Definition of technology convergence

Since Rosenberg (Reference Rosenberg1963) coined the term convergence, scholars have defined it differently. There remains little consensus about how convergence should be defined (Nordmann, Reference Nordmann2004); thus, despite its frequent use, ‘convergence’ is more a buzzword than a recognized concept (Curran & Leker, Reference Curran and Leker2011). To generalize the fundamental definition of various researchers, convergence can denote a blurring of boundaries between at least two previously disparate areas of science, technology, markets, or industries (Hacklin, Reference Hacklin2008; Curran & Leker, Reference Curran and Leker2011; Karvonen, Lehtovaara, & Kässi, Reference Karvonen, Lehtovaara and Kässi2012). Through this process of merging heterogeneous knowledge, products, or markets, a new segment is created in which interchangeability and connectedness between the respective areas increase, and correspondingly a new value is created within the blurred boundaries (Curran & Leker, Reference Curran and Leker2011).

A similar concept has also been denoted as ‘fusion’ or ‘interdisciplinarity’ in the literature (Curran & Leker, Reference Curran and Leker2011). According to Curran and Leker (Reference Curran and Leker2011), fusion involves the similar concepts as convergence, with the terms often seeming to be interchangeable, whereas interdisciplinarity can be considered a subsegment of convergence (Curran & Leker, Reference Curran and Leker2011). However, convergence, in a strict sense, differs from fusion (Kodama, Reference Kodama1992) in that the latter creates subsegments of the original segment, whereas the former creates a new presence (Curran & Leker, Reference Curran and Leker2011). In addition, the use of interdisciplinarity is limited to specific fields of academic research (see Qin, Lancaster, & Allen, Reference Qin, Lancaster and Allen1997)Footnote ¹, whereas convergence comprises not only scientific knowledge but also technological knowledge, market, and industry (Curran & Leker, Reference Curran and Leker2011).

Convergence can be refined into the following three subcategories based on the process of developing new products/services (Curran & Leker, Reference Curran and Leker2011): (i) science convergence merges different scientific disciplines or areas, (ii) technology convergence combines the technologies of different application areas, and (iii) industry convergence unites sets of companies that have different technology bases, application fields, and target groups in various markets. However, this typology does not imply that convergence within each subcategory is independent. Scholars posit that convergence occurs through a sequential path from knowledge/science to technology and market/industry (Hacklin, Reference Hacklin2008; Curran & Leker, Reference Curran and Leker2011; Karvonen et al., 2012). That is, technology convergence occurs in interaction with science convergence and triggers industry convergence (Katz, Reference Katz1996; Hacklin, Reference Hacklin2008; Curran & Leker, Reference Curran and Leker2011).

Organizational context for technology convergence

Technology convergence implies the integration of different technological elements (Kodama, Reference Kodama1991, Reference Kodama1995); technological changes derived from such convergence play a key role in reforming the associated industry (Rosenberg, Reference Rosenberg1983; Katz, Reference Katz1996). Scholars have thus given considerable attention to the advent of technology convergence and have explored from diverse perspectives the contexts in which its underlying activities (i.e., R&D) are carried out. In particular, researchers have focused on institutional obstacles and structural aspects, such as the use of converging technology and convergence incentives, based on the idea that organizational context affects the willingness to promote convergence and determines its success (Gibbons, Limoges, Nowotny, Schwartzman, Scott, & Trow, Reference Gibbons, Limoges, Nowotny, Schwartzman, Scott and Trow1994; Klein, Reference Klein1996; Llerena & Meyer-Krahmer, Reference Llerena and Meyer-Krahmer2003; Stokols et al., Reference Stokols, Misra, Moser, Hall and Taylor2008).

By incorporating the aforementioned factors and concerns, this paper reviews how differences in organizational context (according to triple helix organizational type) influence technology convergence. By focusing on how the context of organizational type can differ when an organization forms an R&D alliance, it thus examines how strategic collaboration within the triple helix affects the advent of technology convergence.

How organizational context within the triple helix affects technology convergence

In general, R&D can be categorized as either explorative or exploitative. Exploration refers to the search, discovery, and development of new knowledge, whereas exploitation indicates the refinement, extension, and intelligent use of existing competencies (March, Reference March1991; Bercovitz & Feldman, Reference Bercovitz and Feldman2007). Universities, industry, and government (i.e., the R&D organizations in the triple helix) differ in terms of the characteristics of their R&D activities (Lee, Bae, & Lee, Reference Lee, Bae and Lee1991; Wernerfelt, Reference Wernerfelt1995; Etzkowitz & Leydesdorff, Reference Etzkowitz and Leydesdorff2000; Eom & Lee, Reference Eom and Lee2010).

Public research institutes such as universities and government research centers typically conduct R&D related to exploration. Specifically, universities aim to discover new scientific/technological knowledge, and government research institutes focus on the adaptation of advanced technology and the development of applied technology for introduction to industry. In contrast to these public research institutes, private firms chiefly perform R&D activities related to exploitation; in other words, they concentrate on developing technology for their own use (Lee, Bae, & Lee, Reference Lee, Bae and Lee1991; Bercovitz & Feldman, Reference Bercovitz and Feldman2007).

When developing technology, the ultimate risk borne by each sector varies because of sector-specific purposes and the innate characteristics of R&D activities. After development concludes, firms usually need to carry out and pay for commercial activities such as marketing and manufacturing (Håkanson, Reference Håkanson1993; Stevens & Burley, Reference Stevens and Burley1997). In addition, although firms can trade some developed technologies in technology markets through licensing arrangements (i.e., licensing out), the amounts involved in such transactions are at best marginal (Gambardella, Giuri, & Luzzi, Reference Gambardella, Giuri and Luzzi2007). Accordingly, when planning and conducing R&D projects, firms are more concerned about the additional resources required for further commercial processes than are public research institutes: the primary concern of public research institutes is not commercial profit but successful knowledge production (e.g., patents and papers).

Because of the different levels of responsibility for further innovation steps such as marketing and manufacturing, the perceived risk inherent in high-risk technology development is much higher for firms than it is for public research institutes; firms are therefore more shortsighted and risk averse about R&D activities than are public research institutes (Miotti & Sachwald, Reference Miotti and Sachwald2003). Further, firms involved in government-supported R&D programs tend to be reluctant to develop high-risk technology, even if it suggests high-potential returns, given the backdrop of bureaucrats’ obsession with R&D outputs and unpredictable policy changes (Seo & Yang, Reference Seo and Yang2011).

Firms that conduct government-supported R&D programs could therefore be relatively reluctant to develop converging technologies. The development of converging technologies requires the profound exploration of opportunities across technological domains in the developmental phase and involves more risk in the commercialization phase than that of ordinary technology (Nordmann, Reference Nordmann2004; Tegart, Reference Tegart2005). Although the use of public R&D funding makes it attractive for firms to conduct more explorative research (such as developing converging technologies) than they would do using their internal resources, firms focus mainly on short-term and low-risk economic benefits, unlike public research institutes. In other words, the vague outlook of the development process and difficulty exploiting technological assets (Tegart, Reference Tegart2005) may lead researchers at firms to reconsider and even avoid the pursuit of converging technology.

By contrast, the risk of the future use of converging technology does not seriously endanger the status quo of researchers at public research institutes, which operate with different incentives, goals, routines, and decision-making structures than for-profit entities (Bercovitz & Feldman, Reference Bercovitz and Feldman2007). That is, exploiting opportunities across technological domains, public research institutes are bound with relatively low risk from the commercial success, compared with the risk private firms confront. Although public research institutes concern about the commercial success of technology because of the licensing income after technology transfer, relatively few complementary activities (e.g., marketing and manufacturing of final products) are required for them after the development of converging technologies, compared with private firms. Furthermore, commercial success of developed technology is not the primary concern of public research institutes – they chiefly focus on knowledge production or education.

Rather, the novelty potentially generated from the development of converging technologies may help such researchers receive more attention from policymakers and the public, which would thus help them acquire further R&D resources from funding agencies (Winter, Reference Winter2004). Therefore, since a number of researchers at private research institutes put laborious efforts into uncovering funding opportunities (Goldfarb, Reference Goldfarb2008; Bozeman, Fay, & Slade, Reference Bozeman, Fay and Slade2013), and financial resources are a key determinant of research output, they could be more motivated than those at private firms to develop converging technologies that would improve their reputations. This leads to the following hypotheses:

Effect of strategic collaboration on technology convergence within the triple helix

Organizations – especially private firms – tend to adopt a collaboration strategy when confronting risk-related issues such as R&D activities (Becker & Dietz, Reference Becker and Dietz2004; Belderbos, Carree, & Lokshin, Reference Belderbos, Carree and Lokshin2004). This phenomenon can be understood from the resource-based view of the firm, in which an organization is considered a portfolio of its own competencies. According to this theory, an organization seeks to collaborate with partners whose resources are mutually complementary (Prahalad & Hamel, Reference Prahalad and Hamel1990). This is because diverse technological competencies are required for firm success, but the accumulation of technological competencies involves risks such as the uncertainty of commercial success and high-potential cost (Zahra, Reference Zahra1996; Teece, Pisano, & Shuen, Reference Teece, Pisano and Shuen1997). Therefore, firms implement a collaboration strategy in order to mitigate risk by sharing it with their partners; such a strategy is in widespread use among R&D networks (Liebeskind, Oliver, Zucker, & Brewer, Reference Liebeskind, Oliver, Zucker and Brewer1996; Tether, Reference Tether2002; Miotti & Sachwald, Reference Miotti and Sachwald2003; Becker & Dietz, Reference Becker and Dietz2004) because it improves the outcomes of technological innovation (Kogut, Reference Kogut1988; Das & Teng, Reference Das and Teng2000; Hagedoorn, Link, & Vonortas, Reference Hagedoorn, Link and Vonortas2000).

By investigating the behavior of firms engaging in R&D collaboration, scholars, especially those interested in organizational learning, have found that firms strategically choose their R&D partners based on suitability for the collaborative purpose (i.e., either exploration or exploitation of innovation). This is because the effective use of collaboration strategy positively contributes to firm performance (Bercovitz & Feldman, Reference Bercovitz and Feldman2007).

Therefore, firms tend to collaborate with public research institutes when undertaking explorative R&D activities such as those relating to radical product innovation (Eom & Lee, Reference Eom and Lee2010) or market penetration (Belderbos, Carree, & Lokshin, Reference Belderbos, Carree and Lokshin2004). In general, when forming cross-boundary research alliances (e.g., firm–firm, firm–university, or firm–government research institute), firms value diversity for exploration research and complementarity for exploitation research (Bercovitz & Feldman, Reference Bercovitz and Feldman2007). Since collaboration with public research institutes, especially universities, provides firms with diverse and novel scientific and technological knowledge as well as low-cost labor (Belderbos, Carree, Diederen, Lokshin, & Veugelers, Reference Belderbos, Carree, Diederen, Lokshin and Veugelers2004), they are likely to conduct explorative R&D rather than exploitative R&D when researching as part of a public-sector alliance (Bercovitz & Feldman, Reference Bercovitz and Feldman2007). Moreover, such collaboration garners opportunities to find new technological paths (Sakakibara, Reference Sakakibara2001) and eventually improves the sourcing firms’ outcomes (Sanchez & Tejedor, Reference Sanchez and Tejedor1995).

Likewise, when opting against undertaking R&D activities alone, firms can be sufficiently motivated to collaborate with public research institutes for the exploration of technology convergence. As converging technologies contribute greatly to the formation of cross-market entry incentives (Katz, Reference Katz1996) and entail high risk for firms in the development process (Tegart, Reference Tegart2005), the reduced risk and strengthened knowledge pool of an R&D alliance attracts firms to collaborate with public research institutes. In addition, the garnered opportunities for finding new technological paths may increase the spontaneous chances of discovering opportunities for technology convergence that require paths bridging disparate technology domains. This leads to the following hypotheses:

It could also be argued that public research institutes have incentives to collaborate with industrial firms because they lack the complementary assets necessary for the commercialization of the developed technology and because licensing income is their primary economic return from the generation of new knowledge; indeed, R&D interaction between the private and public sectors is nowadays unsurprising.

In general, the linear model of innovation conceptualizes universities and government research institutes as being upstream of firms. In particular, universities are usually involved at the earliest stage of knowledge creation, followed by government research institutes (Rothaermel & Deeds, Reference Rothaermel and Deeds2004). Accordingly, for public research institutes, collaboration with firms may cause R&D activities to be more mission oriented (Goldfarb, Reference Goldfarb2008), which reduces the opportunities to explore technology convergence. Further, unlike public research institutes, firms tend to demand that their R&D partners have a discernible outlook and adopt clear operational work processes (Goldfarb, Reference Goldfarb2008) in order to ensure that R&D activities of the partners are commercially lucrative. In other words, in their collaboration with public research institutes, the majority of firms look for short-term technical solutions (Meyer-Krahmer & Schmoch, Reference Meyer-Krahmer and Schmoch1998), which restricts traditional academic goals (i.e., the discovery of new findings; Goldfarb, Reference Goldfarb2008).

Converging technologies are not defined within traditional technological boundaries, and they create value in new fields across technological domains; this necessitates delicate exploration between experts (Llerena & Meyer-Krahmer, Reference Llerena and Meyer-Krahmer2003). However, it is always challenging to exchange and transfer techno-scientific knowledge across organizational boundaries, and this challenge is increased when the recipient is not capable of absorbing knowledge because of limited direct experience with the technology (Cohen & Levinthal, Reference Cohen and Levinthal1990). Moreover, firms mainly have expertise in exploitation, while public research institutes have it in exploration. Accordingly, because of the difficulty exploring converging technologies, universities and government research institutes can access more opportunities to find new approaches to technical issues, potentially leading to a greater likelihood of creating converging technologies, when they are not collaborating with firms. This leads to the following hypotheses:

There have been very few discussions about collaboration between universities and government research institutes because both types of research entity share similar R&D purposes (Bozeman, Reference Bozeman2000). However, each has distinctive organizational features and thus different R&D characteristics. While government research institutes are likely to focus on contributing to their nations and industries by using their well-equipped infrastructure and developing technology to be more applicable, universities tend to focus on the production of academic knowledge by exploring broad techno-scientific domains.

Thus, it makes sense that government research institutes wish to collaborate with universities to develop converging technologies: they have comparable resources. Such institutes require a source of new scientific/technological knowledge to apply to their research in order to adapt advanced technology, incorporate cutting-edge infrastructure, and pursue academic goals (Bozeman, Reference Bozeman2000). For government research institutes, such new scientific/technological knowledge may also provide further novel possibilities for merging other technologies in partnership with universities.

This type of collaboration also provides universities with potentially fruitful opportunities to explore potential applications of their own technologies because, unlike private firms, government research institutes are largely exempt from the financial burden necessarily incurred during the commercialization process. Thus, collaboration between universities and government research institutes nurtures more opportunities for technology convergence than does independent development. This leads to the following hypotheses:

Data sources

This study uses data from the National Science and Technology Information Service in South Korea to demonstrate the different organizational contexts within the triple helix. These data include 9 years (2001–2009 inclusive) of information about the features of government-supported R&D programs and their outputs in South Korea. Researchers who undertake any government-supported R&D project must register its R&D outputs such as patents. The use of this novel database featuring a broad time period and technology domains makes it possible to show the generalizable behavior of R&D entities that develop technologies, especially those entities that use public R&D resources.

The data include 66,244 patents in South Korea that resulted from government-supported R&D projects. I exclude patents that do not include sufficient information for the present study, such as the macro-technology domain and identification code of the associated R&D project; thus, I use 51,837 patents in the final sample. Some of these patents have already been issued, but others have not yet been judged by the Korean Intellectual Property Office. However, it is sensible to recognize even patent applications as the results of R&D activities (Thursby & Kemp, Reference Thursby and Kemp2002); therefore, I include all applications regardless of whether the patents have actually been issued.

Measurements of technology convergence

Because there are so few empirical studies of technology convergence, a review of the means of measuring science convergence is required to judge whether a technology is converging. Although scholars have debated the appropriate indicators to measure science convergence (Morillo, Bordons, & Gómez, Reference Morillo, Bordons and Gómez2003), they generally use bibliometric methods (van Rijinsoever & Hessels, Reference Van Rijinsoever and Hessels2010). A number of studies analyze interconnectivity among disciplines by assessing cross-disciplinary citations in journal articles (Small, Reference Small1999; van Leeuwen & Tijssen, Reference Van Leeuwen and Tijssen2000; Porter, Cohen, Roessner, & Perreault, Reference Porter, Cohen, Roessner and Perreault2007; Porter & Rafols, Reference Porter and Rafols2009), the co-classification of journal subfields (e.g., Tijssen, Reference Tijssen1992; Morillo, Bordons, & Gómez, Reference Morillo, Bordons and Gómez2003), or co-wording in journal articles (Palmer, Reference Palmer1999). Industry convergence studies use similar methods; for example, Curran and colleagues (see Curran & Leker, Reference Curran and Leker2009, Reference Curran and Leker2011; Curran, Bröring, & Leker, Reference Curran, Bröring and Leker2010) define industry convergence as the rate of multi-assignation between industrial fields based on the International Patent Classification.

However, such bibliometric methods cannot be applied to technology convergence studies because academic journals cannot represent technical/commercial knowledge and innovative activity. Instead, patent documents constitute an ample information source that describes such knowledge and helps create an understanding of the linkages among industries, nations, or technologies in terms of technological innovation and knowledge flow (Lee & Kim, Reference Lee and Kim2012).

In this paper, I thus suggest an alternative means of measuring technology convergence based on the multi-assignation of patent documents. Regardless of the different paths that eventually comprise technological emergence, an upwelling of technology accompanies a set of linked knowledge (Llerena & Meyer-Krahmer, Reference Llerena and Meyer-Krahmer2003). The key measurement issue is how to define the link and the original source of technological knowledge, as similarly discussed in science convergence studies. R&D projects that result in patents as the outputs of R&D activities are a useful and practical source of technological knowledge for multi-assignation of patent documents. Under the premise that technological knowledge derives from R&D activities, the techno-scientific domain of R&D projects can be assumed to be the primary techno-scientific domain of R&D activities. In fact, some scholars argue that convergence can also be gauged in terms of proposals or projects (Porter et al., Reference Porter, Cohen, Roessner and Perreault2007). In this paper, the linkage between the techno-scientific domains of R&D projects and patents is used to establish a basis for technology convergence.

Another essential issue is the classification of technology. Although such classifications already exist in various forms, including the International Patent Classification and the National Science and Technology Standard Taxonomic SystemFootnote ², a taxonomic framework should consider the standpoint of governmental policy on technology convergence in order to derive practical implications.

According to the government initiatives of major developed countries, technologies can be categorized into discrete domains. The National Science Foundation in the United States proclaims four major technology convergence domains: nanotechnology (NT), biotechnology (BT), infotechnology (IT), and cognitive science (Roco & Bainbridge, Reference Roco and Bainbridge2002). The European Commission defines a similar typology (NT, BT, IT, social science, and humanities) (Nordmann, Reference Nordmann2004), as does the Japanese government in its third Science and Technology Basic Plan (NT, BT, IT, and environment technology [ET]; Kim, Jung, & Jeong, Reference Kim, Jung and Jeong2009).

In South Korea, the Office of Science and Technology Innovation's Promising New Future Technologies (6T) consist of six major technology domains for convergence: NT, BT, IT, ET, space technology (ST), and culture technology (CT)Footnote ³; all R&D projects in South Korea are classified as per these domains (Oh, Kim, & Ahn, Reference Oh, Kim and Ahn2010)Footnote ⁴. This South Korean typology for technology convergence is widespread and has been used to plan practical technology convergence strategies and in the processes of planning, investing in, and assessing national R&D programs (Kim, Jung, & Jeong, Reference Kim, Jung and Jeong2009). Thus, it is sensible to make this typology the fundamental framework for analyzing technology convergence – at least in the South Korean context.

During planning, South Korean researchers declare the macro-technology domain to which the R&D project belongs and then submit a proposal to funding agencies. If accepted, they undertake these projects. During R&D activities or after completion, they register the produced patents along with information on what these projects contributed to their creation.

Researchers are careful to reference the appropriate R&D projects and contribution ratios when reporting patentsFootnote ⁵. Superior research agencies (e.g., the National Research Foundation of Korea) review the registered research outputs (i.e., patents) with peer evaluators and investigate the relevance of the outputs to the R&D activities to prevent researchers from overselling their research outputs. In addition to this ex post evaluation, the legal issue of intellectual property rights leads researchers to be meticulous in reporting their research outputs to ensure the fair distribution of licensing fees and ownership shares.

Patents can have either multiple assignations or only a single assignation of sourced R&D projects. Among the patents with multiple assignations, some patents derive from sourced R&D projects with homogeneous macro-technology domains, while some other patents derive from sourced R&D projects with heterogeneous macro-technology domains. For the present study, I presume that the latter type of multiple assignations represents technology convergence because it indicates that researchers have referred to and combined technological knowledge from heterogeneous technology domains; hereafter, I define such patents as ‘converging patents.’ The definition of converging patents is depicted in Figure 1. For example, this figure shows that Patent 5 is a converging patent because it originates from two distinct technology domains, whereas others are associated with only a single technology domain.

Figure 1 Conceptualized structure of technology development

Variables

Based on the foregoing, I constructed a dependent variable, Convergence, that equals 1 if the patent is a converging patent and 0 otherwise. I further set dummy variables according to organizational type to demonstrate the effects of R&D entity involvement on technology convergence. Indu_Involved is equal to 1 if the patent is a result of single or multiple R&D projects in which any industrial firm participated and 0 otherwise. Likewise, Univ_Involved is equal to 1 if the patent is a result of R&D project(s) in which any university participated, and Gov_Involved is equal to 1 if the patent is a result of R&D project(s) in which any government research institute participated. Since some R&D projects are collaborations, Indu_Involved, Univ_Involved, and Gov_Involved are not mutually exclusive.

In the next step, I set mutually exclusive dummy variables designated by type of collaboration between R&D entities to determine whether different organizational contexts or merely the forming of collaboration results in the advent of technology convergence. Indu is equal to 1 if an industrial firm developed the technology without any collaboration with other R&D entities and 0 otherwise, whereas Indu–Indu is equal to 1 if the patent is a result of R&D project(s) conducted through R&D collaboration between industrial firms. The variables Univ, Univ–Univ, Gov, and Gov–Gov follow along the same lines. Likewise, the dummy variables of collaboration between heterogeneous R&D entities (i.e., Indu–Univ, Univ–Gov, Indu–Gov, and Indu–Univ–Gov) follow the same rule as applied in Indu–Indu, Univ–Univ, and Gov–Gov.

Finally, I introduced the macro-technology domains for controls. Neither convergence nor technology innovation occurs uniformly across all techno-scientific domains (Ziman, Reference Ziman1994; Malerba, Reference Malerba2002), and not every domain is conducive to convergence (Ziman, Reference Ziman1994). For this reason, Carayol and Thi (Reference Carayol and Thi2005) also introduce controls for each discipline in their empirical analysis. As before, I set dummy variables as controls based on the 6T typology of macro-technology domains (i.e., IT, BT, ST, CT, NT, ET, and ETC). In this respect, I assume that the technology domain with the highest share of contribution to a patent is the technology domain of the patent; hereafter, I call this the ‘key technology domain.’ In cases where there is more than one key technology domain as determined by this method, the technology domain associated with the larger amount of funding is assumed to be the key technology domain. The definitions of these variables are summarized in Table 1.

Table 1 Definitions of the dependent and independent variables

Figure 2 depicts the percentage of converging patents among all patents by year. This figure shows that the ratio grows incrementally as the annual government R&D budget increases. This phenomenon indicates that the level of R&D activity corresponding to the need for technology convergence has increased, perhaps because needs have increased or the R&D budget has grown. Further, the trend took a significant downturn in 2008, probably because of the global economic crisis, although it recovered in 2009: economic conditions seem to influence the trend of technology convergenceFootnote ⁶.

Figure 2 Converging patents as a percentage of all patents by year

Table 2 reports the mean values of the independent and control variables. Because Indu_Involved, Univ_Involved, and Gov_Involved are not mutually exclusive, the sum of the mean values of these variables exceeds 1, whereas the sum of the mean values of Indu, Indu–Indu, Univ, Univ–Univ, Gov, Gov–Gov, Indu–Univ, Univ–Gov, Indu–Gov, and Indu–Univ–Gov equals 1. In addition, the data show that the shares of patents in the data steeply increase with each application year, with the exception of Y2008. This may be caused by a gradually increasing R&D budget, as shown in Figure 2, and improving R&D productivity, which affect the output of R&D activities. However, further study may be required to confirm this reasoning.

Table 2 Descriptive statistics

Methodology

I begin the following empirical analysis by estimating the probability of developing converging technology as a function of the above-described variables and technological controls. Since Convergence is a dichotomous variable, I use probit analysis in this estimation (Greene, Reference Greene2003). In the presented model, R&D entities are assumed to make decisions with the objective of maximizing their own utility. I denote developing technology as c, which equals 1 for execution of converging technology and 0 otherwise; the ith technology is developed to be converging technology only if U_{i 1} > U_{i 0}. The utility function that ranks the preference for the ith technology is then assumed to be a function of the characteristics of the given contexts ‘X’ (e.g., organizational factors), with the disturbance term whose mean value is 0:

$$${{U}_{i{\rm{1}}}}(X)\, = \,{{{\rm{\rbeta }}}_{\rm{1}}}{{X}_i}\, + \,{{{\rm{{\repsilon}}}}_{i{\rm{1}}}}$$$ for execution and $$$ {{U}_{i{\rm{0}}}}(X)\, = \,{{{\rm{\rbeta }}}_{\rm{0}}}{{X}_i}\, + \,{{{\rm{{\repsilon}}}}_{i{\rm{0}}}} $$$ for non-execution.

Using a probit model based on a normal distribution for ε, I examined the estimation with the probit option on STATA 12.

Effects of R&D entities’ involvement on technology convergence

Table 3 presents the empirical results of the effects of R&D entities’ involvement on technology convergence. The standard error of each coefficient is displayed in parentheses below each coefficient. I used additional models that vary in terms of the variable set of contexts and controls to explore the possible individual effects within these estimations. Model 1 includes the variables of the involved R&D entity type (i.e., Indu_Involved, Univ_Involved, and Gov_Involved) as well as the year dummy variables; Model 2 adds the technological controls (i.e., the variables related to 6T).

Table 3 Probit estimations of the effects of R&D entities’ involvement on technology convergence

Note. ^†p < .1, *p < .05, **p < .01.

Overall, the results present strong consistency in terms of the estimated signs and significance levels, suggesting robustness of the presented theoretical model. With the exception of ST and ET, the variables in Models 1 and 2 are significant at the 10% level, while the signs of the coefficients do not differ between these two models. In addition, compared with Model 1, inclusion of technological control variables significantly increases the pseudo R ² in Model 2 and thus shows the importance of technological control, although the value of pseudo R ² is still not high. The same pattern can be observed in the classification table, Table 4. I set the probability level (i.e., cutoff probability) to be the mean value of Convergence for relative accuracy of prediction. Model 1 correctly predicts 59.8% of the cases, whereas Model 2 does 66.2% of the cases. The increase is attributed to the inclusion of technological controls, thus indicating that the nature of convergence may vary by the technological domains, especially prediction on non-converging technology.

Table 4 Classification table of Models 1 and 2

Note. ^aPr(Convergence) ≥ probability level ⇒ Predicted Convergence = 1; otherwise 0.

These estimations in Table 3 support Hypotheses 1a and 1b but reject Hypothesis 1c. Like private firms, government research institutes negatively affect the development of converging technology when involved in R&D activities, whereas universities have a positive impact. Before these estimation results, one could infer that government research institutes, which generally involve large research groups and conduct large-scale research projects, invent more converging technologies than do other types of R&D organization. Large R&D groups could be advantageous for covering the costs of convergence (e.g., the cost of merging heterogeneous entities caused by the cognitive distance between distinct techno-scientific domains) because they have more financial resources, social capital, or accumulated knowledge stock than do other types of R&D organization. However, the results suggest that the negative influence of industrial firms is weaker than that of government research institutes because the coefficients of Gov_Involved are higher than those of Indu_Involved. Government research institutes and universities share many characteristics by virtue of being public institutes, but the results show they are significantly different in terms of their tendencies to develop converging technologies.

The same tendency found using the Probit estimation can be found through the values of the odds ratio, as shown in Table 5. The contingency table between Convergence and each organization type variable (i.e., Indu_Involved, Univ_Involved, and Gov_Involved, respectively) presents the frequency of convergent technology development by combination of dichotomous variables. The share of the frequency based on the value of Convergence is displayed in parentheses below each frequency. The odds and odds ratio using the values in the contingency table are calculated and presented below the contingency table. While the odds ratios of Indu_Involved and Gov_Involved are both ~0.6, that of Univ_Involved is 3.29; that is, a technology developed with the participation of a university is 3.29 times more likely to be a convergent technology than a technology developed without the participation of a universityFootnote ⁷. However, note that the absolute probability of inventing a converging technology at a university (i.e., the odds of Univ_Involved, 0.13) is quite low.

Table 5 Contingency table and odds ratio by type of R&D entity

In summarizing the aforementioned results, the influence of R&D entity type on technology convergence can be depicted as shown in Figure 3.

Figure 3 The effects of research and development entities’ involvement on technology convergence

The estimations comprising the 6T variables illustrate the intriguing nature of technology convergence. The results derived from Model 2 show that IT and BT (CT and NT) are less (more) likely to be involved in technology convergence than the base group of dummy variables (i.e., ETC). In other words, CT and NT are more likely to combine with other technology domains than are IT and BT. Because NT is often considered a ‘building technology’ that suggests both a new way of manufacturing and novel functions when converging with other promising technologies (Roco & Bainbridge, Reference Roco and Bainbridge2002), its generality could allow it to combine with other technologies more readily.

Moreover, CT may relate positively to technology convergence because of its close technological distance from IT. Therefore, CT–IT convergence might account for a large share of the CT domain but not of the IT domain, probably there are more IT-related patents than CT-related patents (see Table 2). The influence of CT–IT convergence on the probability of IT-related convergence might thus be marginal because of this size difference.

Effects of R&D entities’ collaboration on technology convergence

Table 6 shows the empirical results of the effects of R&D entities’ collaboration on technology convergence. The coefficients and standard errors of each variable are presented in the same way as in Table 3. Models 3, 4, and 5 include observations of R&D projects conducted by industrial firms, universities, and government research institutes, respectively, in order to demonstrate the collaboration effect by R&D entity type. For example, the observations used in Model 3 are of patents created by R&D projects in which industrial firms participate. Further, because Indu, Univ, and Gov work as base groups, they are omitted from Models 3, 4, and 5, respectively, while redundant variables are also excluded. For example, Univ, Univ–Univ, Gov, Gov–Gov, and Univ–Gov are not necessaryin Model 3 to demonstrate the behavior of industrial firms.

Table 6 Probit estimations of the effects of R&D entities’ collaboration on technology convergence

Note. ^†p < .1, *p < .05, **p < .01.

The presented results show that organizational differences, rather than the collaboration type itself, influence the probability of creating technology convergence, because none of the coefficients of the variables indicating collaboration between homogeneous R&D entities (i.e., Indu–Indu, Univ–Univ, and Gov–Gov) is significant at the 10% level. This finding implies that organizational context is crucial for understanding the nature of technology convergence.

Specifically, the results derived from Model 3 show that industrial firms are more likely to create converging technology in collaboration with universities than when they conduct R&D activities alone. The coefficients of Indu–Univ and Indu–Univ–Gov are positively significant at the 1% level; thus, Hypothesis 2a is supported. Meanwhile, the coefficient of Indu–Gov is not significant, although its sign is the same as Hypothesis 2b; thus, Hypothesis 2b is not supported.

As expected, the coefficients of Indu–Gov and Indu–Univ–Gov are negatively significant in Model 4 at the 1% level; thus, Hypothesis 3a is supported. These results illustrate that universities are less likely to create converging technology in collaboration with industrial firms than when they conduct R&D activities alone. However, the coefficient of Univ–Gov is not significant, with the expected sign as in Hypothesis 4a; thus, Hypothesis 4a is not supported.

The results derived from Model 5 suggest that government research institutes are more likely to create converging technology in collaboration with universities but less likely to do so in collaboration with industrial firms than when they conduct R&D activities alone. The coefficient of Indu–Gov is significant at the 1% level with a negative sign, whereas that of Univ–Gov is significant with a positive sign. Therefore, Hypotheses 3b and 4b are both supported. However, Indu–Univ–Gov is not significant at the 10% level, probably because the negative effect of industrial firms and the positive effect of universities co-exist in the variable. To summarize the aforementioned results, the influence of type of R&D collaboration on technology convergence can be depicted as shown in Figure 4.

Figure 4 The effects of research and development entities’ collaboration on technology convergence

The results of technological controls (6T) derived from Models 3, 4, and 5 are largely similar to those presented for Models 1 and 2. As before, IT and BT seem to be less likely to be involved in technology convergence than CT and NT (see Model 2). However, the coefficient of NT is not significant in Model 5. Instead, the coefficients of ST and ET, which are not significant in any other model, are negatively significant. In terms of ST, this finding may occur because government research institutes play a significant role in national space programs, which are highly mission oriented. Therefore, ST is less likely to be involved in technology convergence than CT, which readily evolves with other technologies in commercial markets. Likewise, because a number of government research institutes work on government-owned power plants, the technology convergence of ET tends to follow that of ST.

In addition, consistent with Models 1 and 2, the models each correctly predict over 70% of convergent technology cases and over 60% of overall cases, as shown in Table 7. On the condition of the productivity level set in the same way for Models 1 and 2, the percentages of correct cases suggest the models are valid at estimating the probability of developing converging technology and require improvement in prediction power as well.

Table 7 Classification table of Models 3, 4, and 5

Note. ^aPr(Convergence) ≥ Probability level ⇒ Predicted Convergence = 1; otherwise 0.

Discussion and policy implications

This study examined how the organizational contexts of R&D entities affect the advent of technology convergence. By adopting the multi-assignation analysis of technology domains, this paper used a rich and novel data set based on the technologies derived from government-supported R&D programs in South Korea from 2001 to 2009 to demonstrate the effects of different types of R&D entities and alliances.

Overall, the presented findings illustrate the important role of universities and the negative impact of private firms on technology convergence. Although both firms and government research institutes tend to be reluctant to or even incapable of independently developing converging technologies, collaboration with universities increases the likelihood of such development. The empirical results further show that firms’ collaboration strategies for radical innovation and exploration can extend to their technology convergence strategies, supporting the important role of the organizational context in leading technology convergence. This extended view suggests that policymakers need to design support systems for post-R&D processes (e.g., marketing and seeking consumers) to alleviate the risk associated with the development of technology convergence, especially among industrial firms.

In addition, proper incentive systems that relate to the challenges inherent in technology convergence may offer fruitful opportunities for development. The presented findings showed that industry–university R&D alliances negatively affect technology convergence on the university side, but positively affect technology convergence in private firms. As reasoned in this study, differences in incentive systems and attitude toward exploration between the public and private sectors complicate the production of converging technologies. Therefore, appropriate reward structures for the exploration of converging technologies (e.g., specialized large-scale joint R&D programs for the creation of technology convergence) might lead university researchers and industrial firms to collaborate in the future. In particular, since the commercialization of converging technologies is relatively difficult (Tegart, Reference Tegart2005), flourishing industry–university relationships based on such a reward structure might expedite the diffusion of technology convergence, thereby potentially contributing to the generation of novel economic benefits for universities, industrial firms, and even the nation.

An interesting question is why government research institutes do not play a university-like role in technology convergence, especially given the homogeneity between the two entities (Tether, Reference Tether2002). The distinct pattern of South Korean government research institutes in technology convergence may be attributed to their local characteristics, such as firms’ weak preference for government research institutes in high-risk innovation projects, the strongly mission-oriented R&D goals of government research institutes, or their inflexible, goal-oriented R&D perspective.

According to Noh, Chung, and Rah (Reference Noh, Chung and Rah2010), South Korean firms are less likely to choose government research institutes than other firms or universities as partners in technology innovation programs because they do not trust the competency levels of government collaborators. Consistent with the findings presented by Noh, Chung, and Rah (Reference Noh, Chung and Rah2010), Min (2010) suggests that researchers from South Korean government research institutes perceive that their lack of capability and creativity in adopting new technologies derives from their hierarchical organizational systems and inflexible research assessments. Given the extended understanding of technology-innovation strategy, this suggests a low degree of capability that could influence the selection of collaboration partners.

The case of government research institutes informs policymakers that stable funding, strong organizational competency, and appropriate combinations of research collaborators can promote technology convergence. In countries in which government research institutes have a strong academic reputation, a high degree of autonomy and self-sufficiency in terms of funding, and well-established relationships with industrial firms (e.g., Fraunhofer and Max Planck in Germany, VTT in Finland, and TNO in the Netherlands), they might play a role similar to that of universities in this study. That is, in those countries, universities and industrial firms would be likely to collaborate with government research institutes in developing converging technology. However, in countries in which government research institutes largely depend on given government funding and universities have much stronger competencies than government research institutes (e.g., national laboratories in the United States), government research institutes might have a role similar to the one shown in this study. Support for these measures may require additional review and demonstration.

Limitations and further research

This study has certain limitations. First, the methodology used to measure technology convergence does not distinguish converging technology that results from strategically planned R&D activities for technology convergence from that resulting from spontaneous discovery. The advent of technology does not necessarily occur in line with strategic direction, and some R&D entities could consider the possibility of technology convergence and then adopt an alternative collaboration strategy.

Second, in line with previous studies (especially those on science convergence, like Morillo, Bordons, & Gómez, Reference Morillo, Bordons and Gómez2003, that use co-authorship and multi-assignation analysis), the methodology and data herein do not include every type of existent technology convergence. Given that technological knowledge is known to evolve through a confluence of existing, relevant technologies (Arthur, Reference Arthur2009), the findings of this study should thus be interpreted with the understanding that the revealed empirical results reflect only the patenting activities of R&D organizations. Although patenting activity chiefly represents the technological innovations of R&D organizations, the linkage between patents and projects merely shows the consequences of R&D activities, not the processes. In addition, though rare, some R&D projects are planned to be interdisciplinary; thus, some convergent technology patents are not counted as converging technology because they involve no interaction with previous research achievements. As a matter of course, this limitation of measurement also occurs in any other measurements of convergence/interdisciplinarity based on linkage (e.g., co-citation analysis, citation flow analysis, and co-classification analysis). Thus, this study neither illustrates the entire techno-scientific picture nor gives conclusive answers to the research question of this study but only delivers essential insights on the advent of technology convergence.

Finally, despite the rich national-level data set used in this study, it lacks the explanation of diverse contexts outside the environment of government-supported R&D, which would be necessary to understand the overarching nature of technology convergence. In particular, the firm's motivation to participate in government-supported R&D programs could affect organizational behavior. Furthermore, other contexts such as resources, social capital, and technological characteristics could affect the tendency for technology convergence, although this paper mainly focuses on the organizational context. A model that includes such contexts as well as the organizational context might have better predictive power, thus possibly providing a more profound understanding of the nature of convergence.

This study also suggests certain directions for future research. For one, the actual economic benefits of technology convergence could be studied. For instance, citation analysis or assessments of patent quality could widen the current understanding of technology convergence and the strategies of R&D entities. For another, cross-cultural analysis might help explain organizational contexts for technology convergence. Macro-organizational behavior can differ by culture and established institution; therefore, analyzing the effects of such differences might help policymakers build better environments for technology convergence. Moreover, a study with a more generic data set involving different regions and approaches to technology convergence should follow in the future to strengthen the generalizability of the presented results.