Collected worker experiences, knowledge management practices and service innovation in urban Norway
2021; Elsevier BV; Volume: 100; Issue: 6 Linguagem: Inglês
10.1111/pirs.12633
ISSN1435-5957
AutoresSverre J. Herstad, Marte C.W. Solheim, Marit Engen,
Tópico(s)Management and Organizational Studies
ResumoKnowledge-intensive services firms prefer to locate in cities that provide access to rich information flows and abundant opportunities for learning-by-recruiting. Focusing specifically on such locations, this paper explores how innovation is associated with work experiences "collected" by employees through their recent career paths and the implementation by current employer firms of practices to manage knowledge. Strong complementarities are found using a unique Norwegian dataset: The statistical association between practices and innovation outcomes depends strongly on variety of experience-knowledge among employees. Conversely, while said variety does not affect innovation in the absence of dedicated practices, it strongly does in their presence. Las empresas de servicios intensivos en cuanto a conocimiento prefieren ubicarse en ciudades que ofrecen acceso a ricos flujos de información y abundantes oportunidades de aprendizaje mediante la contratación. Centrándose específicamente en estos lugares, este artículo estudia cómo está asociada la innovación a las experiencias laborales "recopiladas" por los empleados durante sus recientes trayectorias profesionales y la aplicación por parte de las empresas empleadoras actuales de prácticas para gestionar el conocimiento. Se encontraron fuertes complementariedades mediante el uso de un conjunto de datos único de Noruega: La asociación estadística entre las prácticas y los resultados de la innovación depende en gran medida de la variedad de la combinación de experiencia-conocimiento de los empleados. A la inversa, mientras que dicha variedad no afecta a la innovación en ausencia de prácticas especializadas, sí lo hace fuertemente si están presentes. 知識集約型サービス産業の企業は、豊富な情報フローと豊富なlearning-by-recruiting(実際に雇用することによって学習する)の機会が得られる都市部に立地することを好む。本稿では、そのような場所に焦点を当てて、従業員の最近のキャリアパスを通じて「集められた」職務経験とイノベーションはどのように関連しているか、また、現在の雇用主である企業の知識マネジメントの実践方法について調査した。ノルウェーの独自のデータセットから強い相互補完の関係が認められる。すなわち、実践とイノベーションの成果の統計的関連性は、従業員の経験知識の多様性に強く依存している。逆に、献身的な実践がなければ従業員の経験知識の多様性がイノベーションに影響を与えることはないが、献身的な実践があれば強く影響を与える。 Corporate innovation has traditionally been studied from the perspective of scientific knowledge, systematic research and development (R&D) work and formal business network configurations (e.g., Bogliacino & Cardona, 2014; Schøtt & Jensen, 2016; van Beers & Zand, 2014). Yet, structural change has for long favoured services industries where even firms in the knowledge-intensive segment (hereby KIS) might depend less on R&D than on large amounts of human resources (Niosi et al., 2012; Pina & Tether, 2016; Teece, 2003). The strong and persistent preferences exhibited by KIS for locating in the "madding crowd" (Shearmur, 2015) of large city regions (Jøranli & Herstad, 2017; Keeble & Nachum, 2002; Torres & Godinho, 2020) suggest that firms in these locations develop innovation models that are geared towards tapping external knowledge pools through recruitment and localized information search (Glaeser, 1999; Malmberg & Power, 2005; Power & Lundmark, 2003). Drawing inspiration from management studies (e.g., Dokko et al., 2009; Mawdsley & Somaya, 2015) and evolutionary economic geography (Boschma & Frenken, 2011), efforts have in recent years been made to explore how organizational performance reflects experience-knowledge "collected" (Östbring et al., 2018) by employees at prior places of employment (Boschma et al., 2009; Herstad et al., 2019; Solheim et al., 2020). In parallel, efforts have also been made to broaden the traditional focus of innovation research on R&D to consider impacts of other organizational practices implemented to stimulate creativity, foster learning and facilitate knowledge integration (Arundel et al., 2007; Donate & Sánchez de Pablo, 2015; Lorenz et al., 2016; Revilla & Rodríguez-Prado, 2018). However, the fundamental question of whether experience-knowledge interact with organizational practices to shape innovation in firms remains open. This question is particularly pressing in the context of KIS due to revealed locational preferences and intrinsic industry characteristics. To address it, we use data from the Norwegian Community Innovation Survey in 2010 (CIS2010) that provide information on different knowledge management practices, and on innovation outcomes. Supplementary employer-employee register data allow us to describe the work experiences that firms" current employees have collected in the past. Our model that predicts innovation probabilities from practically zero to almost 70% finds strong complementarities between variety of collected worker experiences and practices implemented by firms. Innovation can generally be described as a process of accessing or developing cognitive resources, and combining them into new products or practices (Schumpeter, 1934; Witell et al., 2017). In contrast to manufactured goods that are physical manifestations of resources used in development and production, services are intangible and depend on how 'service encounters' (Sørensen et al., 2013; Voorhees et al., 2017) are structured and what interacting agents bring into them. When encounters take the form of face-to-face interactions between clients and service professionals, the challenge arises for firms of how to link the practices of front-line employees to resources and learning in the larger organization (Dougherty, 2004; Engen & Magnusson, 2018). Encounters can also be facilitated by digital technologies, allowing provision without face-to-face interaction. This tightens interdependencies between "service" and the process by which it is provided. As a result, services firms tend to innovate differently from manufacturing (Sundbo & Gallouj, 2000; Tether, 2005; Tuominen & Toivonen, 2011) in that their efforts are more incremental, interactive and distributed (Sørensen et al., 2013; Sundbo & Gallouj, 2000; Tether & Tajar, 2008) with strong complementary relationships (Amara et al., 2009) and often blurred boundaries between "product," "process" and the larger "organization" (Gallouj & Savona, 2008; Witell et al., 2016). The need to innovate along multiple, interrelated dimensions of the business means that creation of workspaces where employees can develop and share their knowledge is crucial. This can be facilitated by different "knowledge management practices instead of # defined by Donate and Sánchez de Pablo (2015, p. 362) as "activities, initiatives and strategies that companies can use to generate, store, transfer and apply knowledge for the improvement of organisational performance." Such practices include systematic research and development work (R&D), but extend into what Popper and Lipshitz (1998) referred to as integrated, dual-purpose learning mechanisms, that is, those that at the outset are interwoven with task performance (Arundel et al., 2007; Jensen et al., 2007) but facilitate also reflection and sharing outside of ongoing work practices (Dougherty, 2004; Engen & Magnusson, 2015). Three such practices have received considerable attention over the years (Arundel et al., 2007; Leiponen, 2006; Lindbeck & Snower, 2000; Nonaka, 1995), recently under the heading of "creativity stimulating mechanisms" (Doran & Ryan, 2017; Revilla & Rodríguez-Prado, 2018): (i) Rotation of jobs (tasks) between employees; (ii) rotation of employees between teams; and (iii) brainstorming sessions (Arundel et al., 2007; Chuang et al., 2016; Lindbeck & Snower, 2000; Santos et al., 2017). Job rotation can reveal to managers and other employees specialized skills or talents that would otherwise be hidden (Ortega, 2001; Santos et al., 2017). Moreover, it can nurture creativity, lubricate interactive learning, and foster the collective understanding of interlinked business processes that makes knowledge more fluid and easier to put into practice (Eriksson & Ortega, 2006; Nonaka, 1994; Revilla & Rodríguez-Prado, 2018). Similarly, teamwork defined as collaborative activities that locate, share, create and apply knowledge among groups of people (Chuang et al., 2016), often from different functional areas (Keller, 2001), is emphasized as highly important—(e.g., Batt-Rawden et al., 2019; Erhardt, 2011; Leiponen, 2006) because teams with different compositions (e.g., project groups) can be established in response to shifting needs. Teamwork enhances the range of information that is available and ease coordination and overlap of tasks (Eisenhardt & Martin, 2000: 1109). Finally, brainstorming sessions can be arranged to access on a broader basis the organisations' distributed knowledge, stimulate creativity (Doran & Ryan, 2017; Revilla & Rodríguez-Prado, 2018) and compile insights that employees have gained through their practices (Donate & Sánchez de Pablo, 2015; Engen & Magnusson, 2018). Based on this, a first hypothesis can be formulated: H1.The implementation of creativity-stimulating mechanisms (CSMs) is positively associated with innovation in knowledge-intensive services firms. The knowledge management framework of Donate and Sánchez de Pablo (2015) follow in the tradition of innovation research more generally by paying particular attention to different aspects of the practice that is R&D; defined by the OECD (2015, p. 28) as "systematic work undertaken in order to increase the stock of knowledge [….] and to devise new applications of available knowledge." Due to this attention, the importance of corporate R&D for in-house knowledge development and the capacity to absorb from the external environment (Cohen & Levinthal, 1989; Cohen & Levinthal, 1990) is well documented (e.g., Cassiman & Veugelers, 2006; Grimpe & Kaiser, 2010). In the discussion by Popper and Lipshitz (1998), R&D is considered a dedicated, non-integrated learning practice, namely, one that operates in (relative) isolation from ongoing business activities for the sole purpose of learning and innovation. The service innovation literature reflects this view in that R&D generally is considered less relevant for innovation due to lower direct technological content and dependence instead on knowledge that is practice-based and distributed among employees (Gallouj, 2002; Karlsson & Skålén, 2015; Sørensen et al., 2013; Tuominen & Toivonen, 2011). While accepting that R&D might generally be less common in services industries, documented intra-industry heterogeneity (e.g., Pina & Tether, 2016) forces the expectation that such work might well be conducted and then impact on innovation: H2.R&D is positively associated with innovation in knowledge-intensive services firms. In extension of this, we need to consider interaction effects between the two knowledge management practices that are CSMs and R&D. A complementary relationship is implied when Grant (1996), Jensen et al. (2007) and others on a general basis emphasizes that success in R&D work depends on access to a broader range of knowledge and ideas than can be contained within dedicated R&D departments. In this perspective, the use of CSMs can serve to mobilize practice-based knowledge from the broader organization, and thereby increase the productivity of R&D. At the same time, it can also be argued that the absence of R&D efforts increases the importance of non-R&D efforts, that is, CSMs; and/or expected that effects of R&D on innovation override effects of CSMs when firms engage in both practices. This suggests a substitutive relationship. More generally, management scholars have found "too much of a good thing" effects to be widespread in organizational design and strategy (Pierce & Aguinis, 2013). Such effects could here take the form of diminishing returns to individual practices due to growing overall complexity of organizational processes when they are combined. Accordingly, we formulate a third hypothesis that leaves the sign of the expected interaction effect open for the empirical analysis to explore: H3.The association between R&D and innovation in knowledge-intensive services firms depends on whether or not CSMs are implemented. While the service research field discusses how to foster and then capture for innovation learning from ongoing task execution in the organization (Engen & Magnusson, 2018; Sundbo & Gallouj, 2000), the geography literature points out that employees enter into firms with knowledge (Timmermans & Boschma, 2014), networks (Agrawal et al., 2006) and behavioural attributes (Dokko et al., 2009; Herstad et al., 2015) that reflect their learning at prior places of employment. The relevance of viewing this as "resources" available for integration into "service" (Witell et al., 2017) is underscored by contributions finding recruitment to be the most important mechanism for competence upgrading in KIS (Jøranli, 2018; Keeble & Nachum, 2002; Tether, 2003) and the networks of employees to be extensively used for information search (Deprey et al., 2011; Reihlen & Apel, 2007; Tödtling et al., 2006). This more than hints that knowledge intensive services prefer to locate in cities for access to "the large amounts of human resources" on which they depend. Urban labour markets facilitate ongoing employer-employee matching (Duranton & Puga, 2004; Eriksson & Rodríguez-Pose, 2017; Helsley & Strange, 1990), and provide individuals with opportunities for "learning-by-doing" that "speed the accumulation of human resources" (Glaeser & Maré, 2001) as evident from persistent wage premiums on careers in cities (Glaeser, 1999; Gordon et al., 2015). Thus, a focus on what individuals carry from their prior work experience (cf. Dokko et al., 2009) to current employer KIS is required, and fully in line with the literature on knowledge spillovers through labour market mobility (Almeida & Kogut, 1999; Møen, 2005; Oettl & Agrawal, 2008; Singh & Agrawal, 2011). To approach this, we draw on the concept of "collected worker experiences" introduced by Östbring et al. (2018). In this perspective, whether experiences can be integrated into (new) services depends how well they interact and combine in firms. At a general level, "cognitive resource diversity theory" in human resource research (cf. Horwitz, 2005) proposes that firms benefit from diverse experiences because they represent a range of knowledge and ideas, networks and perspectives that trigger learning and facilitate innovation through "new combinations" (Van Engen & Van Woerkom, 2010, p. 133). Against this, work within the "similarity attraction paradigm" (cf. Horwitz, 2005; Horwitz & Horwitz, 2007) argues that more effective communication between less divergent perspectives enhance the execution of tasks and provide the basis for continuous incremental improvements of products and practices. This, however, comes with the risk of "myopia" prohibitive of more fundamental changes (Levinthal & March, 1993). Very diverse attributes, on the other hand, could lead to distrust, a lack of shared understanding and a high risk of conflicts that may also reinforce the firm's focus on retaining rather than adjusting established products and practices (Horwitz, 2005; Jehn et al., 1999; Madsen et al., 2003). The economic geography literature (cf. Östbring et al., 2018) echoes this debate when discussing benefits from (Marshallian) specialization versus (urban) diversity (Beaudry & Schiffauerova, 2009; Firgo & Mayerhofer, 2018; Frenken et al., 2007). Translated to the KIS sector under consideration here, this debate is one of whether firms in cities benefit from specialized labour pools that reflect concentration of services employment, or the different recruitment channels that urban diversity open to other domains of the economy (cf. e.g. Jøranli & Herstad, 2017). Focusing on the firm level, Timmermans and Boschma (2014) agree with the earlier Boschma et al. (2009) that positive productivity effects from inflows generally demand that labour dispatching domain are sufficiently different so that a learning potential exists yet not too distant so that the potential can captured; that is, "related." However, they also find that firms in urban locations, and in extension services, respond differently. Similarly, while, Herstad et al. (2015) inflows form "related" industries positively associated with innovation in a sample of manufacturing and services firms, the later Herstad (2018) found the services segment of the same sample dependent on more diverse cognitions than the manufacturing counterpart (cf. also Firgo & Mayerhofer, 2018). In line with this, Herstad et al. (2019) found KIS benefitting for product innovation from combining a broad range of experience-knowledge drawn specifically from the labour markets of large cities. This is consistent with the notion that innovation in the rapidly changing market environments of services demand bridging of different cognitive domains (Bugge, 2011; Jøranli, 2018; Pershina et al., 2019). Still, the discussion of what amount and type of variety is required extends into questions of definition and operationalization (Whittle & Kogler, 2020). Thus, we here simply assume that variety of experiences matter for innovation and leave open for the empirical analysis to address what is the requisite type. A fourth hypothesis is formulated on this basis: H4.Variety of collected worker experiences is positively associated with innovation knowledge-intensive services firms. Even though the literature on absorptive capacity suggests that the type and amount of cognitive variety that firms effectively can assimilate and exploit depend on efforts made in this respect (Cohen & Levinthal, 1990; Zahra & George, 2002), research on collected worker experiences (Herstad et al., 2019; Östbring et al., 2018) and the antecedent literature on aggregate movements in the labour market (Boschma et al., 2009; Herstad et al., 2015; Timmermans & Boschma, 2014) has insofar only occasionally (cf. Solheim et al., 2020: on R&D in the manufacturing industry) considered moderating effects of knowledge management practices (here understood as R&D and/or CSMs). Similarly, research on such practices has paid limited attention to the actual composition of organizational knowledge bases that they are implemented to manage. In the context here, a complementary relationship exists if the implementation of practices allow firms to exploit a broader range of experience-knowledge for innovation, whereas a substitutive relationship exists if collected experiences foremost matter for innovation in the absence of dedicated learning and knowledge integration efforts on the side of firms. Accordingly, a final hypothesis is formulated: H5.The relationship between collected worker experiences and innovation in knowledge-intensive services firms depend on the implementation of knowledge management practices. The analysis uses innovation data sampled by the governmental agency Statistics Norway in the seventh round of the pan-European Community Innovation Survey (CIS2010) that build on definitions and guidelines provided in the third edition of the Oslo Manual (OECD, 2005). In contrast to many other European countries, participation in the Norwegian surveys is compulsory for sampled firms. The 2010 survey is unique in that information on the use of "creativity stimulating mechanisms" during the reference period 2008–2010 is provided in addition to the standard information on R&D, innovation activities and outcomes. For the purpose here, the data have been merged with Linked Employer-Employee Registers (LEED) covering the years 2004–2008. KIS are defined as described in Table A1 in the Appendix. To allow growth, labour replacement and collected worker experiences to be captured as detailed below, only firms established before 2006 are included in the analysis. The sample is also restricted to the firms with 10 employees or more that responded to the full CIS questionnaire, thus providing all required information. These criteria give 1,124 KIS observations, of which 795 were located in one of the four urban housing and labour-market regions of Norway as originally defined by Jukvam (2002) and updated in Gundersen and Jukvam (2013). Given that non-conventional location choices suggest different business strategies and innovation models (e.g., Herstad et al., 2019; Shearmur, 2015), KIS located outside these urban labour market regions are excluded from the analysis. Being typically described as fluid and dynamic, with blurred boundaries between product, process and organization (Gallouj & Savona, 2008; Witell et al., 2016), two dependent variables are used in the analysis to capture innovation "in services." On the one hand, it is strictly defined as "service innovation" with the variable SERVINNO taking on the value 1 if firms reported that they during the reference period introduced a new or significantly improved service onto their markets. On the other, a broader definition is applied as the variable SUPPINNO for "supportive innovation" takes on the value 1 if firms reported process, organizsational or marketing innovation as specified in the CIS.1 As would be expected given these definitions and their theoretical backdrop, Table 1 shows that SUPPINNO is the most common form while SERVINNO without SUPPINNO is particularly rare. As previously noted, we focus on the two types of knowledge management practices that are R&D traditionally defined and those more recently labelled "creativity stimulating mechanisms" (CSMs). The binary variable "R&D" takes on the value 1 if firms reported intramural research and development work during the reference period. The binary variable CSM takes on the value 1 if firms reported using at least two of the three following practices: (i) brainstorming sessions; (ii) multidisciplinary or cross-functional work teams; or (iii) job rotation of staff to different departments or other parts of their enterprise group. The strict definition is applied to capture only firms that engage systematically, and broadly, in such efforts. As can be seen from Table 2 below, a somewhat larger proportion of firms implement CSMs compared to R&D. However, the most striking pattern is that more firms engage in both practices than in either one. To describe the composition of experience-knowledge present in firms at the start of the CIS2010 reference period in 2008, matrixes have been generated that uses industry codes to categorize the past workplaces of employees as observed over a period. Constructing such matrixes demands consistent sector classifications. Standards have changed, and entirely new classes have been added in response to structural change in the economy. For the period 2004–2008, the data allow the previous SN2002 (building on NACE Rev. 1.1) to be harmonized with the current and more detailed SN2007 (building on NACE Rev. 2). Thus, the analysis focuses on experiences collected during this five-year period.2 As an example, consider the median firm with 36 employees in 2008. As observed here, the firm contains 36 (size in 2008) × 5 (years employees are observed) = 180 experience-years. Each experience-year has been assigned to the five-digit industry codes in which it was obtained. Years during the period 2004–2007 in which employees of the focal firm in 2008 were not in employment are not counted because no work experiences were gained. Based on these matrixes, the essential collective dimensions that is how employees' experiences relate to one another when combined in the focal employer firms has been described using entropy measures computed in accordance with Jaquemin and Berry (1979). The variable VAR_BETWEEN is the distribution of experience-years between two-digit main industry groups. The variable VAR_WITHIN is the weighted sum of distributions at the 3-digit level within 2-digit main groups where the weights are the proportions of all experience-years accounted for by each 2-digit group. Total variety (VAR_TOTAL) is the sum of _BETWEEN and _WITHIN, which equals the entropy of the distribution between 3-digit groups. This operationalization is as originally applied by Frenken et al. (2007) to describe the composition of employment in regions and later adapted to capture the composition of experience- knowledge in firms (Östbring et al., 2018; Solheim et al., 2020); however, we refrain from using the terminology "related" versus "unrelated" skills to avoid making substantive assumptions from a method for empirically distinguishing the composition only of industry experiences without taking into account occupations or educations. All workers present in the firm in 2008 are considered, independently of their roles, whether they worked full-time or part-time in 2008 or previous years, and their education levels. Restricting the analysis to certain roles or occupations would be in conflict with the proposition in service innovation theory that learning and innovation have distinct collective, distributed and informal dimensions. Excluding part-time employees in 2008 would suggest also excluding individuals who worked part-time at certain points in earlier years, potentially introducing biases due to over-representation of demographic groups less inclined to work part-time (e.g., middle-aged male workers). Finally, restricting the analysis to include only employees with education above a set threshold level would entail making decisions about what that level generally should be that are non-trivial given sector heterogeneity and the real option open for individuals of following the vocational training rather than higher education qualification route. Examples of relevant vocational training programs include, for example, web design, media and communication that rank low on the education level scale yet qualify for skill-intensive apprenticeships and subsequent knowledge-intensive work. Moreover, skills acquired through experience might well compensate for lack of formal training, academic or vocational. Stability of staff inherently reduces the experience-variety hypothesized to influence innovation positively, while high turnover, for example, of low-skilled workers that are mobile between firms and industries might increase variety with limited influence on innovation. To control for this, the variable CHURN0608 is included that is the proportion of employees present in 2006 that was replaced with new employees during the two-year period leading to the start of the CIS reference period in 2008. Similarly, experience variety can be affected by the growth rate of the firm e.g. through absorption of unskilled labour during fast expansions, and growth can be dynamically interlinked with itself (positive or negative serial correlations cf. e.g. Coad (2007) as well as with innovation (Bogliacino et al., 2017; Herstad & Sandven, 2020). Therefore, the variable GROWTH0608 captures employment growth during the two-year period leading up to the start of the CIS2010 reference period in 2008. Aspects of urban location that we need to control for to isolate the interaction of practices with collected experiences include market orientation (Isaksen, 2004; Keeble & Nachum, 2002), proximity to (a broad range of) potential collaboration partners (Herstad & Ebersberger, 2015), exposure to information flows more generally (Doloreux & Shearmur, 2012; Shearmur, 2015; Storper & Venables, 2004), and overall higher education levels among employees in cities compared to outside (e.g., Aslesen et al., 2008). The variable FORMAR is included that takes on the value 1 if the firms" most important market is outside Norway. Collaboration is strictly defined in the CIS to capture only active participation for innovation with other enterprises or non-commercial institutions. As firms can access external resources also from other sources than partners, and by other means than active participation, we implement instead the control variable SEARCH that captures the use by firms of information from partners as well as non-partner organizations (e.g., competitors) and public sources such as publications and databases. Following Laursen and Salter (2006), the variable is a count of how many of the 11 sources specified in the CIS that are used. The variable EDULEVEL08 is the average education level of firms' employees in 2008, described on the standard eight-level scale used in the registers. Variety measured as entropy is influenced by the size of the firm, which may also affect innovation propensities. This is captured by the variable SIZE that is the natural logarithm of employment in 2008. Knowledge intensive services is a heterogeneous industry, meaning that labour market linkages and innovation propensities might differ substantially between subgroups. Therefore, 14 dummy variables are included in all regressions as controls for 15 two-digit industry groups described in Table A1 in the Appendix. Dummy variables are also included to capture previously documented inter- and intra-regional differentiation (cf. Herstad & Ebersberger, 2015) in the Norwegian urban hierarchy: OSLO C captures locations within the capital city itself. OSLO W captures locations in the bordering south-western municipalities where ICT and other technical services remain heavily concentrated (Isaksen, 2004; Jøranli & Herstad, 2017), while OSLO O captures locations in the outer dwelling municipalities of the capital labour market regions. BERGEN is the second-largest housing and labour market region; STAVANGER is the third and TRONDHEIM the fourth. As the two outcome variables are binary and service innovation theory suggest they are interdependent (Amara et al., 2009; Gallouj & Savona, 2008; Witell et al., 2016), both are estimated simultaneously using the bivariate probit estimator in which the (potential) correlation between the binary outcomes is captured by the conditional correlation of the error terms (Filippini et al., 2018; Greene, 2018). When interaction effects are included in probit (or logit) models as
Referência(s)