Patent data analysis of innovation behaviours in technology hardware & equipment industry

Patent data are widely used in literature as a measure of innovation, being the only formally and publicly verified output of inventive activities and. we use patent data to study innovation strategies related to exploitative vs. explorative activities, closed vs. open processes and incremental vs. radical outputs. in this paper, we propose an integrated framework based on the combination of such variables, in order to analyse capabilities, activities and competencies related to r&d processes. twelve different behaviours were detected considering four dimensions of innovative processes. if combined together, such Abstract


Introduction
Patent data are widely used in literature as a measure of innovation, being the only formally and publicly verified output of inventive activities and.we use patent data to study innovation strategies related to exploitative vs. explorative activities, closed vs. open processes and incremental vs. ________________________________________________________________________________________________________________ ______________ Francesca Michelino, Antonello Cammarano, Emilia Lamberti and Mauro Caputo (2015), Journal of Innovation & Business Best Practice, DOI: 10.5171/2015.316698behaviours describe the overall innovation strategy of a company.our research question is: how innovation strategies of companies can be analysed and technological evolution can be tracked through patent statistics? the research is based on objective data detected from patstat database, and on some variables already acknowledged and operationalized in scientific literature, that we combine in an integrated framework.the output of our analysis consists of a map summarising the strategies of companies towards innovation and providing information about the prevailing innovation paths (firm level).the framework can be also applied for mapping innovation strategies in specific industries and describing their technological evolution (industry level).
the methodology is tested on a sample of 133 r&d intense companies from technology hardware & equipment (the) industry, by analysing over 300,000 patents applied from 2008 to 2012, validating both the framework applicability and its explicative power and usefulness.from the analysis of the behaviours of companies, exploitation strategies that lead to radical outputs seem to be the most relevant within the sample.we also detected a growing adoption of open innovation and an increasing concentration of r&d efforts on radical outputs. in what follows, after a brief literature review on patent-based metrics for innovation, the measurement framework is presented and then applied to the sample.results are discussed and conclusions will close the work.

Literature Review
A large number of scholars have used patent data as a proxy for innovation (Acs and Audretsch, 1989;Chakrabarti, 1991;Grupp, 1992;Belderbos, 2001;Frietsch and Grupp, 2006;Hanel, 2006).Patent data offer a valuable source of information, useful to both keep track of the technological strategy evolution of companies and make comparisons, as they contain standardized data, stored for a long period and continuously updated (Griliches, 1990).Furthermore, patents are the only formally and publicly verified outputs of inventive activities (Ma and Lee, 2008).
The integrated framework we suggest is based on the combination of variables already acknowledged in scientific literature concerning: 1) exploitative vs. explorative activities, 2) closed vs. open processes and 3) incremental vs. radical outputs.In what follows, a deepening of the operationalization of each dimension under investigation is reported.Since March's (1991) work, a wide debate has raged over the need for a balance between exploiting the knowledge an organization already holds (local search) and exploring for knowledge that is different and new to the organization (distant search).Exploitation is associated with current viability and thus leads to more capability at current activities, while exploration is related to the acquisition of diverse and novel body of knowledge that will serve as the seed for future technological developments.Such concepts can be operationalized through the investigation of patent classification codes: international patent classification (IPC) codes allow analysing the technology field in which patents impact and can be considered as a proxy of skills developed by the firm in a specific technology domain.A patent is considered as an explorative one when it is situated in a technology domain that is new or unfamiliar to the firm, i.e. the firm did not patent in the technology domain in the past five years (Belderbos et al., 2010 Further, although many contributions aimed at identifying the evolution of technological patterns of companies, they show only a partial overview of the innovation strategies pursued, e.g.Suzuki and Kodama (2004) defined technological trajectories and technology diversification strategies by analysing IPCs.Therefore, an integrated patent-based map of innovation capabilities, processes and competencies seems to be lacking.In this paper we aim at investigating patent data after a multidimensional point of view, in order to: 1) analyse the whole innovation process in terms of capabilities, activities and competencies, 2) keep track of the evolution of such innovation process and 3) study the innovation strategies of companies.

Methodological Framework
Starting from the literature review, we designed a framework that combines all the aforementioned variables with the aim of defining the innovation strategies adopted by companies after a multidimensional perspective.By simulating innovation through an input-process-output model, we believe that innovation strategies are pursued through management choices on capabilities, activities and competencies.In particular, capabilities are considered as the input of our model and can be related to exploitation and exploration strategies on ________________________________________________________________________________________________________________ ______________ Francesca Michelino, Antonello Cammarano, Emilia Lamberti and Mauro Caputo (2015), Journal of Innovation & Business Best Practice, DOI: 10.5171/2015.316698 each technological domain in which the firm is involved.Activities are linked to the organization of R&D efforts and are here summarised by the choice of collaborating or not with other firms.Finally, the patent, which is the concrete manifestation of competencies developed by the company at the end of the innovation process, can be considered as a proxy of radical or incremental innovation, depending on the potential pioneering of the output.
A fourth dimension is added to the inputprocess-output model: the relevance of the process, that can be defined high if the capabilities which gave raise to the process are core, low otherwise (Figure 1) .

Figure1: Input-process-output model for simulating innovation strategies
The starting point of our work is the extraction of patents from PATSTAT database: for each analysed company we considered all its patents applied in the investigated time interval, and recorded patent classification codes, number of applicants and number of backward citations, in order to analyse capabilities, activities and competencies.
As suggested by scientific literature, technology fields can be analysed by detecting IPCs recorded in patent applications.In our framework, we used the cooperative patent classification (CPC) system, a new patent classification systemjointly developed by the European Patent Office and the United States Patent and Trademark Office -which can be considered as an evolution of IPC, since it is more specific and detailed.At least one CPC is related to a patent application, in order to define the technological areas on which patents have impact.CPCs categorize technological fields into a five-level hierarchical system, from the broadest to the very specific: section, class, subclass, main group and subgroup.For example, the "details of semiconductor or other solid state devices" field belongs to a main group expressed in an alpha-numerical code, H01L23: H identifies the section "electricity", H01 the class "basic electric elements", and H01L represent the subclass "semiconductors devices or other solid state devices".
Finally, the subgroup is represented by adding a slash symbol (/) and numerical digits, e.g."H01L23/02" specifically defines patents related to "containers for semiconductors devices or other solid state devices".In order to identify the technology field, we decided to cut the code and consider only the information before the slash, since the operationalization of the variable capability clearly requires more generalization.After data extraction, for each company we obtained a list of all the CPCs detected in the patents it filed in the selected time horizon.CPCs can be labelled as core or non-core and exploitative or explorative.In particular, each CPC is defined core if it is declared in at least 10% of the patents filed in the previous five years, non-core otherwise; exploitative if the company filed patents in such technology domain in the past five years, explorative otherwise.Obviously, from these two definitions, no core and explorative CPCs can be found.In the company filed more than 10% of its patents of the previous five years in the technology domain described by the CPC non-core exploitative the company filed some patents, but less than 10% of those registered in the previous five years, in the technology domain described by the CPC non-core explorative the company did not file any patent in the previous five years in the technology domain described by the CPC Further, by analysing patents that declare the specific technological field and detecting the number of owners and backward citations, we can define their nature as (see Table 2): Given that a CPC can be detected in more than one patent for each company, both closed and open patents, as well as both incremental and radical ones, can be found, i.e. the competencies can be used by companies in both their closed and their open innovation processes and can give raise to both incremental and radical innovation.
Thus, our framework describes the innovation processes adopted for each CPC through four dimensions.Each patent that contains the analysed CPC is described with four different labels, the first two inherited from the belonging CPC, and associated to only one of the twelve available different behaviours in R&D processes explained in Figure 2. technological class is used to study the overall behaviour of a firm, summing the results obtained from all the CPCs.Therefore, our framework can evaluate the weight of a single behaviour on the mix of innovation strategies of companies in a specific time interval, i.e. the map of innovation processes.
A share indicator that summarises the impact of a specific combination on the overall innovation strategy describes each behaviour.
Two of the most discussed trade-off in the scientific literature, are exploration vs. exploitation and radical vs. incremental innovation strategies.
Through our framework, we can evaluate the adoption of such strategies, or their mix (Table 3).Starting from the behaviours described through the capabilities-competencies matrix, by adding the information about relevance and process organization, we can define a simple nomenclature for the twelve different combinations (Table 4).Further, by selecting a sample of companies, the framework provides information about innovation in specific industries, allowing to perform an industry level analysis.

Label
By comparing results obtained in different time intervals, we can study the innovation paths undertaken by companies and find the continuous innovation strategies of companies or industries, verifying the evolution on the adoption of the twelve different combinations.

Findings
The framework was applied to a sample of 133 R&D intense companies from THE industry ranked by their investment in R&D, according to The 2012 EU Industrial R&D Investment Scoreboard (JRC, 2012), excluding the firms whose 2012 annual reports were not available and those for which the list of subsidiaries was not found in such documents.We choose this industry because it is R&D intense, uses patents as a means of appropriation of innovation (Pavitt, 1984) and is characterised by companies incorporating strategic technology alliances into the core of their technology strategies.In order to consider the impact of R&D activities on the corporate group, we searched patents developed by both the parent company and its subsidiaries disclosed in annual reports, also taking into account patents related to acquired companies and applied after the acquisition.We considered five years of analysis -from 2008 to 2012 -and gathered data from PATSTAT database, downloading patents applications from 2003 to 2012 and analysing 316,015 documents.As a matter of fact, for each year of analysis we have to consider also the patents in the previous five years in order to define the relevance and the exploitation vs. exploration of capabilities (Table 5).For example, for the year 2012 we downloaded all patent applications of companies, including documents related to their subsidiaries, identifying the CPCs registered and verifying if they were core/non-core and exploitative/explorative by analysing patents data from 2007 to 2011 and recording information about number of applicants and backward citations.We applied our framework in order to estimate the overall innovation strategy of each company and, by comparing results obtained from a group of firms, a benchmark is available.Table 6  Strengthening and advancement strategies are the most relevant ones, and in particular, the latter was detected in over half of the patent applications.In the analysed industry -where the development pace is very fast and the life cycle of products is short, exploitation strategies are strongly preferred.Furthermore, radical innovation seems to strongly characterize this industry, with about 54% of R&D activities leading to a radical output and patents without backward citations.

Year of analysis
In order to detect the evolution of the technological patterns of the sample and analyse the innovation strategies, we applied the framework, year by year, on patent applications from 2008 to 2012 and studied the result obtained on the eight cumulative share indicators.In Figure 3  are reported, being those of non-core, exploitation, closed and incremental complementary to the former.In the fiveyear period the shares of exploration and core innovation do not vary significantly, the former always being less than 4% and the latter on average around 14%.A continuous leveraging of existing capabilities through exploitation processes is observed.A growth in the adoption of open strategies is observed from 17% to 28% patents of the sample being registered by two or more applicants.Finally, the growth of the radical share indicator from 39% to 54% denotes an increased concentration of activities towards outputs that are pioneering for the companies of the sample.Only in 2012 radical outputs prevail on incremental ones.

Conclusions
The paper aims at contributing to the current literature on innovation management by describing how companies manage technological evolution and organize R&D activities, through the use of patent data.A capabilities-competencies matrix is defined, pointing out the differences between exploitation vs. exploration strategies and incremental vs. radical outputs.Noteworthy, these concepts are often treated alternatively in current literature but they can be evaluated separately since they describe two different dimensions of innovation activities.Mapping innovation strategies on the basis of patent applications, allows to consider the direct outcomes of the inventive process and, more specifically, of those inventions which are expected to have a commercial impact.
The paper addresses the need for operative, practical instruments, which can help managers to monitor and control their innovation paths.Given the availability and objectivity of patent documents, studying innovation through the analysis of patent data can help decision-makers to assess the status of their own strategies and compare it over time and space, also allowing the benchmarking with competitors.
A sample of 133 companies from THE industry over the period2008-2012 was studied, validating both the framework applicability and its explicative power and usefulness.Advancement strategies seem to be characteristics in such industry, while exploration activities are negligible as a whole.Two operationalization issues arise.First, the results are affected by our definition of core and non-core activities -CPCs are considered core if they are declared in at least 10% of the patents filed in the previous five years -as well as by the decision of cutting CPCs without considering the subgroup number, in order to avoid excessive detail on the definition of the capabilities of companies.Second, as to the definition of exploitative activities, companies may lose experience if they did not patent in a specific technology domain in a previous time interval lower than five years, since in THE industry the development pace is faster and the life cycle of product is shorter.Therefore considering a shorter time interval may probably improve the definition of exploitative and explorative R&D processes.We preferred to follow the approach already acknowledged in scientific literature but our consideration suggests a deepening of the operationalization of exploration and exploitation variables in such industry.Some general limitations can be defined for the work.First, the use of patent data as a proxy of technological activities might underestimate the phenomenon, since not all R&D efforts will result in an application for a patent.Second, the use of patent data for investigating the adoption of open innovation could be questionable, since not all collaborations will be captured by copatenting activities (Hagedoorn et al., 2003)

Figure 2 :
Figure 2: Twelve different combinations of labels defining specific behaviours in R&D processesAt a firm-level, for each company we obtained the number of patents related to each configuration: the combination of Behaviour Core -Exploitation -Closed -Incremental Core closed strengthening Non-core -Exploitation -Closed -Incremental Non-core closed strengthening Core -Exploitation -Open -Incremental Core open strengthening Non-core -Exploitation -Open -Incremental Non-core open strengthening Non-core -Exploitation -Closed -Radical Core closed advancement Core -Exploitation -Closed -Radical Non-core closed advancement Core -Exploitation -Open -Radical Core open advancement Non-core -Exploitation -Open -Radical Non-core open advancement Non-core -Exploration -Closed -Incremental Closed expansion Non-core -Exploration -Open -Incremental Open expansion Non-core -Exploration -Closed -Radical Closed explosion Non-core -Exploration -Open -Radical Open explosion Therefore, our framework supports us in identifying innovation strategies of firms in a specific time interval and provides a useful instrument for benchmarking.Journal of Innovation & Business Best Practice ________________________________________________________________________________________________________________ _____________ Francesca Michelino, Antonello Cammarano, Emilia Lamberti and Mauro Caputo (2015), Journal of Innovation & Business Best Practice, DOI: 10.5171/2015.316698 the trends for core, exploration, open and radical shares Journal of Innovation & Business Best Practice ________________________________________________________________________________________________________________ _____________ Francesca Michelino, Antonello Cammarano, Emilia Lamberti and Mauro Caputo (2015), Journal of Innovation & Business Best Practice, DOI: 10.5171/2015.316698