Product and process innovation causal loop diagram Produ ct Inno vatio n Rate (SP PDI). + +. +. +. -. +. +. + R13. Produ ct At tractiv enes s. Desired Pro duct Inn ovatio n Pri ce. Production Capacity. R&D capaci ty and S&T Transfer. + + Process Inn ovatio n. +. +. Sales. Desired Pro cess Inn ovatio n. + R12. +. -. Mark et Sh are. +. +. Product Inn ovatio n. +R14. + +. Process Inno vation Rate (SP PTI ). +R15. + Rev enues. +. 6. R&D Budget. Policy sensitivity tests. Policy sensitivity exists when a change in assumptions reverses the impacts or desirability of a proposed policy (Sterman, 2000). Parameter sensitivity tests are able to reveal the degree of robustness of model behaviour, as well as indicate the degree that policy recommendations are influenced by uncertainty in parameter values. This test can also help to show the risk involved in using the model for policymaking.. 6.1 Time policy test From the simulation of product innovation and process innovation, Figure 7 shows a relationship of ‘reciprocal causation’. A growing sophistication of the product itself causes process innovation. Process innovation in turn may enable or require further innovation in products, leading to an increasingly tighter linkage between product change and process change. These same concerns have been voiced by Utterback (1979). In the case of Taiwan’s IC industry, process innovation (it is fair to say, process development) emerged at an early stage, then the design houses sprouted up quickly (especially when the manufacturing style of the dedicated semiconductor foundry was introduced by Morris Chang, a Chairman of TSMC) in turn followed by speeded-up product innovation. Tracing back to the development history of semiconductors, we see the rate of technological progress in this industry following the rapidly rising portion of the S-shaped curve – positive first then negative (Klein, 1979)..

(15) An alternative approach to technology policy assessment Figure 7. 135. The simulation of product and process innovation. Policymakers often seek to adjust the state of the system until it equals a goal or desired state. The discrepancy is the gap between the desired state of the system and the actual state. The adjustment time represents how quickly the IC industry tries to respond to innovation from other departments or innovation challenges from competitors. The original assumptions for Process Technology Innovation (PTI) adjustment time and Product Design Innovation (PDI) adjustment time are respectively set to 0.5 year and 1 year. Below we will assess the effect of changing the response time of PTI or PDI on each other, and see if the performance of innovation shows improvement or retrogression. There are six tests that will be experimented with, shown in Table 2. In general, reducing adjustment times makes the system less stable in response to temporary changes, but raises the performance of innovation. Regarding the changing status of these six tests, please refer to the Appendix 2. In sum, the shorter response time reveals the following things: •. The rate of innovation speeds up (the slope is steep).. •. The volume of innovation increases. When there is a high speed of innovation, the volume of innovation rises.. •. The reaction of behaviour (such as the inflection point or high peak) comes earlier. The innovation life cycle has condensed, and as a result the reaction comes early.. •. The speed of process technology innovation is slower than product design innovation. This is because the delay in PTI is longer than in PDI. The longer the delay, the longer the time to adjust to the desired goal.. •. When response time is less than two months, the system becomes unstable (refer to Appendix 2)..

(16) 136. T.L. Lee. Table 2. The result of adjustment time policy tests Original. Test 1. Test 2. Test 3. Test 4. Test5. Test 6. Time for PTI adjustment. 0.5 year. 1 month. 2 months. 0.5 year. 1 year. 0.25 year. 1 year. Time for PDI adjustment. 1 year. 1 month. 2 months. 0.5 year. 1 year. 0.5 year. 2 years. PTI performance. 106.83. +62.72. +30.69. +1.71. –22.71. +16.84. –25.37. PDI performance. 397.5. +99.87. –6.81. +6.37. –39.07. +8.93. –70.95. 6.2 Science and Technology policy tests In the first 20 years of developing the IC industry, technology purchases from overseas were necessary and proved to be an efficient way forward. Gradually, a few big firms, such as TSMC and UMC, insisted on developing their own in-house R&D capacity and not depending on importing foreign technologies in order to compete in the world market in the next era. For the next ten years, if Taiwan reduces or even stops purchasing foreign technology, what will happen to innovation performance and R&D capacity? If this kind of S&T policy is Taiwan’s only way of standing in the international arena in the future, what kind of policy portfolio would be appropriate? Next we attempt to choose four policy tests under the principle of no purchase of foreign technology. From Table 3 it can be seen that both R&D capacity and innovation performance in Test 2 do not decrease as much. If there is no purchase of foreign technology but instead a strengthening of cooperation with domestic research institutes, still the original level cannot be maintained. Doubling government grants, in Test 3, seems unable to save the poor innovation performance. It can be established that raising government grants is not an effective way to promote innovation performance. In other words, this is not a proper solution to improve innovation performance. Another interesting thing to notice in Test 4 is that the PTI still drops, and PDI only slightly increases, although the R&D budget has been raised by 5%. On this basis, it can be inferred that raising the R&D budget alone is also not enough to maintain the original level when deciding not to purchase foreign technology. Table 3. The results of science and technology policy tests The assumption of policy set. Test 1. No S&T purchase, ceteris paribus. Test 2. No S&T purchase but raising institute spin-off. Test 3. No S&T purchase but raising government grants. Test 4. No S&T purchase but raising industry R&D budget by 5%. R&D capacity (units). Innovation performance PTI (units) PDI (units). –1236.51. –6.74. –34.87. –769.52. –2.57. –13.43. –1236.51. –6.71. –34.06. +22 977.49. –3.76. +0.96. Policy 5% IC finance was appropriately allocated to S&T purchase and R&D budget increase suggestion.

(17) An alternative approach to technology policy assessment. 137. The policy simulation part of this study explores what is the best S&T policy for Taiwan’s IC industry. As these policy simulations indicate, Taiwan cannot reach a good innovation performance level without purchasing foreign technology or cooperating with foreign companies. The most important thing is that no single S&T policy can be emphasised alone; instead it is a matter of arranging the appropriate portfolio policy in terms of the needs of industry. If there is a 5% increase in IC finance, ceteris paribus, this 5% may be allocated to S&T purchases and R&D budget increases, to see if the R&D capacity and innovation performance are better than under the previous tests. Table 4 shows four kinds of portfolios. With as neutral an attitude as is possible, Portfolio 3 seems better than the others from the point of view of a ‘win-win’ strategy. Table 4. Policy portfolio suggestions. 5% financial policy portfolio Innovation performance. Portfolio 1 R&D S&T purchase 1% 4% +5,122 PTI: +8.57 PDI: +53.37. Portfolio 2 R&D S&T purchase 2% 3% +9,284 PTI: +7.97 PDI: +55.22. Portfolio 3 R&D S&T purchase 3% 2% +14,070 PTI: +7.19 PDI: +55.28. Portfolio 4 R&D S&T purchase 4% 1% +19,181 PTI: +6.26 PDI: +53.74. As mentioned above, no one S&T policy can be emphasised alone and no one S&T policy can be suitable for all companies. There is a great heterogeneity existing in each company even in the same industry. S&T policy needs to be fitted to the needs of industry; moreover, the current industry environment (domestic and international), competitors, all the suppliers and demanders within the value chain, etc. also need to be taken into consideration. By doing so, the industry can make the right decision at the right time.. 7. Findings from the SD model structure. From the causal loop diagram we found that more process innovation leads to more pressure for desired product innovation; in turn, the greater the product innovation, the greater the need for desired process innovation. The greater the product innovation, the more attractive the product becomes, the higher the market share, hence more sales, more revenues and then more R&D budgets reinvested in product innovation. Similarly, the greater the process innovation, the higher the production capacity, causing prices to be reduced, then raising product attractiveness; the more attractive the product is, the greater the market share, sales, revenues, R&D budgets and investment in process innovation. The Taiwan IC industry appears to have displayed many of the characteristics of a ‘virtuous circle’, but of course it is easy in principle to reverse all these positive feedbacks. Here it is worth noting that all the indicators for generating R&D capacity are carried or embedded in ‘people’ themselves; for this reason, the generation of R&D cannot be conducted without ‘people’. Put in another way, the indicators selected cannot directly produce R&D capacity; instead it needs to be produced by people. Similarly, the process technology and product design innovation themselves are unable to produce new.

(18) 138. T.L. Lee. products; instead, this comes about via the process of production. Otherwise, it is easy to make logical and structural errors, and some had to be overcome in developing the present model (refer to Appendix 3, about the HR sub-system and ST sub-system). This result reflects how causal relationships always tend to be simplified, which could easily lead to ignoring the structural factors of relevant questions. Deeper thinking, with the aid of the insights from the proposed model structure, would reveal the complex, dynamic and non-linear relationship for the NIS. Although System Dynamics (SD) does present the process coherently, it has a weakness in describing why and how the people interactive learning process occurred within the Taiwanese innovation system model. For example, it explores the ‘time for response’ as the key factor for innovation performance in the policy tests, but without fully investigating why and how the time for response plays a crucial part in the model of the innovation system. This could be explored further in future studies.. 8. Conclusions and policy implications. There is always a major rate and level to represent the activities of each sub-system, such as monetary flow/level in the financial sub-system; people flow/level and capacity flow/level in the human sub-system; science and technology flow/level in the S&T transfer sub-system; product and process flow/level in the innovation commercialisation sub-system; and sales and revenues flow/level in the product market sub-system. By means of changing a rate variable, we can see how the behaviour changes. Facing a dynamic environment, we can build up the policy simulations and proceed with policy rehearsal by using computer modelling in order to pre-assess and avoid policymaking errors in the real world. Pardue et al. (1999) pointed out that R&D is the main driving force for technical change and commercialising technical advancement. However, from the experiment, it was concluded that ‘time to respond to the desired PTI/PDI’ is the main driving force for raising commercialisation performance. According to the results of S&T policy Tests 2 and 3 in Table 3, the innovation performance did not show obvious improvement when testing for ‘no S&T purchase’ while raising domestic technology spin-off or else raising government grant/subsidy. These results reflect how foreign S&T transfer is still an essential factor in Taiwan’s IC industry. The exceptions to the improvements in process technology, relative to other state-of-the-art technology, in the IC industry in Taiwan more and more deeply depend on the introduction and transfer of foreign advanced technology. The questionnaire indicates that the role of ‘strategic alliances’ is becoming more important in getting new technologies, instead of direct technology purchases. In addition, according to the results of the questionnaire in both improving performance of S&T and innovation commercialisation, the USA and Japan are thought to be the best partners for Taiwan’s IC firms. Regarding the portfolio of the science and technology policy test in Table 3, the subsidy from government is no longer the key ‘propeller’ which pushes forward the improvement of the IC industry. Based on the classification of Rothwell and Zegveld (1985), the effects of innovation policy on S&T activities are divided into three aspects: supply policy, demand policy and environmental policy. In supply policy, the firms of the IC industry claim to supply more S&T manpower in respect of education policy, and.

(19) An alternative approach to technology policy assessment. 139. more information. Following the massive accumulation of financial and technology capacity, the IC industry/firm no longer seems to need ‘demand policy’ support from government. Because the IC industry is a globalised industry, it aims at the global market, and therefore is in no need of the protectionist approach previously applied. Similarly, the technical norms also follow the technology leader in this field, but not the mother country. From the environmental policy perspective, firms claim a maturity of infrastructures, preferential taxes and eased restrictions in the last four instruments. According to the interviews, they all think that less government is the best government, and appeal to free market mechanisms. Based on the characteristics of innovation management, S&T policy needs to be fitted to the needs of industry in the current industrial environment (domestic and international); competitors and all the suppliers and demanders within the value chain need to be taken into consideration as well. As Coyle (1977, p.29) pointed out: “….so the problem is to select a harmonious collection of economic and social policies which will be robust for the whole economy in the face of external shocks from the economic controller’s complement and environment.”. Once all the elements of all policies have been considered, it may be possible to make the right decision at the right time. As already noted, no one S&T policy can be emphasised alone. This result gives confirmation to the fact that policy (government S&T policy or the industry’s own policy) cannot place its emphasis just on one side of policy.. References Balzat, M. and Hanusch, H. (2004) ‘Recent trends in the research on national innovation systems’, Journal of Evolutionary of Economics, Vol. 14, pp.197–210. Chang, P.L., Shih, C. and Hsu, C.W. (1994) ‘The formation process of Taiwan’s IC industry-method of technology transfer’, Technovation, Vol. 14, No. 3, pp.161–171. Chen, C.F. and Sewell, G. (1996) ‘Strategies for technological development in South Korea and Taiwan: the case of semiconductors’, Research Policy, Vol. 25, pp.759–783. Coyle, R.G. (1977) Management System Dynamics, Chichester: John Wiley and Sons. Edquist, C. (1997) ‘Institutions and organizations in systems of innovation’, in C. Edquist (Ed.) System of Innovation: Technologies, Institutions and Organizations, London: Pinter. Edquist, C. (2001) ‘The systems of innovation approach and innovation policy: an account of the state of the art’, Paper Presented at the Nelson and Winter DRUID Summer Conference, Aalborg Congress Center, Aalborg, Denmark, 12–15 June. Fagerberg, J., Mowery, D.C. and Nelson, R.R. (Eds.) (2005) The Oxford Handbook of Innovation, Oxford: Oxford University Press. Forrester, J.W. (1978) ‘Market growth as influenced by capital investment’, in E.B. Roberts (Ed.) Managerial Applications of System Dynamics, MA: The MIT Press. Freeman, C. (1987) Technology Policy and Economic Performance: Lessons from Japan, London: Pinter. Freeman, C. (1995) ‘The “national system of innovation” in historical perspective’, Cambridge Journal of Economics, Vol. 19, pp.5–24, reprinted in C. Edquist (Ed.) (2000) Systems of Innovation: Growth, Competitiveness and Employment, Glos: Edward Elgar. Graham, A.K., Morecroft, J.D.W., Senge, P.M. and Sterman, J.D. (1994) ‘Model-supported case studies for management education’, in J.D.W. Morecroft and J.D. Sterman (Eds.) Modeling for Learning Organizations, Productivity, Portland..

(20) 140. T.L. Lee. Hsinchu Science Park Administration (2004) An Introduction of Science-Based Industrial Park, Taiwan. Johnson, A. and Jacobsson, S. (2000) ‘The emergence of a growth industry: a comparative analysis of the German, Dutch and Swedish wind turbine industries’, Paper Presented at the Schumpeter Conference in Manchester. Klein, B.H. (1979) ‘The slowdown in productivity advances: a dynamic explanation’, in C.T. Hill and J.M. Utterback (Eds.) Technological Innovation for a Dynamic Economy, NY: Pergamon Press. Lee, Y-L. (2002) ‘Dynamic analysis of the national innovation systems model – a case study of Taiwan’s integrated circuit industry’, Unpublished PhD Thesis, PREST, University of Manchester. Liu, X. and White, S. (2001) ‘Comparing innovation systems: a framework and application to China’s transitional context’, Research Policy, Vol. 30, pp.1091–1114. Lundvall, B. (1998) ‘Why study national systems and national styles of innovation’, Technology Analysis and Strategic Management, Vol. 10, No. 4, pp.407–421. Lundvall, B. (2000) ‘Introduction’, in C. Edquist and M. Mckelvey (Eds.) Systems of Innovation: Growth, Competitiveness and Employment, an Elgar Reference Collection, Cheltenham, UK: Edward Elgar Publishing. Lundvall, B. (Ed.) (1992) National Innovation Systems: Towards a Theory of Innovation and Interactive Learning, London: Pinter. Mathews, J.A. (1997) ‘A silicon valley of the east: creating Taiwan’s semiconductor industry’, California Management Review, Vol. 39, No. 4, pp.26–54. Metcalfe, S. (1995) ‘The economic foundations of technology policy: equilibrium and evolutionary perspectives’, in P. Stoneman (Ed.) Handbook of the Economics of Innovation and Technological Change, London: Blackwell. Mowery, D.C. and Ziedonis, A.A. (1998) ‘Market failure of market magic? – Structural change in the US national innovation system’, STI Review – Science/Technology/Industry, Vol. 22, pp.101–136. Mytelka, L. (2000) ‘Local systems of innovation in a globalized world economy’, Industry and Innovation, Vol. 7, No. 1, pp.15–32. Mytelka, L. (2001) ‘Promoting scientific and technological knowledge for sustainable development’, Paper for the Third UN Conference on Least Developed Countries, Round Table: Education for All and Sustainable Development in LDCs. Mytelka, L. (2003) ‘Competence building and policy impact through the innovation review process: a commentary’, Paper Presented on the IDRC-UNESCO Joint Workshop on Future Directions for National Reviews of Science, Technology and Innovation in Developing Countries, UNESCO, Paris, 23–24 April. Nelson, R. (Ed.) (1993) National Innovation Systems: A Comparative Analysis, NY: Oxford University Press. Nicolis, G. and Prigogine, L. (1994) The university, facing up its European responsibilities’, Paper for the Fourth European Framework Program and Research on Complex Systems, Pisa, November. OECD (1994) Main Definitions and Conventions for the Measurement of Research and Experimental Development (R&D) – A Summary of the FRASCATI Manu, Paris. OECD (1997) ‘National innovation systems’, The NIS Project, Paris. Pardue, J.H., Clark, Jr., T.D. and Winch, G.W. (1999) ‘Modelling short- and long-term dynamics in the commercialization of technical advances in IT producing industries’, System Dynamics Review, Spring, Vol. 15, pp.97–105. Patel, P. and Pavitt, K. (1994) ‘The nature and economic importance of national innovation systems’, STI Review, Vol. 14..

(21) An alternative approach to technology policy assessment. 141. Rickne, A. (2000) New Technology-Based Firms and Industrial Dynamics: Evidence from the Technological Systems of Biomaterials in Sweden, Ohio, and Massachusetts, Department of Industrial Dynamics, Chalmers University of Technology. Rothwell, R. and Zegveld, W. (1985) Reindustrialization and Technology, Harlow: Longman. Smith, K. (2000) ‘Innovation as a systemic phenomenon: rethinking the role of policy’, Enterprise and Innovation Management Studies, Vol. 1, No. 1, pp.73–102. Sterman, J.D. (2000) Business Dynamics-Systems Thinking and Modelling for a Complex World, MA: McGraw-Hill. Utterback, J.M. (1979) ‘The dynamics of product and process innovation in industry’, in C.T. Hill and J.M. Utterback (Eds.) Technological Innovation for a Dynamic Economy, NY: Pergamon Press. Xue, L. (1997) ‘Promoting industrial R&D and high-tech development through science parks: the Taiwan experience and its implications for developing countries’, International Journal of Technology Management, Special Issue on R&D Management, Vol. 13, Nos. 7–8, pp.744–761.. Notes 1. 2. 3. Johnson and Jacobsson (2000) mention five functions: to create ‘new’ knowledge; to guide the direction of the search process; to supply resources, i.e., capital, competence and other resources; to facilitate the creation of positive external economies; and to facilitate the formation of markets. Rickne (2000) provides a long list of functions: to create human capital; to create and diffuse technological opportunities and products; to incubate in order to provide facilities, equipment and administrative support; to facilitate regulation for technologies, materials and products that may enlarge the market and enhance market access; to legitimise technology and firms; to create markets and diffuse market knowledge; to enhance networking; to direct technology, market and partner research; to facilitate financing; and to create a labour market that the NTBF can utilise. Liu and White (2001) identify five fundamental activities. These are R&D, implementation, end-use, linkage, and education..

(22) 142. T.L. Lee. Appendix 1 Questionnaire SECTION ONE: National Innovation System (NIS) According to Stan Metcalfe: “A national system of innovation is that set of distinct institutions which jointly and individually contribute to the development and diffusion of new technologies and which provides the framework within which governments form and implement policies to influence the innovation process.” Questions Item Q1-1. Compared with Ticks in order, 5 is very good, 4 is good, USA, Europe and 3 is moderate, 2 is bad, 1 is very bad Japan, do you think that 5 4 3 2 1 Taiwan has a good innovation performance or process? Q1-2. Are you aware of 5 4 3 2 1 the concept of NIS? Q1-3. What institutes would you include in the National Innovation System in Taiwan? (multiple choice). Government Industry (company) Association of Industry Education and University system Public R&D Institute Private R&D Institute (quasi-government cooperation) Science Park Others Government Industry (company) Association of Industry Education and University system Public R&D Institute Private R&D Institute (quasigovernment cooperation) Science Park Others. Q1-4. Which of these institutes do you have practical interaction with in order to innovate? (multiple choice). Q1-5. Do you consider that these types of institutes strongly or weakly contribute to the National System of Innovation in Taiwan?. Ticks in order, 5 is the very strong, 1 is the very week 5. 4. 3. 2. 1. 1. Government. . . . . . 2 3 4. Industry (company). . . . . . Association of industry. . . . . . Education and University system. . . . . . 5 6. Public R&D Institute. . . . . . Private R&D Institute (quasi-government cooperation). . . . . . 7 8. Science Park. . . . . . Others. . . . . .

(23) An alternative approach to technology policy assessment. 143. Appendix 1 Questionnaire (continued) Questions Q1-6. Where did the driving force come from for the Taiwanese Innovation System? And what extent?. Item Ticks in order, 5 is the most important, 1 is the least important 5. 4. 3. 2. 1. 1 2 3. Government intervention. . . . . . Industry cooperation. . . . . . Individual company face the shift of business style. . . . . . 4. External technology opportunity. . . . . . 5. External environment change. . . . . . 6. Others. . . . . . Ticks in order, 5 is the most important, 1 is the least important. Q1-7. What extent of the following types of interactions need to be strengthened in building the Taiwanese Innovation System?. 5. 4. 3. 2. 1. 1. Cooperation between university-industry. . . . . . 2. Cooperation between institute-industry. . . . . . 3. Cooperation between institute-university. . . . . . 4. Interaction between government-industry. . . . . . 5. Cooperation between industry association-industry. . . . . . 6. Cooperation between industry association-university. . . . . . 7. Cooperation between industry-foreign university or institute. . . . . 8. Others. . . . . .

(24) 144. T.L. Lee. Appendix 1 Questionnaire (continued) Questions. Item. Q1-8. To what extent should the following items need to be strengthened in 1 building the Taiwanese Innovation System? 2. Ticks in order, 5 is the most important, 1 is the least important 5. 4. 3. 2. 1. . . . . . Better higher educational university system. . . . . 3. Producing more graduates. . . . . . 4 5. Producing more skills. . . . . . More explicit science and technology development policy. . . . . 6. Increasing government R&D expenditures. . . . . . 7. Public sector R&D capacity. . . . . . 8. Private sector R&D capacity. . . . . . 9. Strengthen technological equipment and facilities. . . . . . 10. Increase the output of research publications and reports. . . . . 11. Increase patenting and protection of IPRs. . . . . . 12. Better legal and financial institutions. . . . . . 13. Capability of technological commercialisation. . . . . . 14. Encouraging technological spin-offs. . . . . 15. Greater cooperation with foreign companies. . . . . . 16. Increased foreign direct investment. . . . . . 17. Greater interaction with consumers in innovation. . . . . . 18. Others. . . . . . Better secondary educational system. .

(25) An alternative approach to technology policy assessment. 145. Appendix 1 Questionnaire (continued) SECTION TWO: Scientific and Technological Capacity Producing, absorbing, improving, transferring, and diffusing Scientific and Technological knowledge (Note: In this section, we do not distinguish between Science and Technology) Ticks in order, 5 is the very strong, 1 is the very weak. Q2-1. How do you regard the Scientific and Technological capacity of the IC Industry compared with other industries in Taiwan? Q2-2. To what extent do the following policies input need to be strengthened in order to raise Scientific and Technological performance in the IC industry?. Q2-3. To what extent do the following indicators of R&D capacity?. 5. 4. 3. 2. 1. . . . . . Ticks in order, 5 is the most important, 1 is the least important 5. 4. 3. 2. 1. 1. Government R&D capacity. . . . . . 2. Private institutes’ R&D capacity (quasi-government cooperation). . . . . . 3. Industrial (company) R&D capacity. . . . . . 4. Cultivating technological human resources. . . . . . 5. Establishing private technology suppliers (including spin-offs) and consultants. . . . . . 6. Cooperation with international companies. . . . . . 7. Others. . . . . . Ticks in order, 5 is the most important, 1 is the least important 5. 4. 3. 2. 1. 1. Level of education. . . . . . 2. Work experience. . . . . . 3. Patents. . . . . . 4. Publications. . . . . . 5. Process technology capability (µm). . . . . . 6. R&D expenditure. . . . . . 7. Training on the job. . . . . . 8. Others. . . . . .

(26) 146. T.L. Lee. Appendix 1 Questionnaire (continued) Questions. Q2-4. To what extent do the following rates of change of input or output need to be speeded up in order to raise Scientific and Technological performance?. Item. Ticks in order, 5 is the most important, 1 is the least important. 3. 5 The R&D expenditures of government The R&D expenditures of industry (companies) The amounts of publications . 4. The numbers of patents. . . . . . 5. The numbers of researchers. . . . . . 6. The inflows of foreign direct investment. . . . . 7. Others. . . . . 1 2. . 4. 3. 2. 1. . . . . . . . . . . . . SECTION THREE: Innovation and Commercialisation Capacity Producing innovations, including product innovation and process innovation, up to the point of commercialising the innovation Q3-1. How do you regard the Innovation and Commercialisation capacity of IC Industry compared with other industries in Taiwan?. Ticks in order, 5 is the very strong, 1 is the very weak 5 4 3 2 1. Q3-2. To what extent do the following policies input most need to be strengthened in order to raise Innovation and Commercialisation performance in the IC industry?. Ticks in order, 5 is the most important, 1 is the least important. . . . . . 5. 4. 3. 2. 1. 1. In-house industrial (company) R&D capacity. . . . . . 2. Cultivating technological human resources. . . . . . 3. More government generous financial resources. . . . . . 4. Cooperation with international companies. . . . . . 5. Cooperation with customers and consumers. . . . . 6. Government support for commercialisation. . . . . . 7. Others. . . . . .

(27) An alternative approach to technology policy assessment. 147. Appendix 1 Questionnaire (continued) Questions. Item. Q3-3. To what extent do the following rates of change most need to 1 be speeded up in order to raise Innovation and 2 Commercialisation 3 performance? 4. Q3-4. How can we reduce the impact of delays in innovation in the IC industry?. Ticks in order, 5 is the most important, 1 is the least important 5 4 3 2 1 Rate of producing new inventions The number of production. . . . . . . . . . . The number of sales Domestic market shares. . . . . . . . . . . 5. Foreign market shares. . . . . . 6 7. Rate of profit Rate of investment. . . . . . . . . . . 8. Others. . . . . . Ticks in order, 5 is the most important, 1 is the least important 5. 4. 3. 2. 1. 1. Improving process innovation. . . . . . 2. Increasing the sales to more effective channels. . . . . . 3. Continuing product innovation. . . . . . 4. Strengthening the cooperation with foreign firms. . . . . . 5. Increasing capital from foreign direct investment Increasing R&D expenditures as percentage of GDP Increasing R&D expenditures relative to sales revenue. . . . . . . . . . . . . . . . . 6. 7. 8. Others. .

(28) 148. T.L. Lee. Appendix 1 Questionnaire (continued) SECTION FOUR: Source of Technology Q4-1. How to get new technologies to respond to the short IC product life 1 cycle? 2 3 Q4-2. What should be the style of cooperation with a foreign company?. Q4-4. Which region for cooperation with foreign companies would be most appropriate for improving Taiwan’s Innovation and Commercialisation performance?. In-house R&D. . . . . . Purchasing them outside Strategic alliance. . . . . . . . . . . Ticks in order, 5 is the most important, 1 is the least important 5. 4. 3. 2. 1. 1. Purchase technological skill. . . . . . 2. Purchase patents or licenses. . . . . . 3. Purchase of turn-key plants. . . . . . 4 5. Joint ventures Alliances. . . . . . 6. Marketing and sales agreements Others. 7 Q4-3. Which region for cooperation with foreign companies would be most appropriate for improving Taiwan’s Scientific and Technological performance?. Ticks in order, 5 is the most important, 1 is the least important 5 4 3 2 1. . . . . . . . . . . . . . . . Ticks in order, 5 is the most important, 1 is the least important 5. 4. 3. 2. 1. 1 2. USA Europe. . . . . . . . . . . 3. Japan. . . . . . 4 5. Other Asia Pacific area Mainland China. . . . . . . . . . . 6. Others. . . . . . Ticks in order, 5 is the most important, 1 is the least important 5. 4. 3. 2. 1. . . . . . . . . . . 1 2. USA Europe. 3. Japan. . . . . . 4 5. Other Asia Pacific area Mainland China. . . . . . . . . . . 6. Others. . . . . .

(29) An alternative approach to technology policy assessment. Appendix 2 The changing status of ‘Adjustment Time Policy Tests’ Adjustment Time Policy Test – original. Adjustment Time Policy Test 1 (1 month versus 1 month). Note:. The original diagram shows the 0.5- and 1-year innovation performance forecasts (from year 2000 to year 2010) based on the original model and the same conditions. In other words, after 1999, the forecasting simulation data are based on the parameter values of the last year – 1999. The patterns in all tests approximately exhibit the same behaviour. Irrespective of whether process or product innovation is considered, they basically show an S-shape growth before 1999.. 149.

(30) 150. T.L. Lee. Adjustment Time Policy Test 2 (2 month versus 2 month). Adjustment Time Policy Test 3 (0.5 year versus 0.5 year). Note:. The original diagram shows the 0.5- and 1-year innovation performance forecasts (from year 2000 to year 2010) based on the original model and the same conditions. In other words, after 1999, the forecasting simulation data are based on the parameter values of the last year – 1999. The patterns in all tests approximately exhibit the same behaviour. Irrespective of whether process or product innovation is considered, they basically show an S-shape growth before 1999..

(31) An alternative approach to technology policy assessment Adjustment Time Policy Test 4 (1 year versus 1 year). Adjustment Time Policy Test 5 (3 month versus 6 month). Note:. The original diagram shows the 0.5- and 1-year innovation performance forecasts (from year 2000 to year 2010) based on the original model and the same conditions. In other words, after 1999, the forecasting simulation data are based on the parameter values of the last year – 1999. The patterns in all tests approximately exhibit the same behaviour. Irrespective of whether process or product innovation is considered, they basically show an S-shape growth before 1999.. 151.

(32) 152. T.L. Lee. Adjustment Time Policy Test 6 (1 year versus 2 year). Note:. The original diagram shows the 0.5- and 1-year innovation performance forecasts (from year 2000 to year 2010) based on the original model and the same conditions. In other words, after 1999, the forecasting simulation data are based on the parameter values of the last year – 1999. The patterns in all tests approximately exhibit the same behaviour. Irrespective of whether process or product innovation is considered, they basically show an S-shape growth before 1999..

(33) Sale s. Profit Share. Net Profit. Ave Fore Depo Receipt. IC Financial. ~. Eff of NPDI. Ratio of TRD. Time for PTI Adjust. Ratio of RD Invest. Annu IC RD Budg. IC RD Budget. Time for PDI Adjust. Perceived Lead Time. ~. Eff of Lead Time. Reinvest in TRD. Ratio of PEP. Retained Earnings. Purch Equip & Plant. Ave VC Invest IC. Annu IC Finan. ~. Eff of price. Prod Attrac. Capital Market(Financial). Revenue s. ~. Pric e. Retain Earn Conv to Cash Capital Incre. Ratio of CCCI. Profit Share Ratio. Net Profit Mgn. ~. Global Market Demand. Annu Revenues. Annu Sales. Taiwan Market share. ~. Product Market. ~. Total ST Researchers. Annu IC RD Budg. Wafer Fab. Time to Mask. Anul RD Expend per Res. Ave Work Dura per Res. Ave Training per Res. Ave Level of Edu. Ave Pat per Res ~. Time to Layout. Layout to Mask. Layou t. Time for PTI Adjust. Expe Wt. Training Wt. Edu Wt. Pat Wt. Eff of Annu RD Expend. ~. Eff of Expe Annu RD Expend Wt. ~. ~. Eff of Trianing. ~. Publi Wt. Eff of Publi. ~. PTI Proj. Desired FPTI. ~. ~. RD Cap Generation. ~. Year Grow Rate. ~. RD Payoff from PNST. PNS T. Total ST Researchers. Current Person Cost. ~. ~. New comers. Profit Share. Eff of Profit Share. Time to Hire. HR Recrut Rate. Total ST Researchers. RD Cap Loss. NS T. Ave Train Dura per Res. Ave Work Dura per Res. HR Quit Rate. RD Loss per Exit. Ave RD Level per Res. Training U. Quality of Manpower. Ave Cost per NST. NST Gain from Foreign. ~. Ratio of ST Trans. ST Trans Budg. Expe U. Expe Researchers. Edu U. Level of Manpower. HR Assim Rate. Total ST Researchers. ~. SP NST Trans. NST Tran from Foreign. NST Trans from TRDP. Annu IC Finan. Science & Technology Transfer. Ave NST Trans from TRDP. Ave NST Coop from TRDP. Internal from PNST to RD Cap. Annu Wages. ST Res Discrep. Person Cost Ratio. Annu IC RD Budg. Predi IC RD Budg. Predi Person Cost. Desired ST Researchers. Resource. RD Cap. Ave Cost Per PTI. ~. SP PTI. ~. Effect of PNST. Effect of RD Cap. ~. Ave Cost per PDI. IC Grants from Gov. Ratio of Grants in PTI IC Grants from Gov Human. PTI PI. Annu RD Expend U. Expe U. Fine Proc Tech Inno. Cur Ave RD Level per Res. Training U. Edu U. Pat U. Publi U. Ratio of PTI RD. PTI Budget. ~. SP PDI. Desired NPDI. New Prod Desi Inno. Ave Product Cap. NPDI Based on FPTI. Proc Tech Inno R1. Annu IC RD Budg. Eff of Edu. Eff of Pat. Ave Publi per Res. PTI Del Rate. Mas k. PDI PI. PDI Proj. Ratio of Grants in PTI PDI Budget. FPTI Ask from NPDI Proc Tech Inno Proc Tech Inno L2 L1. Mask to Wafer. Proc Tech Inno R2. Proc Tech Inno L3. Time to Wafer. Wafer to Pack&Test. Proc Tech Inno R3. Fine Proc Tech Inno. Time to Pack&Test. Prod to Market. Pack & Test. Prod Desi Inno L1. Annu IC RD Budg. Prod Desi Inno R1 Time for PDI Adjust. Prod Desi Inno L2. Prod Desi Inno R2. Prod Desi Inno L3. Prod Desi Inno R3. New Prod Desi Inno. PDI Del Rate. Innovaion Commercialization(Product & Process Innovation) Ratio of PDI RD. An alternative approach to technology policy assessment 153. Appendix 3 The system dynamics model of Taiwan’s IC industry.

(34)