• No results found

Applicability of Patent Information in Technological Forecasting: A Sector-specific Approach

N/A
N/A
Protected

Academic year: 2022

Share "Applicability of Patent Information in Technological Forecasting: A Sector-specific Approach"

Copied!
9
0
0

Loading.... (view fulltext now)

Full text

(1)

Applicability of Patent Information in Technological Forecasting:

A Sector-specific Approach

Byungun Yoon

Department of Industrial & Systems Engineering, Dongguk University-Seoul, 26, 3-ga, Pil-dong, Joong-gu, Seoul, 100-715, South Korea

and Sungjoo Lee†

Department of Industrial & Information Systems Engineering, Ajou University, San 5, Woncheon-dong, Yeongtong-gu, Suwon 443-749, South Korea

Received 19 September 2011, revised 31 October 2011

Since the strategies of firms for protecting their innovations could vary between patents and trade secrets according to the characteristics of industries, patent analysis might not always be appropriate for forecasting technological trends in industries. This paper aims to identify relevant industries where patent information can be effectively utilized to scrutinize the trends and effects of technological activities. To this end, first, sectoral differences in patenting activities are explored by analysing the community innovation survey (CIS) data. Second, the applicability of patent trend analysis for technological forecasting is examined in each industry through the S-curve fitting process with patent data. Finally, correlation analysis between R&D data (R&D investment and loyalty income) and patent application data is performed to demonstrate the explanatory power of patent information in R&D management, by investigating the relationship between the inputs and outputs of a R&D system. The results of this paper will help support a strategic process for using patent analysis to envisage future trends and comprehend diverse characteristics of a technology.

Keywords: Patent analysis, S-curve, sectoral patterns of innovation, technological forecasting, R&D investment

Technological forecasting aims to anticipate the direction and speed of change in technological trends, enabling the early detection of revolutionary technologies.1 The early detection of critical signals in technological trends assists researchers and practitioners in preparing dynamic scenarios in terms of economic as well as technological competition. Accordingly, technological forecasting is an unavoidable process for devising successful policies that can meet both public and private needs.2-4 In general, technological forecasting techniques are classified into two categories:

exploratory forecasting to extend historical trends of past data into the future, and normative forecasting to predict the capability that will be available to meet future needs.5 Quantitative techniques such as trend extrapolation for exploratory forecasting have been mostly suggested to forecast promising technology in a short-term way.6-8 Since patent information provides practical past evidence of technology planning, technological forecasting on the basis of patent analysis can facilitate short-term forecasting (1 - 5 years). The

underlying idea of patent-based technology forecasting is that the increasing number of filed patents in a technology area indicates high value of the technology, stimulating new technology development. The historical patterns of patent applications often indicate the trends of growth in a technology area.9 This approach, thus, generally utilizes the extrapolation technique for the purpose of short-term forecasting rather than scenario analysis that aims at providing a better understanding of the range of possible business and technological environments in the mid or long-term future. Fig. 1

__________

†Email: Corresponding author: sungjoo @ajou.ac.kr Fig. 1  Position of patent-based technology forecasting

Environmental scanning

Delphi

Morphology analysis

TRIZ

Technology roadmap

Scenario analysis

Patent analysis

Bibliometric analysis

Trend exploration Long-term

Short-term

Normative Exploratory

(2)

depicts the position of patent analysis in the landscape of technology forecasting models. Consequently, this research deals with the application of patent information to conduct exploratory, short-term technology forecasting.

In engineering management, patent analysis has long been regarded as a crucial method for strategic decision-making. From a macro perspective, patent statistics are adopted to evaluate innovation processes and provide indicators of national technological capacity.10-12 Meanwhile, from a micro perspective, considerable studies analyse patent information to assess the effectiveness of R&D activities and identify potential research areas.13-15 Patent analysis has also served as a valuable reference for priority-setting in R&D investment by investigating financial efficiency.16 Many firms have come to consider patents as a source of competitive power and emphasize patent acquisition as a part of their business strategies.17 In particular, patent portfolios are increasingly regarded as being of major interest for strategic business development decisions, leading to the implementation of well-defined patent strategies.18 Generally, managerial tools like patent maps and patent portfolios require patent information, which enables analytical work due to a large range of advantages with respect to the availability of databases, scope of coverage, and richness of information.19 Lately, the importance of patent analysis is being increasingly highlighted in high- technology management as innovation processes become more complex and the cycle of innovation becomes shorter.20 In addition, most of the technological information is available in patent documents and hence patent databases. According to a survey of 435 patents, mainly US patents, it was concluded that 70 per cent of new technology was not disclosed later elsewhere, 13 per cent partly disclosed later on, and 16 per cent completely disclosed in non- patent literature subsequently.21 Moreover, another study shows that just 5.77 per cent of technological information in patent documents is later published in non-patent literature, indicating that patent information is unique and imperative in that it can be scarcely collected from other literature sources.22

However, patent analysis has several drawbacks when it is used as an indicator to measure technological progress. First, not all inventions are patented because a patent office excludes a lot of inventions that cannot meet the patentability criteria

(novelty, non-obviousness, and industrial applicability).23 Second, inventors might strategically decide to exploit the inventions through other types of intellectual property, such as copyright and trade secrets.24 Basically, such a strategy originates from firms’ concerns that the exposure of their inventions might allow competitors to easily replicate their innovations. In general, process innovations are more often protected in the form of trade secrets rather than patents because a patented process can be easily imitated by competitors. Third, not all patents have equal value because a company might retain a lot of valueless patents, implying that companies strategically apply for patents in insubstantial technologies to secure a favourable position with regard to competitors. Finally, an invention is not identical to innovation because all inventions may not result in commercial success. Since many patents tend to be characterized as inventions, patent analysis might not be useful for analysing economic profits through technological progress. Thus, a framework needs to be proposed for identifying appropriate cases wherein patent information can be used with a high level of confidence to analyse technological practices.

Consequently, appropriate industries where patents of high quality are actively sought and companies that achieve significant benefits through patenting activities should be derived; resolving the issue of the applicability of patent analysis in technology management. Basically, the underlying idea of this research is based on sectoral innovation systems, which means that different sectors operate under different technological regimes characterized by specific combinations of opportunity, appropriability conditions and characteristics of knowledge bases.

The characteristics of innovation systems have been analysed at the industry level, wherein innovation is regarded as an interactive process involving various players, such as firms, universities, research centres, and government agencies. Thus, the role of persistence and heterogeneity of innovative activities at the firm level is explored in determining the patterns of technological change in different industries. Relations and networks are frequently regarded as core elements of innovation and production processes.25-27 The patent system and property rights have different effects on different sectoral systems as a consequence of the different features of the systems.28 Particularly, Pavitt defined the taxonomy of industries by considering unique

(3)

patterns of innovation and comparing different intellectual property strategies of each industry. For example, while supplier-dominated sectors such as agriculture and housing generally prefer trade secrets, innovations in scale-intensive sectors such as automobiles and steel can be protected by natural and lengthy technical lags in imitation (i.e. patents).29 If patent analysis can be used for technological forecasting in all sectors, researchers and practitioners can utilize patent information in their decision- making process with regard to R&D policies; without any uncertainty on the applicability of patent analysis.

However, many experts argue that since intellectual property right (IPR) strategies are different according to sectors as mentioned in the work of Pavitt, the use of patent information should be tailored to ensure the validity of patent analysis.30 The approach described in this article will be helpful to analyse and forecast technological activities by considering distinctive IPR strategies of diverse industries. Although many studies have employed patent analysis in order to investigate various aspects of R&D activities, including technological entry/exit and knowledge flow, researchers have made little effort to validate the application of patent data in order to enable more accurate technological forecasting and more sophisticated innovation analysis.

This paper aims at identifying suitable industries where patent information can be usefully applied in the activities of technology management, by conducting three analyses. First, differences among all sectors in devising patent strategy are examined to extract the sectors merit patent analysis. Second, sectors where patent-based technological forecasting is successfully able to anticipate technological trends are identified. Finally, the correlation between patent application and R&D investment (including patent application and profits through commercialization) is analysed to investigate the exploratory power of patent information with respect to R&D activities.

Research Framework

Data

This paper exploits the results of the community innovation survey (CIS) carried out by the national statistical office of Korean government, and patent data from the USPTO database in order to identify sectors where patent information can be used for technological forecasting. Although the CIS data are based on the innovative activity of companies in

Korea and patent data are collected from the USPTO database, the combined use of the two sources offers various advantages in terms of considering two different innovation databases. The CIS data are beneficial for understanding the variations across sectors and reveal the specific methods of protection of innovations in each sector. In this paper, the CIS data of 5241 Korean companies include the survey of innovation information collected in 2005 and 2006.

The CIS data cover the basic information of the firms, product and process innovation, innovation activity and expenditure, effects of innovation, source of information for innovation patents and so on. On the other hand, the patent data provide ample information for forecasting the future of technology and evaluating the technology. In this paper, information on the number of patent applications was collected in a time period ranging from 1985 to 2002. In addition, R&D expenditure and dollars of technology export in each sector were gathered between 1994 and 2007 in order to investigate their relationship with the number of patent applications.

Overall Process

The overall process of this research consists of four steps:

(i) To define the industrial categories to reflect an insightful perspective on technological trajectories.

Basically, since this paper concentrates on identifying appropriate sectors for which patent analysis can be applied, the definition of sectors is critical for deriving meaningful results and implications.

(ii) To scrutinize the various strategies of each sector for protecting its innovations. In general, companies in a specific sector adopt a specific type of IPR strategy from various options such as patents, trademarks, and secrecy.

(iii) To analyse the usefulness of patent analysis for technological forecasting by estimating errors in forecasting through growth curves. Although companies in a sector typically select the patenting process to safeguard their inventions, patent data may not be relevant for forecasting the future of technology in other sectors where other IPR strategies are viable for protecting the technological capabilities of those sectors.

(iv) To calculate the correlation between the number of patent applications and other indices, such as R&D investment and royalty income through technology export in each sector, to investigate the applicability of patent analysis.

(4)

Analysis

IPR Strategies of Industries

A company that accomplishes a technological invention normally wants to capture the rents from its innovation, viz., appropriability. Even though most technological innovations are inherently difficult for competitors to replicate, some of them might be easily copied or imitated, often leading to a critical lawsuit.

Therefore, companies seek for a system to protect their intellectual property; most countries offer regulations for legal protection in the form of patents, trademarks, copyright, and trade secret laws. However, sectoral differences in terms of the underlying features of the technological and industrial structure enable the adoption of diverse strategies across industries.

The differences between companies in each industry regarding the choice of IPR strategies were investigated by analysing the CIS data. For this, relevant data like production innovation, process innovation, patent, trade secret and time-to-market, from the abundant information of the CIS were utilized to scrutinize different IPR strategies of various industries. The proportion of each type in each industry was derived by summing the proportion of IPR types of companies which are included in a specific category, and ANOVA and t-test were employed to statistically validate the differences in such proportions over industries as well as in each industry, showing that patenting activity is more active in some industries than in others. In addition, the analysis of IPR strategies in product as well as process innovation was performed to examine the differences in strategies according to diverse types of innovation. Table 1 summarizes the parameters of the analysis.

Applicability of Patent Analysis in Technological Forecasting

Even if patent activity in a sector is relatively active, the forecasting power of patent analysis could be weak

in that sector. Forecasting accuracy with patent application data over time is therefore, estimated. First, patent data are collected from the database of the USPTO, focusing on historical patent application data in each sector. Second, the time-series data are divided into training data (for estimating parameters) and test data (for calculating the forecasting error). In this study, patent data was retrieved from 1985 to 2002. While the first 15 time-series data points from 1985 to 1999 were chosen as the training data, the latter three data points were adopted for examining forecasting errors by anticipating the number of patent applications on the basis of the derived trends. To this end, both linear and non-linear regressions were applied to fit the data into five growth curves (exponential, logistic, Gomperz, Bass, and Harvey models) and the mean average percentage errors (MAPEs) with regard to the test data were measured in order to calculate the forecasting errors in the estimation. The five growth-curves shown in Table 2 are representative growth-curve models for obtaining good performance in technological forecasting. If the ranks of each sector are consistent with regard to the forecasting accuracy, the contention – that patent information in those sectors that are highly ranked on the list can be significantly applied for technological forecasting – will be statistically validated.

The consistency of ranks is analysed with ‘Kendall’s W’

(also known as Kendall’s coefficient of concordance).

Since this coefficient provides a measure of total agreement within the group of observers, it is employed to measure the consistency of ranks of growth-curves.

Applicability of Patent Analysis in the Study of R&D Investment Efficiency

Several scholars have applied patent databases for analysing, evaluating, and planning activities related to technology management, including the investigation of the efficiency of R&D investment and technology valuation. However, in many studies that estimate the

Table 1  Parameters of the analysis

Subject Data Analytic methodologies

IPR

strategies CIS data t- test, ANOVA Forecasting

accuracy

Patent data Non-linear regression, ANOVA, non- parametric analysis (Kendall’s W) Correlation

analysis with other R&D data

Patent data, R&D investment, Dollars of technology export

Pearson’s correlation coefficient

Table 2  Regression equations of growth-curve models Models Regression Regression equations Exponential Non-linear Yt=Kβ eαt+εt

Logistic Non-linear t t t

e

Y K ε

α β +

= + 1

Gompertz Non-linear Yt=Keβeαt+εt

Bass Linear yt=β +βYt +β Yt +εt

2 1 2 1 1

0 ( )

Harvey Linear lnyt=β0+β1t+β2ln(Yt1)+εt

(5)

productivity of R&D investment, the time lag between R&D investment and the granting of patents is ignored.

Therefore, appropriate industries wherein patent analysis is applicable should be identified by using the time-series of R&D investment after applying the time lag. For this purpose, the dynamic analysis of the correlation between the two sets of data is conducted by changing the time lag between R&D investment and the output of technological innovation, such as patent applications, from 0 to 5 years. Thus, the time-series data of R&D investment from 1994 to 2007 were collected and patent data also retrieved during the same time period. The correlation analysis between the two sets of data is executed by calculating Pearson’s correlation coefficient.

Applicability of Patent Analysis in the Study of Technology Exploitation

Some researchers assert that the usefulness of patent analysis is often compromised because many companies file a large number of valueless patents.

They consider patent activity to be part of managerial strategies to earn profits from the resulting technological superiority. To clarify this, industries which made valuable patent applications to the patent office needed to be identified. The worth of technology export is a good proxy measure of technological quality because valuable patents can be exported abroad and earn considerable royalty. Since all types of incomes that patents generate in a business can be hardly estimated, the income through export of technology can be a clear indicator to explain patent quality and investigate the applicability of patent information in analysing technology exploitation activities. The data on the income of technology export and number of patent applications were collected from the Korean National Statistical Office and USPTO, respectively. In the second step, Pearson’s correlation coefficient was examined to analyse the relations between the two datasets.

Results

Industrial Classification for Innovation

Industrial classification has been implemented for the purpose of both research and industrial policy-making.

Although various classifications are suggested by governments, institutes, and so on, the conventional taxonomy is the international standard industry classification (ISIC), which suffers from having too many detailed categories and giving inadequate consideration to technological criteria. Thus, in this

study, the classification scheme of Pavitt, which is often employed for scrutinizing the characteristics of innovation, was adopted and revised to reflect changes in the economic situation, technological trends, and industrial structures. Pavitt suggested four main categories of industry based on innovation patterns:

(i) supplier dominated, (ii) scale intensive, (iii) specialized supplier, and (iii) science based; service sectors are excluded from his taxonomy. Even though patent activities generally concentrate on manufacturing sectors, the coverage of sectors needs to be expanded to include service sectors in order to allow the consideration of business method patents as a class of patents which disclose and claim new methods of doing business with technical approaches. The taxonomy of industry is defined such that it covers all potential industries by reflecting the taxonomy of Pavitt and the ISIC because Pavitt proposed a manufacturing sector oriented classification and the recent approach of ISIC complements the Pavitt’s classification by including service industry classifications.

As a result, the two main categories (manufacturing and service) can be divided into a total of nine detailed categories (supplier dominated, scale intensive, specialized supplier, biotechnology (BT), information technology (IT), business service, distribution service, financial service, and communication service). Table 3 presents the industrial classification of this research and the main industries in each category.

Sector-specific IPR Strategies

Most of the companies in the nine sectors tend to implement different IPR strategies according to the basic characteristics of their industries. ANOVA and

Table 3  Proposed industrial classification Super-level Sub-level Main industries

Supplier dominated

Agriculture, textiles Scale intensive Power systems, motors Specialized

supplier

Medical instruments, measurement Biotechnology Drugs, biotechnology Manufacturing

Information technology

Semiconductor, computer

Business service R&D, professional services

Distribution service

Wholesale, retailing services

Financial service

Banking, insurance services

Service

Communication service

Postal,

telecommunication services

(6)

t-test of CIS data were applied to investigate the unique IPR policies of each sector. The difference of IPR policies among sectors was statistically tested through ANOVA. If the p-value of ANOVA is lower than the significant level (p=0.05 in this research), there are significant differences among the aforementioned sectors. The results are presented in Table 4, indicating that the importance of IPR strategy (patents, trade secrect, and time-to-market) varies significantly across sectors in relation to both product and process innovation. In addition, three sectors (scale intensive, specialized supplier, and information technology) were inclined to maintain technological superiority in the form of patents than in other forms, in comparison with other manufacturing sectors and all the service sectors. Regarding sector-wise analysis, the supplier dominated, distribution service, financial service, and communication service sectors normally protected their technological capability through trade secrets than through patents in product innovation by t-test. On the contrary, the specialized supplier and information technology sectors actively engage in patenting activity. In the scale intensive, biotechnology,

and business service sectors, there are no significant differences between the two strategies. However, regarding process innovation, nearly all the sectors (supplier dominated, scale intensive, specialized supplier, distribution service, financial service, and communication service) adopted the trade secret strategy because process innovation can be secured better through trade secrets than patents, implying that non-technical innovations such as processes could be easily imitated by competitors if patented. Consequently, patent information has an analytical power when it is applied in the analysis of product innovation in the scale intensive, specialized supplier, biotechnology, and information technology sectors.

Sectoral Differences in the Forecasting Power of Patent Analysis

Patent analysis should be appropriately employed to forecast the future of technology. Thus, specific industries that exhibit high predictability in trends in technology development need to be identified by calculating the forecasting accuracy of patent analysis in each industry. Table 5 depicts the forecasting accuracy of each sector through five growth curves

Table 4  Importance of IPR strategies

Product innovation Process innovation

Categories No of

firms

Patents Trade secrets

Time-to- market

p-value (t-test)

Patents Trade secrets

Time-to - market

p-value (t-test)

Supplier dominated 87 2.21 2.69 2.66 0.00* 2.48 3.14 3.01 0.00*

Scale intensive 272 2.90 2.85 2.85 0.66 2.46 2.88 2.75 0.00*

Specialized supplier 230 2.94 2.57 2.69 0.00* 2.23 2.48 2.34 0.01*

Biotechnology 27 2.65 2.43 2.74 0.51 2.22 2.59 2.41 0.15

Information technology 168 3.12 2.70 2.75 0.01* 2.50 2.68 2.60 0.11

Business service 218 1.10 1.31 0.91 0.14 0.50 0.71 0.55 0.10

Distribution service 282 0.31 0.69 0.63 0.00* 0.10 0.43 0.40 0.00*

Financial service 95 0.31 1.14 0.95 0.00* 0.04 0.85 0.57 0.00*

Communication service 299 0.99 1.26 1.27 0.00* 0.38 0.81 0.68 0.00*

p-value (ANOVA) 0.00* 0.00* 0.00* 0.00* 0.00* 0.00* 0.00*

* indicates that the p-value is significant at the 95% level

Table 5  Forecasting accuracy in each sector MAPE

Categories Exponential

model

Logistic model

Gompertz model

Bass model

Harvey model

Average rank

Kendall consistency coefficient

Supplier dominated 0.36 0.33 0.05 0.07 0.30 6.4

Scale intensive 0.30 0.27 0.04 0.01 0.04 3

Specialized supplier 0.57 0.38 0.05 0.02 0.14 5.4

Biotechnology 0.25 0.23 0.03 0.03 0.07 2.8

Information technology 0.79 0.05 0.06 0.01 0.10 4.4

Business service 0.64 0.61 0.08 0.05 0.21 7

Distribution service 0.05 0.04 0.01 0.05 0.14 3

Financial service 0.84 0.77 0.12 0.09 0.39 9

Communication service 0.34 0.22 0.03 0.07 0.08 3.8

0.026*

* indicates that the p-value is significant at the 95% level

(7)

with patent information, which reveal the MAPE of forecasting. The concordance of ranks is analysed by employing ‘Kendall’s W’. Sectors that show high forecasting accuracy are the biotechnology and information sectors; this result is similar to that of the analysis of IPR strategies. However, although the communication service sector features a low proportion of firms in patenting, its forecasting accuracy is greater than that of the other manufacturing sectors. This indicates that the communication service sector has several firms that protect innovations through patents, and that the intrinsic features of the service are mostly based on technologies or products. The hypothesis that MAPE varies across sectors is statistically significant through ANOVA (p-value: 0.003) and the ranks of the nine sectors in forecasting accuracy are consistent in the forecast through the five growth curves (p-value of Kendall’s W: 0.025).

Relation between Patent Applications and R&D Investment

R&D investment enables a company to execute a R&D project, allowing the invention of one or more patents. However, R&D efforts often do not accelerate patenting activity, leading to a low degree of correlation between the time-series of patent applications and R&D investment. Thus, if the appropriate sectors where the correlation coefficient is significantly high can be identified, the applicability of patent analysis probably will be enhanced. Table 6 presents the output of this analysis, indicating that in three sectors (specialized supplier, biotechnology, and information technology), the correlation coefficient between R&D investment and patent application is significantly high. Patent analysis related to R&D investment in these sectors would be very valuable

because patent information is closely related to R&D inputs. In addition, the result implies that while the time lag in the specialized supplier and information technology sectors is two years, it is just one year in the biotechnology sector.

Relation between Patent Information and Technology Export Information

Technology valuation can be depicted as a kind of technological forecasting in that it must estimate the future value of technology. Even if the quality of technology is insufficient for patenting, many companies convert their technological capabilities to patents in order to gain a positive position, as a sort of technological strategy. Thus, the value of patents must be ascertained to distinguish high-quality patents from valueless patents. In this study, royalty income from technology export is considered to be a good proxy measure for that. Correlation analysis between patent applications and the income from technology export was also conducted. Consequently, the scale intensive, specialized supplier, and information technology sectors were appropriate in the analysis when patent data were used to investigate the effects and values of technology exploitation. Table 6 depicts the results on the correlation coefficient between patent applications and income from technology export.

Discussion

The various analyses in this article suggest that there are specific situations where it is appropriate to apply patent analysis. First, the technological capability from product innovation is most commonly protected through patents unlike that from process innovation. Thus, patent information can be employed to analyse the characteristics of the technologies that

Table 6  Correlation analysis with R&D data

Pearson’s correlation coefficient with R&D investment (considering a time-lag in years) Categories

0Y 1Y 2Y 3Y 4Y 5Y

With technology export

Supplier dominated 0.191 0.027 0.212 -0.104 -0.225 -0.126 0.203

Scale intensive 0.334 -0.027 -0.438 -0.376 0.380 0.982 0.671*

Specialized supplier 0.585* 0.0710* 0.744* 0.712* 0.531 0.812 0.571*

Biotechnology 0. 390* 0.577* 0.007 0.061 0.662 -0.294 0.144

Information technology

0.759* 0.780* 0.947* 0.868* 0.697 0.906 0.718*

Business service 0.521 0.085 -0.516 -0.686* -0.945* -0.910 -0.643

Distribution service 0.437 0.194 0.539 0.341 0.211 0.182 -0.472

Financial service 0.632 0.487 -0.391 0.573 0.439 0.462 -0.357

Communication service

0.450 0.373 0.448 0.091 0.772 0.539 -0.227

* indicates that the p-value is significant at the 95% level

(8)

are associated with product innovations. Second, from the analysis of IPR strategies, the specialized supplier and information technology sectors were found to be generally suitable for patent analysis because many companies wished to get their technologies granted as patents. Third, the forecasting power based on patent information in the biotechnology, scale intensive, information technology, distribution service, and communication service sectors is strong. The trends in patent application in these sectors are more or less predictable because either the IPR strategies of these sectors are dominated by patents or the sectors are technology-based. Fourth, patent analysis in service sectors is not appropriate for analyses of R&D investment or technological quality because patent application in the sectors is too strategic to be adopted in such analyses. Finally, contrary to common belief, patent analysis in manufacturing sectors is generally appropriate. However, since patenting activity in the supplier dominated sector is not significant, this sector needs to be excluded from the set of sectors for which patent analysis can be employed. In the case of the biotechnology sector, while patent applications and R&D investment is highly correlated, the relationship between patent applications and income from technology export is weak, implying that biotechnology is strategically patented in the perspective of loyalty and that it is still at a nascent stage of development. Thus, patent analysis in this sector should be carefully employed for investigating the profits from technology development.

Conclusion

This paper aims at identifying appropriate sectors where patent analysis can be employed for technology analysis, thereby enabling analysts to confidently utilize patent analysis in engineering management.

For this purpose, the principal data, such as CIS data and patent data, were collected from the USPTO, and various statistical analyses such as non-linear regression, correlation analysis, ANOVA, and t-tests were executed with the new industrial classification.

Consequently, manufacturing sectors were more relevant than service sectors for the use of patent analysis. In particular, the biotechnology, specialized supplier, and information technology sectors can derive great benefits from patent-based technology analysis. Although the proportion of firms that use a patenting strategy in the communication service sector was low in comparison with most of the

manufacturing sectors, this sector exhibited the highest potential among the service sectors with regard to the applicability of patent analysis. In addition, the forecasting accuracy of the sector is relatively high, which suggests the utility of patent analysis for technology analysis.

However, since this paper tackles a very complex issue on protecting innovations, there are several limitations in terms of data collection and interpretation. First, since different databases are used to analyse the applicability of patent analysis, the integration of results is difficult, hampering the incorporation of implications. While CIS data are based on the survey conducted by the Korean government, patent data are retrieved from the USPTO database. Second, only patent statistics are investigated for evaluating the usefulness of patent analysis; full sets of patent information are not considered. This paper limits the range of patent information to the number of patent applications over time. Thus, unfortunately, content analysis on patent documents is excluded in this paper. Third, indices for evaluating patent-based technology analysis are still insufficient to ascertain that the proposed approach is valuable. In this paper, only two measures, R&D investment and income from technology export, are used to examine how practical the proposed approach is.

Therefore, future research needs to address the aforementioned issues. Firstly, the differences between the USPTO, EPO, and JPO could be explored to find appropriate patent systems where patent analysis can be effective for technology analysis. This process might also involve a method for adjusting the original approach to enhance the forecasting accuracy in each patent system. Also, patent documents can be analysed to identify relevant sectors to which technological forecasting can be successfully applied. Since patent documents contain enormous amounts of information on products as well as technologies, the text mining of such technological documents can yield profound insights for identifying potential technology development and visualizing trends in technology. Therefore, it will be useful to clarify the right sectors where text mining- based technological forecasting can be employed.

Finally, a number of indices need to be proposed for validating the usefulness of patent analysis in technology analysis. For example, patent quality can be measured by additional indicators such as overall profits summing the income from technology export and revenue of products that are developed with patented technologies.

(9)

The results of this paper can provide a strong rationale for analysts who intend to use patent analysis in forecasting the future of technology. In practice, since some sectors that might show poor potential with regard to the utility of patent analysis for technological forecasting can be screened out, the expected costs of conducting technological forecasting obviously will be reduced. In addition, the differences of intellectual property activities across various industries can allow researchers to understand an idiosyncratic technological strategy of firms. Moreover, in the perspective of government, the aforementioned results can be reflected in a process of making a technological policy. For example, a different time lag between R&D investment and fruitful R&D outputs/outcomes among sectors needs to be applied to effectively draw technology roadmaps. In particular, policy makers can derive relevant industries where a patent activity is substantially conducted to gain competitive edge without developing a tangible product, earning technological loyalty.

Acknowledgement

This research was supported by the Basic Science Research Program through the National Research Foundation (NRF) and funded by the Ministry of Education, Science, and Technology (Grant no: 2009- 0073285).

References

1 Perez Carlota, Technology change and opportunities for development as a moving target (UNCTAD, New York), 1999, http://www.unctad.org/en/docs/ux_tdxrt1d9.en.pdf (1 August 2011).

2 Griliches Z, Patent statistics as economic indicators: A survey, Journal of Economic Literature, 28 (4) (1990) 1661-1707.

3 Archibugi D and Pianta M, Measuring technological change through patents and innovation surveys, Technovation, 16 (9) (1996) 146–451.

4 Coates V et al., On the future of technological forecasting, Technological Forecasting and Social Change, 67 (1) (2001) 1–17.

5 Martino J P, Technological Forecasting for Decision Making (McGraw-Hill, New York), 1993.

6 Mann D L, Better technology forecasting using systematic innovation methods, Technological Forecasting and Social Change, 70 (8) (2003) 779–795.

7 Zhu D and Porter A L, Automated extraction and visualization of information for technological intelligence and forecasting, Technological Forecasting and Social Change, 69 (5) (2002) 295-506.

8 Porter A, in Futures Research Methodology, edited by J C Glenn and T J Gordon (American Council for the UNU, Washington DC), 2003.

9 Watts R J and Porter A L, Innovation forecasting, Technological Forecasting and Social Change, 56 (1) (1997) 25–47.

10 Ernst H, Patent applications and subsequent changes of performance: evidence from time-series cross-section analyses on the firm level, Research Policy, 30 (1) (2001) 143–57.

11 Paci R, Sassu A and Usai S, International patenting and national technological specialization, Technovation, 17 (1) (1997) 25-38.

12 Tong X and Frame J D, Measuring national technological performance with patent claims data, Research Policy, 23 (2) (1994) 133-141.

13 Narin F, Noma E and Perry R, Patents as indicators of corporate technological strength, Research Policy, 16 (2-4) (1987) 143-155.

14 Jeon J, Lee C and Park Y, How to use patent information to search potential technology partners in open innovation, Journal of Intellectual Property Rights, 16 (5) (2011) 385-393.

15 Phadnis R and Hirwani R, Patent analysis as a tool for research planning: Case study of phytochemicals in tea, Journal of Intellectual Property Rights, 10 (3) (2005) 513-524.

16 Hirschy M and Richardson V, Valuation effects of patent quality: A comparison for Japanese and US firms, Pacific- Basin Finance Journal, 9 (1) (2001) 65-82.

17 Tsuji Y S, Organizational behavior in the R&D process based on patent analysis: Strategic R&D management in a Japanese electronics firm, Technovation, 22 (7) (2002) 417-425.

18 Gassmann O, Patentmanagement: Innovationen erfolgreich nutzen und schutzen (Springer, Berlin), 2007 (in German).

19 The Measurement of Scientific and Technological Activities:

Using Patents as Science and Technology Indicators—Patent Manual (OECD, Paris), 1994.

20 Yoon B and Park Y, A text-mining-based patent network:

Analytical tool for high-technology trend, Journal of High Technology Management Research, 15 (1) (2004) 37–50.

21 Terapane J F, A unique source of information, Chemtech, 8 (5) (1978) 272–276.

22 Liebesny F et al., Scientific and technical information contained in patent specifications - Extent and time factors of its publication in other forms of literature, Journal of the American Society for Information Science, 8 (4) (1974) 165–177.

23 Pavitt K, Patent Statistics as indicators of innovative activities: Possibilities and problems. Scientometrics, 7 (1-2) (1985) 77-99.

24 Hall B, Jaffe A and Trajtenberg M, Market Value and Patent Citations: A First Look (NBER, Cambridge, MA), 2000.

25 Malera F, Orsenigo L and Peretto P, Persistence of innovative activities, sectoral patterns of innovation and international technological specialization, International Journal of Industrial Organization, 15 (1997) 801-826.

26 Edquist C and Jacobsson S, in Systems of Innovation, edited by C Edquist (Frances Pinter, London, UK), 1997.

27 Freel M S, Sectoral patterns of small firm innovation, networking and proximity, Research Policy, 32 (5) (2003) 751-770.

28 Levin R et al., Appropriating the returns from Industrial R&D, Brookings Papers on Economic Activity, 3 (1987) 783–831.

29 Pavitt K, Sectoral patterns of technical change: toward a taxonomy and a theory, Research Policy, 13 (6) (1984) 343-373.

30 Chang J and Zhu X, Bioinformatics databases: Intellectual property protection strategy, Journal of Intellectual Property Rights, 15 (6) (2010) 447-454.

References

Related documents

Various types of the patent-related infor- mation available on such web sites are as follows: (i) infonnation about the industrial property offices, (ii) irformation

Of those who have used the internet to access information and advice about health, the most trustworthy sources are considered to be the NHS website (81 per cent), charity

Harmonization of requirements of national legislation on international road transport, including requirements for vehicles and road infrastructure ..... Promoting the implementation

Fig 3-01.02: Number of public sector undertakings under respective independent departments and Ministry of Science and Technology of the Central Government Ministry undertaking

In recent years, an increasing number of studies have used patent data to analyze innovation and international technology diffusion, in particular in the

The extensive literatures on studies 11 have examined various aspects of impact of product patent protection on Indian pharmaceutical industry such as, incentives for

A Contributory infringement occurs when any person, without authority from the patentee sells or offers to sell within the patent granting States, or imports in such States,

The results shows the impact of foreign investment on Research and Development (R&D) which is indicator of Technological efficiency While testing the hypothesis with regard