Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Diversity of experience and labor productivity in creative industries

Diversity of experience and labor productivity in creative industries This paper studies how the previous experience among workers relates to the labor productivity of the creative indus- tries in Sweden. Eec ff tive knowledge transfers are dependent on the cognitive distance among employees. Using longitudinal matched employer-employee data, I measure the portfolio of the skills within a workplace through (i) the workers’ previous occupation, and (ii) the industry they have been working in previously. Estimates show that diversity of occupational experience is positive for labor productivity, but the diversity of industry experience is not. When dis- tinguishing between related and unrelated diversity, the relatedness of occupational experience is positive for labor productivity, while unrelated occupational experience instead shows negative relationship with productivity. These results point towards the importance of occupational skills that workers bring with them to a new employment, for labor productivity. Keywords: Diversity, Skill relatedness, Previous experience, Labor mobility, Knowledge spillovers JEL classifications: J24, L25 earnings and productivity (Parent 2000; Gathmann and 1 Introduction Schönberg 2010; Sullivan 2010). As people move across Research has often focused on the importance of differ - jobs they bring some knowledge which was specific to ent forms of human capital and firm performance (Del - what they were previously doing to the new employment gado-Verde et  al. 2016; Siepel et  al. 2017). However, the (Almeida and Kogut 1999). From a theoretical stand- productivity of workers within a firm also depends on point, the diversity of the workforce could foster creativ- who they work with (Mas and Moretti 2009; Card et  al. ity and innovation, where new knowledge is created from 2013; Arcidiacono et al. 2017; Neffke 2017). The question the recombination of differentiated skills (Schumpeter that then arises is how the composition of skills relates to 1934; Penrose 1959). However, if skills are too different, firm performance. The purpose of this paper is to exam - misunderstandings and conflicts can arise, which would ine how the diversity of skills which come from previous lead to negative effects on performance. experience within a plant matters for labor productivity. Moreover, for knowledge spillovers and learning to I specifically focus on the diversity of skills which arises happen, workers in a firm, need to have some sort of cog - from previous work experience and labor productivity in nitive proximity among each other (Nooteboom 2000). terms of (i) their previous occupation, and (ii) the indus- Along these lines, I further define diversity by distin - try they have been working in. Since the work of Becker guishing between the relatedness and unrelatedness of (1962), researchers have argued about the importance of experience. While previous literature in these lines meas- industry-specific and occupational-specific human capi - ures the relatedness of skills through either educational tal that people accumulate during their working life on background (Boschma et  al. 2009), previous industry experience (Timmermans and Boschma 2014), or previ- *Correspondence: orsa.kekezi@sofi.su.se ous occupational experience (Östbring et  al. 2017), it is SOFI, Stockholm University, Stockholm, Sweden Full list of author information is available at the end of the article Becker (1962) initially discussed firm-specific human capital, but that is not the focus of this paper. © The Author(s) 2021. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. 18 Page 2 of 21 O. Kekezi all of them which make up the skills of individuals. To my performance (Östbring et al. 2018). To sum up, combin- knowledge, the only previous study which considers mul- ing the project-based type of work, with labor-intensity tiple measures of skills is the one of Östbring et al. (2018) in production, as well as a high innovative potential, who use both education and previous industry experi- creative industries become a good case of study for ence. However, occupations are crucial to add as they are issues regarding the diversity of human capital and labor proxies of skills and abilities of the workforce beyond the productivity. It is also important to note that with the educational background (Bacolod et al. 2009). What peo- increasing focus on the knowledge economy, creative ple work with is sometimes argued to be more important industries are an important sector for regional develop- than their educational degree (Florida 2002). Hence, this ment (Florida 2002; UNESCO 2013). Thus, understand - paper contributes to the existing literature by proxying ing how these sectors become more productive and grow the diversity of skills within a workplace through their has implications for the economy at large. previous work experience, since people can bring with To answer the research questions, I use longitudinal them both industry-specific knowledge but also occu - matched employer-employee yearly data from 2007 to pational-specific one. By disentangling the type of skills 2016 for all the firms and individuals employed in crea - and experience brought into the firm, we can get a deeper tive industries in Sweden. I track the current employees understanding of the micro-mechanisms of knowledge 5 years back to see what type of experience they had. The transfer, knowledge spillovers, and labor productivity. diversity of skills is measured through a fractionaliza- This paper also contributes to the literature by apply - tion index. To disentangle whether diversity is related or ing this research question on creative industries, which unrelated, I use the relatedness index proposed by Nef- are the focus for several reasons. First, as knowledge- fke and Henning (2013), which is based on labor flows. intensive industries, they rely heavily on knowledge as Results show that the diversity of occupational experi- an input. Labor creativity is the main factor of produc- ence is positive for labor productivity, but the diversity of tion (Florida 2002), and they are characterized by tal- industry experience is not. Second, the unrelatedness of ented and high-ability individuals and firms which create industry and occupational experience are both negative new knowledge (Larsen 2001; And and Isaksen 2007). for labor productivity. On the other hand, the related- Employment in such industries is also inherently auton- ness of occupational experience within the workplace is omous and more self-expressive than more traditional positive for firm performance. Third, when experience is workplaces (Howkins 2002; Florida 2002). Second, crea- measured as a combination of industry and occupation, tive industries are characterized by a project-based pro- the relatedness of the two is positive and strongly related duction system and the production is dependent on the to productivity. These results point towards the impor - interaction of multiple agents (Caves 2000), who work tance of occupational specific skills for labor productivity in teams which are put to work together for a short time and indicate that the positive relation between the diver- (Jarvis and Pratt 2006). Interactions among employees sity of the workforce and productivity is mostly driven by are a crucial assumption when studying diversity within relatedness. a firm, because for productivity to be affected work - The paper is organized as follows. Section  2 describes ers need to work together or to interact with each other the theoretical framework and previous literature on for learning to happen. Given the high probability that skills, knowledge spillovers and growth. Section  3 pre- workers within firms in creative industries work together sents the data and variables. In Sect. 4 empirical findings to produce something, they become an interesting case and analysis of the results are shown, and in Sect.  5 the of study. Yet, their skill decomposition has not been stability of the results is checked. Section 6 concludes. extensively studied, with a few exceptions (Taylor and Greve 2006). Third, creative industries are widely seen in the literature as being innovative and the within-firm decomposition is an important determinant of innova- tion (Castañer and Campos 2002; Protogerou et al. 2017). Table  9 in the appendix shows the characteristics of plants that belong to creative industries (as defined on the paper) and the plants that do not for Last, by focusing on a similar set of industries, I am also 2007 and 2016 which is the time studied empirically. The data show that crea - able to mitigate issues arising from sectoral heterogene- tive industries have experienced a much larger growth in terms of employ- ity, which has been shown to give different results regard - ment, number of plants, as well as sales. Productivity growth does not differ between the two groupings, but the growth of wages is lower for creative ing the importance of diversity and relatedness on firm industries. The growth rate of the creative industries during this time period shows that they are an important segment of the Swedish economy, which is growing fast, and it employed about 9 percent of the workforce in 2016.More- over, they also indirectly support the economy by for example facilitating and Table  8 in the Appendix presents the list of industries included, adapted supporting innovation for other sectors in the economy (Müller et al. 2009). from Miguel-Molina et al. (2012). Diversity of experience and labor productivity in creative industries Page 3 of 21 18 et  al. 2014b, a). Boschma et al. (2009) look deeper at the 2 Diversity, relatedness, and firm performance type of educational diversity within firms and find evi - Research on workforce diversity and firm performance, dence that firms with higher education relatedness show broadly defined, is extensive. Some researchers have used higher productivity growth. Similar results are found in case studies and focused on team diversity (Horwitz and Östbring and Lindgren (2013), and the effect is stronger Horwitz 2007) as well as the composition of the top man- for labor-intensive industries than capital-intensive ones. agement and founding team (Bantel and Jackson 1989; However, proxying skills of the workforce through edu- Pitcher and Smith 2001; Visintin and Pittino 2014). Oth- cation has not come without critique in the literature, ers have used linked employer-employee data to examine since the quality of education is heterogenous, not only the within-firm diversity (Söllner 2010; Østergaard et  al. across countries but also across regions within a country 2011; Parrotta et al. 2014a, b; Solheim et al. 2020). On the (Mulligan and Sala-I-Martin 2000; Ingram and Neumann one side, the diversity of skills contributes to the creation 2006). Moreover, skills and human capital are to a large of new ideas and thus better performance (Bantel and extent collected from the working-life experience, some- Jackson 1989; Lazear 1999; Taylor and Greve 2006; Ber- thing that education does not capture. Becker (1962) dis- liant and Fujita 2011). Firms with more diverse knowl- cussed that human capital can be general which increases edge bases also have higher “absorptive capacity”, i.e. productivity no matter the job people have, but it can accumulated knowledge to understand and use the new, also be specific to the firm people are working. Specific incoming one, which is crucial for innovation and growth human capital can therefore not be transferred across (Cohen and Levinthal 1990). On the other side, for cer- jobs. Extending Becker’s work, literature has discussed tain tasks, Kremer’s O-ring predicts that workers with that human capital is also industry (Neal 1995), or occu- similar skills should work together to see higher produc- pation-specific (Kambourov and Manovskii 2009). u Th s, tivity returns (Kremer 1993). Moreover, people might as people change industries or occupations, there are prefer working with others whom they see as similar. If skills which cannot be transferrable. This indicates that diversity leads to misunderstandings, conflicts, or unco - if workers within a firm have very different skills, work - operativeness across workers, negative effects of diver - ing together would not necessarily be beneficial as they sity are observed (Bassett-Jones 2005; Jehn et  al. 1999; would not understand each other, which goes back to the Madsen et  al. 2003; Williams and O’Reilly 1998). Thus, cognitive proximity argument (Nooteboom 2000). how diversity impacts firm performance is an empirical A complementary measure of human capital often question. used in the literature is through different occupations In a theoretical contribution, Lazear (1999) argues that individuals had (Thompson and Thompson 1985; however that for diversity to have a positive effect on per - Florida 2002; Florida et  al. 2008; Scott 2008). Occupa- formance, the skills of the workforce should be disjoint tions measure the practical skills of people, beyond their but still relevant to one another. Moreover, they should formal education (Bacolod et al. 2009; Wixe and Anders- be learnt by the other groups at a not too high cost. Thus, son 2016). The diversity of occupations within a firm has for learning to happen, some level of cognitive proxim- not been extensively studied, but the existing literature ity or complementarity is required (Nooteboom 2000). If suggest a positive effect on innovation (Söllner 2010; Par - the knowledge bases of the firm are too different, people rotta et  al. 2014b). Östbring et  al. (2017) further suggest do not understand each other. Yet, too much cognitive that the positive effect of occupational diversity on pro - proximity might create a lock-in problem that disables ductivity is driven by relatedness because the unrelated- the capability of companies to adopt new technologies or ness of occupations in a firm either displays insignificant market possibilities (Boschma 2005). Nooteboom et  al. or negative effect. Besides education and occupation, (2007) find for instance an inverted U-shaped impact of human capital can also come from industry experience the cognitive distance and innovation of firms, indicating (Neal 1995). Östbring et al. (2018) have studied how the that knowledge shouldn’t be too similar or too different relatedness of industry experience in knowledge-inten- for innovation to happen. To take the cognitive distance sive business services impacts firm performance. Their into account, the notion of relatedness has emerged in results show that for single-plant firms, the variety of the literature, where several studies, stemming from the knowledge and previous industrial experience affect firm work of Frenken et al. (2007), have distinguished between performance positively. related and unrelated diversity (Boschma et al. 2009; Öst- To sum up, the literature has previously investigated bring and Lindgren 2013; Östbring et al. 2017, 2018). the importance of educational diversity, occupational When examining the effect of skill diversity on firm diversity, or diversity of industrial experience on firm performance, most existing studies focus on the diver- performance. Their results point toward a positive impact sity of educational background, where the results often of diversity, but these effects seem to be stronger in the show a positive effect (Østergaard et  al. 2011; Parrotta 18 Page 4 of 21 O. Kekezi case of related diversity. Yet, Timmermans and Boschma particularly challenging in the smaller firms (Christo - (2014) find that it is the unrelatedness which matters for pherson 2004; Hotho and Champion 2011). When it productivity growth of firms in the region of Copenha - comes to the decomposition of the team, Taylor and gen in Denmark. They speculate that it could be because Greve (2006) and Perretti and Negro (2007) find evidence Copenhagen is mostly characterized by service industries that creative industries especially benefit from teams with compared to the rest of Denmark, which might benefit diverse skills. Thus, the literature on firm diversity and mostly from unrelatedness. Therefore, we do not know firm performance discussed at the beginning of Sect.  2, a priori what type of diversity matter most for creative is highly relevant and applicable to the creative indus- industries. tries. Moreover, because the probability of teamwork is Moreover, these studies primarily study the diversity of higher in such industries, the results obtained would give the current occupation individuals have, and not at their a clearer and more accurate picture on the importance of occupational and industrial history. From a theoretical diversity for knowledge spillovers and productivity. perspective, the knowledge of workers is also shaped by Moreover, creative industries are characterized by their previous experiences and job tasks. When people high labor mobility (Florida 2002; Frederiksen and Sed- change jobs, the skills that they have accumulated are ita 2011). Florida (2002) also identifies creative work - not necessarily left behind but instead brought into their ers as mobile in their career choices, since they have the new workplace (Almeida and Kogut 1999). While labor skills and education to change jobs or careers. This can mobility has been extensively studied, we do not know be directly connected to the structure of such industries enough on the type of knowledge and skills are brought which are characterized by a lot of small firms with high into the firm and how that affects performance (Boschma entry and exit rates (Power 2003). Thus, the probability et al. 2009; Timmermans and Boschma 2014). Therefore, that workers have previous experience from other indus- the skills that people bring can come from their previ- tries and occupations is higher. Furthermore, they are ous industry experience, from previous occupations they labor intensive and they usually employ high-skilled indi- might have had, or from a combination of the two. Sul- viduals who create new knowledge (Larsen 2001; Wiig livan (2010) argues that human capital is both connected Aslesen and Isaksen 2007). to the industry and occupation. Similarly, the literature What is also important to note is that skills obtained on job polarization treats a “job” as an occupation-indus- from occupational experience are especially important try interaction (Autor et  al. 2003; Goos and Manning for people working in creative industries. By definition, 2007). The reason for using a combination of the two is creative industries, are characterized by a high concen- that there are industry effects on wages, after controlling tration of creative workers. Florida’s (2002) creative class for the occupation. is based on the occupations people have and what they do in their everyday tasks, rather than the industries 2.1 W hy employee diversity in creative industries? where they are employed. Moreover, the occupational The literature covered so far does not specifically focus distribution across industries can be heterogenous. For on creative industries, raising the questions on how it instance, a high-tech firm employs accountants, engi - relates to them, as well as what can we learn from study- neers, manufacturing jobs, as well as service jobs at the ing the diversity of skills in such sectors. Creative indus- food court (Mellander 2009). Along these lines, Barbour tries are a good case of study for this research question and Markusen (2007) discuss that the occupational struc- for several reasons. ture of high-tech industries in California is different from Researchers have increasingly argued that workers in the rest of the US. Thus, these results hint towards the creative industries are likely to collaborate and work in idea that industry-specific skills might not be equally teams (Caves 2000; Jarvis and Pratt 2006; Uzzi and Spiro important for creative industries. 2005; Savino et  al. 2017). Moreover, due to the project- based character of these industries, the workforce if 3 Data, variables, and method constantly required to readjust and form new teams To examine the relatedness of the previously acquired since projects are often short-term, which can become skills among workers on labor productivity in the crea- tive industries, I use register longitudinal matched employer-employee yearly data, collected from Sta- tistics Sweden, during 2007–2016. To allow plants to In a related strand of literature, studies have indeed looked at the impor- reach some skill diversity, similar studies drop plants tance of industry or occupational experience (not combined) in a firm, for with less than 10 employees (Parrotta et  al. 2014a, b). wages, firm survival as well as productivity (Timmermans and Boschma 2014, Martynovich and Henning 2018, Jara-Figueroa et al. 2018). However, the focus However, creative industries in are characterized by of these studies is on relatedness to the current job rather than relatedness small firms which is clearly shown in Fig.  1 below. To across workers within the workplace. Diversity of experience and labor productivity in creative industries Page 5 of 21 18 Fig. 1 Distribution of plant size Table 1 The most common occupations and industries where workers employed in creative industries come from Occupation Industry Computing professionals Computer programming activities Physical and engineering science technicians Computer consultancy activities Architects, engineers and related professionals Construction and civil engineering activities and related technical consultancy Writers and creative or performing artists Advertising agency activities Finance and sales associate professionals Industrial engineering activities and related technical consultancy Managers of small enterprises Business and other management consultancy activities Shop, stall and market salespersons and demonstrators Engineering activities and related technical consultancy in energy, environ- ment, plumbing, heat and air-conditioning Business professionals Architectural activities Computer associate professionals Other software publishing Artistic, entertainment and sports associate professionals Technical testing and analysis make the visualization clearer, all plant with more than 3.1 Variables and method 50 employees are put together in the last bar.3.1.1 Dependent variable The figure shows the distribution of firm size, where The outcome variable is labor productivity, measured about 62 percent of the firms only have one employee as value-added per employee, in its logged form. Previ- and an additional 12 percent have only two employees. ous research usually measures the effect of relatedness To not exclude too many of the firms in the creative of skills on productivity growth with a time lag of more industries, and to be able to give a representative pic- than one year due to the time it may take for the knowl- ture of the creative industries, I keep firms that have at edge spillovers to influence growth. However, due to the least 3 employees, where at least some level of diversity project-based characteristics of some creative industries, is reached. the short-term effects of such spillovers are of interest. I track the current employees five years back in time u Th s, a one-year lag is implemented. Following Tim - to see what type of experience they had. If they have mermans and Boschma (2014), for multi-plant firms, the changed industries or occupation several times, the most value-added across plants is distributed according to the recent is considered. If they have been working in the distribution of wages. same industry and occupation in the past 5 years, the cur- While measuring productivity through value-added rent job is considered. During the time of the study, the is common, using value added for creative industries experience of the workers in creative industries comes might be cumbersome (Maroto-Sánchez 2012). In broad from 113 different occupations and approximately 700 terms, productivity refers at the ability of a firm to gen - industries. Table  1 shows the 10 most common occupa- erate outputs from a set of inputs. Service sectors, in tions and industries that the workers currently employed general, do not have the same inputs or outputs as the in creative industries have experience on. traditional manufacturing firms, creating so difficulties 18 Page 6 of 21 O. Kekezi in measuring labor productivity (Van Ark 2002). There - (2013), industry changes of individuals who earn less fore, besides value added, results are also estimated using than the industry median wage as well as managers are wages. Assuming that wages also reflect labor productiv - excluded since these are individuals who are not very ity (Becker 1964; Mincer 1974), a more efficient flow of likely to have industry-specific skills. The intuition is knowledge across employees would indicate higher pro- that we want to capture industries that require similar ductivity and thus higher earnings. skill sets. Inter-industrial moves of individuals who do not have industry specific skills, would not give us that 3.1.2 Measuring diversity and relatedness of skills information. I then run a zero-inflated negative binomial In the first step, following Parrotta et  al. (2014b), the regression with pairwise industrial flows as the depend - diversity of skills is measured through a fractionalization ent variable. The independent variables are the employ - index (Alesina et al. 2003) which is computed at the plant ment size, average wage, as well as wage growth in the level as one minus the Herfindahl index: origin and destination industries. Using the point esti- mates obtained, the predicted labor flows are calculated for each industry pair. The SR measure is: Fract = 1 − p wt (1) wst s=1 obs ij SR = (2) ij where w denotes the workplace, s is the variable for which ij the diversity is computed, and t is time. p is square of the obs share of workers within each category s, each year. The where F and F are the observed and predicted flows ij ij index takes the minimum value of zero if there is only respectively. A value of larger than 1 indicates that the one category present in the workplace and its maximum observed flows are larger than predicted, making the value occurs when all categories are distributed equally: industries related. A ratio of lower than 1 shows that the (1 − ) . The index is measured for the diversity of occu - industries are skill dissimilar. In the last step, arguing that pational and industry experience. the probability for an individual to move from industry Besides diversity itself, following the discussion pre- i to j is the following, it is possible to statistically test sented in the literature review, it is also interesting to whether the observed flows are exceptionally large: look at whether the degree of diversity matters. Frenken ij et  al. (2007) proposed the entropy measures of related pˆ = (3) ij emp and unrelated variety, which have been often used in the literature to measure the degree of diversity within a SR is significant and higher than 1 in 4 percent of all firm (Boschma et  al. 2009; Östbring and Lindgren 2013; industry combinations. The NACE industrial classifica - Östbring et al. 2017, 2018). However, these measures are tion changed in 2007, where the industries were split and dependent on industry or occupational classifications aggregated differently, creating difficulties into translat - which do not fully capture the degree of relatedness or ing the old industrial codes to the new ones. Thus, the cognitive proximity since they are arbitrarily decided skill-relatedness index is constructed in the same way (Essletzbichler 2015). for the new codes for labor mobility during the years Therefore, to define related and unrelated industries 2010–2013. and occupations, I rely on the revealed skill-relatedness 5 However, human capital is also dependent on the type (SR) measure proposed by Neffke and Henning (2013). of job workers have in the firm. Gathmann and Schön - The main assumption behind SR is that individuals are berg (2010) find that people are more likely to switch more likely to switch jobs across industries where their occupations across those jobs where they can use their skills can partly be used. The steps described below fol - skills more. Thus, the skill relatedness matrix is also con - low the original paper and are based on the labor flows structed for the 3-digit occupational codes in the same of the working population in Sweden. First, a matrix with way as explained above. The main difference between pairwise labor flows for all 5-digit industry codes during this calculation and the industrial relatedness one is that 2004–2007 is constructed. Like in Neffke and Henning While the Neffke and Henning (2013) skills relatedness index is well-estab - The empirical estimations are however relatively stable even when managers lished in the literature, the index does not consider the geography of labor and people who earn less than median wage are included in the SR. They are mobility. People are more likely to switch jobs in the areas where they live or available upon request. work (Manning and Petrongolo 2017), thus industrial mobility is partly con- strained to the industries available in the region. This concern does however Since the period studied is 2007–2016, for 2007–2010, the relatedness of not change the findings of the paper, neither the suitability of the skill-related - experience is calculated through the old classification and for 2011—2016 ness measure for the research question. with the new one. Diversity of experience and labor productivity in creative industries Page 7 of 21 18 labor flows are not measured each year, but rather every calculated with a time lag of one year to allow for the second year. The reason is that Statistics Sweden does not knowledge spillovers to take place. Ŵ represents a vector collect data regarding occupations for the full population of the plant specific control variables, and Z represents each year. After two years approximately 80 percent of the vector of the region-specific characteristics, D and D f t the population is covered, which makes the occupational are fixed effects on the firm, and time. switches more reliable. About 13 percent of the combina- One problem that the literature has pinpointed how- tions are statistically related to each other. ever, is that the error term consists of ω which is a it Following the steps above, the industry- and occupa- productivity shock observed by all firms but not by econ - tion-pairs which are skill-related to each other are iden- ometricians, while η is observed by both firms and econ - it tified. To aggregate this to a plant level, I first identify ometricians as shown below: all possible industry (and occupation) combinations of u = ω + η it it it experience between workers within a plant. Then, the number of all combinations which are statistically signifi - For that reason, the estimates observed by linear cant with a SR above one is divided by the total number regressions usually show upward biases in the coeffi - of combinations to calculate the share of relatedness in cient of labor and the coefficients for capital are down - a firm. In the same way, the number of people with the ward bias. u Th s, the methodology developed by Olley same industry experience is divided by the total number and Pakes (1996) (henceforth, OP). The OP estimation of combinations. The rest is the share of skill unrelated - is a semi-parametric method which is calculated on the ness in a firm. Since the three shares add up to one, the identification of a proxy variable which is assumed to similarity of skills is not included in the estimations. be a function of ω productivity shocks. The proxy vari - it able is often investments which firm make, which are 3.1.3 Method assumed to increase productivity. Therefore, they sug - To study how relatedness of skills relates to on average gest the use of a control function approach, which con- labor productivity, a linear regression model with fixed trols for the endogeneity of labor, where investments are effects is used. The panel is not balanced, since firms used to replace the unobserved productivity shock. Fol- can enter and exit during the time studied. As in many lowing Tao et al. (2019), investments are measured as the studies where the dependent is productivity, the starting change of fixed assets. Similar two-step approaches have point is often the Cobb–Douglas production function, often been used in the literature to infer productivity by where productivity of plant i at time t is a function of observing the input choices of the firms (Parrotta et  al. technology (A), capital (K), and labor (L): 2014a; Serafinelli 2019; Tao et al. 2019). Y = AL K (4) it it it3.1.4 Control variables Following the Cobb–Douglas production function, labor However, since I am interested in productivity per and capital are included in the estimations. Besides, the employee, we can divide everything by L, allowing the diversity of educational background is also controlled Cobb–Douglas to take the following form: for in the empirical model. The main reason for doing α so is to ensure that our measures of diversity of work Y AL K it it it α−1 (5) = y = = AL K it are not driven by the diversity of the education tracks. it it L L it it Previous literature has mostly found a positive effect between the diversity of education and labor productivity In order to facilitate the empirical estimation, the (Østergaard et  al. 2011; Parrotta et  al. 2014b, a). Share model is estimated in its logarithmic form where all the of workers with high education, plant age, and whether control variables were captured in the A parameter in the the firm is multi-plant or not are further controlled for previous equations: in the model (Östbring and Lindgren 2013; Wixe 2015). ln y =δ ln L + β ln K + ϕ Div + ϕ ln Ŵ it it it 1 it−1 2 it Since knowledge can also be region-specific, to examine + ϕ ln Z + ϕ D + ϕ D + u 3 rt 4 f 5 t it the importance of skills acquired in a different region (6) (Timmermans and Boschma 2014; Boschma et al. 2009), where δ = α − 1 and since α< 1 by definition, the coef - the share of workers who have worked in another labor ficient of labor in this case is expected to be negative. Div are the plant diversity and relatedness which are it−1 8 9 In 2014 the occupational codes changed, and the new codes were manually The education tracks are presented in respectively Table 10 in the Appendix. matched with the old ones. 18 Page 8 of 21 O. Kekezi Table 2 List of Variables and descriptive statistics Variables Measured as Mean SD Min. Max. Outcome variables Avg_Prod (000) Value added per labor 937.217 7396.918 0.094 809,103.1 Wages (00) Average yearly wage in the plant 3865.126 1366.513 3.667 25,718.73 Diversity and relatedness measures FRACT_occu 1 minus the Herfindahl index of the diversity of occupation experience 0.641 0.206 0 0.959 FRACT_ind 1 minus the Herfindahl index of the diversity of industry experience 0.568 0.245 0 0.976 Occ_R Share with related occupation experience 0.44 0.231 0 1 Occ_U Share with unrelated occupation experience 0.313 0.238 0 1 Ind_R Share with related industry experience 0.318 0.229 0 1 Ind_U Share with unrelated industry experience 0.337 0.255 0 1 Occ_Ind_R Share with related occupation and industry experience 0.161 0.165 0 1 Occ_Ind_U Share with unrelated occupation and industry experience 0.149 0.177 0 1 Control variables K (000) Capital 29,533.66 1,137,329 0 1.80E + 08 L Labor–plant size 20.859 57.582 3 3331 FRACT_Edu 1- the Herfindahl index of the diversity of education tracks 0.578 0.215 0 0.91 Edu Share with at least a 3-year university degree 0.352 0.296 0 1 Change_LA Share who have worked in another labor market 0.221 0.221 0 1 Age Years of operation 12.577 8.17 1 30 Multiplant Dummy = 1 if the firm has more than 1 plants 0.297 0.457 0 1 Den Population per square kilometer 1575.774 1943.169 0.2 5496.4 All independent variables are measured at time t, besides the diversity and relatedness which are measured in t-1, to allow time for the knowledge spillovers to take place. All monetary values are in SEK market is included. Last, I include population den- 4 Empirical findings and analysis sity in the municipality to account for the importance Table  3 presents the baseline results. In columns 1(a)– of agglomeration economies on labor productivity and (c) the linear regression results are presented when the wages (Wixe 2015; Glaeser and Mare 2001). Table  11 in dependent variable is average labor productivity. The the appendix presents the correlation matrix. No large Olley–Pakes estimations are presented in columns 2(a)– values are shown from there, indicating that multicollin- (c). The last columns, 3(a)–(c) present linear regression earity is not a problem in this dataset. Table 2 presents models when average wages are instead used as depend- the list of variables used in the estimations. ent variables. The last columns of Table  2 present the descriptive sta- Starting with the diversity variables, measured through tistics, when the variables are in their non-logged form. the fractionalization indices (columns 1(a), 2(a), 3(a)), The fractionalization indices show that individuals have results show that the diversity of previous occupation rather broad backgrounds. The diversity of occupational experience between employees is positively related to experience is on average higher than the industrial one. average labor productivity, as well as wages, the year On average about one third of the employees have a later. However, the fractionalization index of industrial higher education. The table also shows that there are experience does not display a significant relationship for many small workplaces where the mean size is 21, but the labor productivity, but it shows a negative and signifi - median size is 9. Small workplaces are not uncommon for cant relation in the OP estimation as well as for wages. creative industries as shown in Fig. 1 above. 30 percent of These results indicate that having people with differ - the workplaces belong to multi-plant firms. ent occupational backgrounds work together is posi- tive for productivity while having individuals who come from many different industries is not. The findings about occupations are in line with Parrotta et  al. (2014b) and Sweden has 81 labor market regions which consist of several municipalities. Söllner (2010), but their outcome is innovation and not Multicollinearity is also tested through the VIF value in the regressions productivity. Regarding the industrial experience, one and the VIF values are very low, indicating that multicollinearity is not an can speculate that the results might be driven by the issue. Diversity of experience and labor productivity in creative industries Page 9 of 21 18 Table 3 Baseline results Value added – FE Value added – OP Wages – FE 1(a) 1(b) 1(c) 2(a) 2(b) 2(c) 3(a) 3(b) 3(c) FRACT_occu 0.063*** 0.038** 0.021*** (0.009) (0.016) (0.005) FRACT_ind 0.007 − 0.174*** − 0.025*** (0.008) (0.016) (0.004) Occ_R 0.026*** 0.042*** 0.024*** (0.008) (0.014) (0.004) Occ_U − 0.010 − 0.184*** − 0.016*** (0.009) (0.015) (0.004) Ind_R 0.014* − 0.010 − 0.003 (0.008) (0.014) (0.004) Ind_U − 0.022*** − 0.235*** − 0.036*** (0.008) (0.014) (0.004) Occ_I 0.031*** 0.126*** 0.027*** (0.009) (0.018) (0.005) Occ_Ind_U − 0.035*** − 0.296*** − 0.040*** (0.009) (0.016) (0.005) Capital 0.021*** 0.021*** 0.021*** 0.020*** 0.022*** 0.021*** 0.004*** 0.004*** 0.004*** (0.002) (0.002) (0.002) (0.005) (0.005) (0.005) (0.000) (0.000) (0.000) Labor − 0.077*** − 0.072*** − 0.072*** − 0.060*** − 0.073*** − 0.076*** − 0.022*** − 0.022*** − 0.022*** (0.003) (0.003) (0.003) (0.006) (0.006) (0.006) (0.002) (0.002) (0.002) FRACT_Edu 0.079*** 0.097*** 0.097*** − 0.073*** − 0.021 − 0.057*** 0.026*** 0.032*** 0.029*** (0.011) (0.011) (0.011) (0.018) (0.017) (0.017) (0.006) (0.005) (0.005) Edu 0.044*** 0.050*** 0.049*** 0.191*** 0.181*** 0.188*** 0.045*** 0.047*** 0.045*** (0.011) (0.011) (0.011) (0.014) (0.013) (0.014) (0.006) (0.006) (0.006) Age 0.002*** 0.002*** 0.002*** − 0.005 − 0.004 − 0.004 0.001*** 0.001*** 0.001*** (0.000) (0.000) (0.000) (0.197) (0.191) (0.189) (0.000) (0.000) (0.000) Multiplant 0.036** 0.035** 0.035** 0.169*** 0.146*** 0.159*** 0.005 0.005 0.005 (0.015) (0.015) (0.015) (0.009) (0.009) (0.009) (0.004) (0.004) (0.004) Change_LA − 0.088*** − 0.083*** − 0.084*** − 0.036** − 0.020 − 0.059*** − 0.050*** − 0.048*** − 0.051*** (0.009) (0.009) (0.008) (0.015) (0.015) (0.015) (0.004) (0.004) (0.004) Den 0.028*** 0.027*** 0.027*** 0.051*** 0.045*** 0.047*** 0.022*** 0.022*** 0.022*** (0.001) (0.001) (0.001) (0.005) (0.005) (0.005) (0.001) (0.001) (0.001) Obs. 88,078 88,078 88,078 88,078 88,078 88,078 88,078 88,078 88,078 Plants 15,983 15,983 15,983 15,983 15,983 15,983 15,983 15,983 15,983 R-squared 0.785 0.785 0.785 0.894 0.894 0.894 Year FE ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ Robust standard errors in parentheses for columns 1 and 3. For the OP estimations, bootstrapped standard errors are presented with 250 replications. *** p < 0.01, ** p < 0.05, * < 0.1. Estimates of the constant term is not reported non-transferability of industry human capital. Having Given the differences in the results for diversity, it is individuals with different backgrounds working together interesting to see whether the results differ for the degree can come with difficulties in communicating due to for of diversity (columns 1(b), 2(b), 3(b)). Previous findings example different routines they are used to. However, this are inconsistent because these skills are obtained through can also have to do with the nature of creative industries different mechanisms and measure different types of where the main focus is put on the creativity of individu- knowledge. Results show that the relatedness of occupa- als which is more connected to the occupational tasks tional experience is positively related to labor productiv- rather than the industry specific ones. ity and wages, but the relatedness of industry experience is statistically insignificant (besides in column 1(b) at a 18 Page 10 of 21 O. Kekezi 10% significance level). The importance of related occu - can be that the firms do not always have control over the pational experience is also supported by Östbring et  al. diversity of the workforce (Parrotta et  al. 2014a). This (2017). However, the negative sign regarding industrial would especially be true in Sweden, where firms are relatedness does not support the positive results found increasingly experiencing difficulties to find the right by Östbring et  al. (2018) for KIBS industries. Unrelated- person for the right job. This problem of job matching is ness of both industry and occupational experience are even more pronounced in knowledge intensive sectors. negatively related to firm performance. These results Another reason can be that even is some skill combina- suggest that the positive sign of the fractionalization tions are unrelated, they might be complementary to index on the diversity of occupational experience is most each other, which can also matter for productivity and likely driven by the related occupational diversity rather wages, as shown by Neffke (2017). Moreover, firms might than the unrelated one. The negative sign for the rela - not have full information on the type of skill mixture tion observed with the unrelated experience, for occupa- would create the highest productivity. Given that creative tions and industry experience, is not surprising given the industries are characterized by mostly smaller plants, this importance of cognitive proximity (Nooteboom 2000). If assumption would not be unrealistic. the skills are too different, no knowledge spillovers would However, it can also be the case that the composition be possible. of the workforce within a plant is endogenous. One issue In the last columns for the three specifications, I now that could be problematic when looking at firm produc - estimate the relationship between the relatedness and tivity, is the endogeneity of hiring where the more pro- unrelatedness of previous experience measured through ductive firms would hire the more productive workers. If the combination of occupation and industry experience. the diversity of previous experience is endogenously cho- While, to my knowledge, this has not been previously sen by the firm to enhance productivity, the regressions estimated in the firm relatedness literature, a branch of presented in Table  3 should be analysed with caution. A labor economics argues that the skills of the individuals common way to deal with this issue is through the use of come from the tasks they perform which is connected to instrumental variables (IV). However, after trying a few both industry and occupation. Results, once more, show different IVs, they showed to be weak ones. Given that that the relatedness of experience is positively related to weak instruments are biased and uninformative (Young productivity and wages, but the unrelatedness of experi- 2019), they were not included in the paper. It is impor- ence is negatively related to the firm performance. The tant to point out that even if there would be endogeneity fact that the magnitude of the coefficients is also much in hiring, relying on economic theory and previous litera- higher than in columns 1(b), 2(b), and 3(b), hints towards ture, the direction of the relationship should be going the the idea that a combination of skills obtained from direction tested in the paper. Diversity is important for industry and occupation is specifically important for the creation of new ideas, and the results shown for relat- knowledge spillovers which are then mirrored in higher edness support the existing research they are based upon. productivity or higher wages. Unrelatedness of experi- Moving on to the control variables, the diversity of ence, in this case, is negatively related to the outcomes. educational experience is positive and significant in the The results hint towards the idea that the diversity of linear specifications, supporting the existing research occupational experience is important for labor produc- (Østergaard et  al. 2011; Parrotta et  al. 2014b, a). How- tivity, but the diversity of industrial experience shows no ever, the OP estimation shows a negative and significant significance. When separating between the type of diver - relationship between education diversity and productiv- sity, the positive results seem to be solely driven from the ity. Thus, the interpretation of that results should be with relatedness of previous occupational experience. These care. The rest of the control variables take the expected results are in general in line with the existing literature on signs. Note that the negative sign taken by the plant knowledge flows and relatedness (Boschma et  al. 2009; size comes because the dependent variables are divided Östbring and Lindgren 2013; Östbring et al. 2017, 2018). with labor (as shown in Eqs. 4–6 above). Labor elasticity Unrelatedness of industry and occupation experience is about 0.93 in all estimations, which is relatively large shows an either insignificant or negative relationship to compared to the norm in the production function litera- productivity and wages. ture. This can however be driven by the that that creative The question that could arise is why firms would build a workforce with unrelated work experience. One reason Table 11 in the appendix shows the Akaike Information Criterion (AIC) for different model specifications to ensure that the inclusion of the diversity and Footnote 12 (continued) relatedness variables improve the model fit. As shown in the table, the inclu - the model improves. This is further suggestive evidence regarding the impor - sion of the variables decreases the AIC value, indicating indeed that the fit of tance of workforce diversity for labor productivity. Diversity of experience and labor productivity in creative industries Page 11 of 21 18 industries are heavily dependent on labor as an input and to the baseline model. Diversity of occupation is how- much less on capital (with an elasticity of about 0.02). ever only significant in the first estimation (column Capital is positive and significant to firm performance. 1(a)). Occupational relatedness is positive for productiv- The share of highly educated and workplace age are ity while unrelatedness is negative. On the other hand, positively related to labor productivity and wages. Multi- industry relatedness shows a positive relation to value plant firms and older firms which show higher labor pro - added in the OP estimation but a negative in the wages ductivity. However, hiring individuals who have worked model. As in Table  3, unrelatedness of industry experi- in other regions negatively relates to productivity. This ences is negatively related to productivity. The last col - can have to do with knowledge being rather localized umns in the three estimations (1(c), 2(c), 3(c)) confirm the and people working in different regions follow different relationship found above. Interestingly, the share of new routines and other ways of solving problems. It can also employees is negative for both productivity and wages. be the case that there is a need for an adjustment period This means that for the positive spillovers to emerge from which is not considered since this variable is measured the churning of employees, more time is needed. Other at time t. Regarding the regional variables, as expected, research also suggests that teams become more produc- the workplaces located in denser regions, are also the one tive the longer they work together (Bercovitz and Feld- showing higher productivity (Wixe 2015). man 2011). These results have broader implications for policy Moreover, the estimations are also run on a subset of when it comes to workforce building for creative indus- plants that do not experience any change in the work- tries. Rather than focusing on the skills of one individual, force during the time they are in the sample. While there it is important to look at how well it matches the skill sets might be a selection of firms that satisfy this condition, of the people currently employed in the firm. A higher this exercise is still useful for us to understand the under- degree of relatedness is positive for labor productiv- lying mechanisms behind labor productivity. Results are ity (no matter if it is measured through value added or presented in Table  13 in the appendix and they show a wages), which is crucial for firm growth. These results slightly different picture where the diversity measures are of great importance in countries like Sweden where show a negative relationship to labor productivity. The knowledge-intensive firms are constantly struggling to relatedness measures for industry or occupation are now find the right person for the right job. Given that most statistically insignificant in all estimations, while their firms hire from the local labor market region, results sug - unrelatedness is still negatively related to productivity. gest that when deciding on the plant location it might be However, a higher share of related and industry experi- of importance to study the composition of skills in the ence is positively related to labor productivity, which is labor market as well. in line with the previous results. Similarly, unrelated- ness of industry and occupation is negatively related 5 Robustness and stability to productivity. What this table shows is that the posi- To further check the stability of the results, three differ - tive results of relatedness of industry or occupation ent sets of specifications are shown and discussed. experience might be partly driven by the churn in the labor force, where new related knowledge is crucial for 5.1 D iversity of experience or churn of employees? the productivity boost of employees. However, even if Labor mobility in creative industries is generally high there are no changes in the employees, they still benefit (Florida 2002; Frederiksen and Sedita 2011). It is there- from the relatedness of experience in both industry and fore important to ensure that the relationship found occupation. between diversity of experiences and productivity is not only driven by the hiring and firing behavior of the firm. 5.2 Plant size Two different estimations are shown to ensure that this To see if there are any differences between the smaller is indeed the case. First, Table  4 below shows the results and larger workplaces, the sample is separated between when share of new hires as well as the share of people the ones that employ at least 10 employees and the ones who have left the firm are included as control variables. that employ less than 10 employees. Note that previ- Even when the churn of the employees if controlled for, ous literature literature drops firms with less than 5 the diversity and relatedness variables behave similarly This is also confirmed when controlling for how long employees have To ensure that changes in occupational codes in 2014 are not driving the worked together. The results on the main variables of interest are unchanged, results, one more robustness test is shown in Table  12 in the appendix. The and therefore that analysis is not included in the paper. The results when con - model is now estimated for the period 2007–2014. Results are in line with trolling for how long the employees have worked together are available upon what has been presented earlier in the paper. request. 18 Page 12 of 21 O. Kekezi Table 4 Productivity estimations when controlling for new hires and those who have left the firm Value added – FE Value added – OP Wages – FE 1(a) 1(b) 1(c) 2(a) 2(b) 2(c) 3(a) 3(b) 3(c) FRACT_occu 0.031*** 0.024 0.004 (0.009) (0.016) (0.004) FRACT_ind − 0.028*** − 0.121*** − 0.044*** (0.008) (0.015) (0.004) Occ_R 0.021*** 0.050*** 0.022*** (0.008) (0.014) (0.004) Occ_U − 0.016* − 0.166*** − 0.019*** (0.009) (0.015) (0.004) Ind_R − 0.004 0.028** − 0.013*** (0.008) (0.014) (0.004) Ind_U − 0.044*** − 0.182*** − 0.047*** (0.007) (0.014) (0.004) Occ_Ind_R 0.019** 0.151*** 0.021*** (0.009) (0.017) (0.005) Occ_Ind_U − 0.053*** − 0.246*** − 0.049*** (0.009) (0.015) (0.004) Share hires − 0.356*** − 0.359*** − 0.357*** − 0.576*** − 0.544*** − 0.572*** − 0.192*** − 0.189*** − 0.187*** (0.009) (0.009) (0.009) (0.015) (0.015) (0.015) (0.004) (0.004) (0.004) Share left 0.017* 0.018* 0.017* 0.066*** 0.054*** 0.056*** 0.013*** 0.013*** 0.012*** (0.010) (0.010) (0.010) (0.016) (0.016) (0.016) (0.004) (0.004) (0.004) Observations 88,078 88,078 88,078 88,078 88,078 88,078 88,078 88,078 88,078 Plants 15,983 15,983 15,983 15,983 15,983 15,983 15,983 15,983 15,983 R-squared 0.791 0.791 0.791 0.898 0.898 0.898 Robust standard errors in parentheses for columns 1 and 3. For the OP estimations, bootstrapped standard errors are presented with 250 replications. ***p < 0.01, **p < 0.05, * < 0.1. The constant term is not reported. Control variables and year fixed effects are included in all estimations 5.3 Plant age employees (Östbring et  al., 2018) or 10 employees (Par- Following Timmermans and Boschma (2014), Table  6 rotta et  al., 2014a, 2014b). The results are presented in presents the results for plants which are at least 5  years Table 5 below. old, since the younger firms suffer from the liability of They are mostly in line with what has been shown newness (Stinchcombe 1965). In the sample, approxi- before for the larger firms. The main difference is that mately 26% of the plants are now dropped. smaller plants do not seem benefit from the relatedness Results are in line with the baseline estimation and of occupations or industries, but rather from the related- show that the diversity of occupational experience is ness of industry and occupation relatedness. Unrelated- positively related to labor productivity, but the diversity ness is however still negatively related to productivity. of industry experience is not. When looking at related- There can be two different explanations to why we do not ness, the relatedness of occupations, the relatedness of observe significant results for the relatedness of occupa - industry experience, as well as the relatedness in their tions. Small firms might not reach as high levels of relat - combination are positive for labor productivity. However, edness as the larger firms, and there too little variation industry relatedness shows no significance on the wage in the variable to show significant results. It can be the estimation. On the other hand, the unrelatedness of both case that due to the low number of workers people in the industry and occupation experience are negatively related plant need to work with different tasks simultaneously to firm performance. This suggests the importance of and work all together rather than to be separated into cognitive proximity among workers, when it comes to teams. It can also be that firms need to reach a specific knowledge spillovers and productivity advantages. size to benefit from relatedness. However, the relatedness Related to this issue, to understand what facilitates the of both industry and occupation experience is significant success of new firms in creative industries, Table  7 pre- for both categories. sents the results for start-ups instead of the older firms. Diversity of experience and labor productivity in creative industries Page 13 of 21 18 Table 5 Results with plants when a cutoff of 10 employees is made Average value added – FE Average value added – OP Average Wages – FE 1(a) 1(b) 1(c) 2(a) 2(b) 2(c) 3(a) 3(b) 3(c) More than 10 employees FRACT_occu 0.063*** 0.042 0.021*** (0.016) (0.031) (0.007) FRACT_ind − 0.019 − 0.263*** − 0.041*** (0.013) (0.024) (0.005) Occ_R 0.073*** 0.173*** 0.052*** (0.017) (0.031) (0.008) Occ_U − 0.019 − 0.193*** − 0.036*** (0.019) (0.033) (0.008) Ind_R 0.017 − 0.026 − 0.001 (0.014) (0.025) (0.006) Ind_U − 0.057*** − 0.410*** − 0.068*** (0.015) (0.028) (0.006) Occ_Ind_R 0.051*** 0.180*** 0.050*** (0.018) (0.033) (0.008) Occ_Ind_U − 0.090*** − 0.582*** − 0.088*** (0.019) (0.038) (0.008) Observations 31,769 31,769 31,769 31,769 31,769 31,769 31,769 31,769 31,769 Plants 7579 7579 7579 7579 7579 7579 7579 7579 7579 R-squared 0.798 0.799 0.798 0.914 0.915 0.914 Less than 10 employees FRACT_occu 0.042*** 0.019 0.019*** (0.014) (0.023) (0.007) FRACT_ind 0.047*** − 0.167*** 0.011* (0.013) (0.022) (0.006) Occ_R − 0.001 0.016 0.011* (0.011) (0.019) (0.006) Occ_U − 0.028** − 0.155*** − 0.013** (0.012) (0.018) (0.007) Ind_R 0.036*** − 0.028 0.014** (0.012) (0.021) (0.006) Ind_U 0.006 − 0.192*** − 0.005 (0.011) (0.017) (0.006) Occ_Ind_R 0.040*** 0.085*** 0.025*** (0.013) (0.023) (0.007) Occ_Ind_U − 0.018 − 0.217*** − 0.014** (0.014) (0.019) (0.007) Observations 37,430 37,430 37,430 37,430 37,430 37,430 37,430 37,430 37,430 Plants 7979 7979 7979 7979 7979 7979 7979 7979 7979 R-squared 0.797 0.797 0.797 0.895 0.895 0.895 Robust standard errors in parentheses for columns 1 and 3. For the OP estimations, bootstrapped standard errors are presented with 250 replications. ***p < 0.01, ** p < 0.05, * < 0.1. The constant term is not reported. Control variables and year fixed effects are included in all estimations Results here are not as clear cut, but rather similar to value added show a positive sign but the other estima- the results shown for the smaller plants in Table 6 above. tions (columns 2(a) and 3(a)) instead display a negative While the diversity of occupation continues to show relationship. Moreover, the relatedness measures are a positive relationship, the results for industry diver- mainly statistically insignificant or do not show consist - sity are not stable where the fixed effect estimations for ent results across the estimations when taken separately. 18 Page 14 of 21 O. Kekezi Table 6 Regression results when only firms that are at least 5 years old are included Value added – FE Value added – OP Wages – FE 1(a) 1(b) 1(c) 2(a) 2(b) 2(c) 3(a) 3(b) 3(c) FRACT_occu 0.069*** 0.051** 0.026*** (0.010) (0.020) (0.005) FRACT_ind 0.010 − 0.107*** − 0.022*** (0.009) (0.017) (0.004) Occ_R 0.041*** 0.081*** 0.031*** (0.009) (0.018) (0.005) Occ_U − 0.001 − 0.174*** − 0.010* (0.011) (0.019) (0.005) Ind_R 0.025*** 0.041** 0.002 (0.009) (0.018) (0.004) Ind_U − 0.014 − 0.178*** − 0.031*** (0.009) (0.016) (0.004) Occ_Ind_R 0.042*** 0.191*** 0.031*** (0.011) (0.021) (0.005) Occ_Ind_U − 0.040*** − 0.267*** − 0.039*** (0.012) (0.019) (0.005) Observations 65,424 65,424 65,424 65,424 65,424 65,424 65,424 65,424 65,424 Plants 12,020 12,020 12,020 12,020 12,020 12,020 12,020 12,020 12,020 R-squared 0.780 0.780 0.780 0.899 0.899 0.899 Robust standard errors in parentheses for columns 1 and 3. For the OP estimations, bootstrapped standard errors are presented with 250 replications. *** p < 0.01, ** p < 0.05, * < 0.1. The constant term is not reported. Control variables and year fixed effects are included in all estimations Table 7 Labor productivity in startups Value added – FE Value added – OP Wages – FE 1(a) 1(b) 1(c) 2(a) 2(b) 2(c) 3(a) 3(b) 3(c) FRACT_occu 0.092*** 0.066** 0.025*** (0.016) (0.028) (0.008) FRACT_ind 0.032** − 0.239*** − 0.024*** (0.015) (0.027) (0.007) Occ_R 0.017 0.034 0.012* (0.014) (0.024) (0.007) Occ_U − 0.003 − 0.170*** − 0.009 (0.015) (0.025) (0.008) Ind_R 0.017 − 0.059** − 0.015** (0.015) (0.025) (0.007) Ind_U − 0.020 − 0.294*** − 0.035*** (0.014) (0.024) (0.007) Occ_Ind_R 0.032** 0.117*** 0.013* (0.016) (0.029) (0.008) Occ_Ind_U − 0.031** − 0.294*** − 0.027*** (0.015) (0.025) (0.007) Observations 32,259 32,259 32,259 32,259 32,259 32,259 32,259 32,259 32,259 R-squared 0.770 0.770 0.770 0.903 0.903 0.903 Robust standard errors in parentheses for columns 1 and 3. For the OP estimations, bootstrapped standard errors are presented with 250 replications. *** p < 0.01, ** p < 0.05, * < 0.1. The constant term is not reported. Control variables are included in all estimations Diversity of experience and labor productivity in creative industries Page 15 of 21 18 When looking at the combination of industry and occu- a combination of industry and occupation, rather than pation experience results are in line with what has been when they are separated. This suggests that the specific previously shown in the paper; relatedness is positive for human capital of the individuals is connected to both productivity but unrelatedness of experience harms the industry and occupation. productivity growth of plants. What these results suggest Besides contributing to the literature regarding the is that when it comes to startups, the experience of the micro-mechanisms of knowledge spillovers and produc- workers needs to be diverse, but not too diverse. Since tivity which arise from the previous experience, these the teams in this case are smaller, and the probability of results are also important from a policy perspective. working together is larger, the diversity of teams needs Given the importance of creative industries in regional to be related both for industry and for occupation expe- development, understanding how labor productivity is rience, at least in the first years of the startups. These enhanced in these firms benefit the economy at large. results support the findings of Koster and Andersson Moreover, these results reflect the importance of finding (2018) who argue about the importance of occupational the right person for the right job. Knowledge-intensive skills on top of industry skills for the survival of startups. firms in Sweden are continuously having difficulties to Focusing only on one of these dimensions when examin- find the competence for the job. The results shown here ing the previous work experience is not enough to show results suggest that one potential way to look for the right positive results on productivity. competence is to consider the composition of the experi- ence of the people within the plant. Hiring people with related experience in terms of occupation or occupation 6 Conclusions and industry, would benefit the firm in the form of higher The paper studies how the diversity of work experiences labor productivity (which is mirrored in both value added among employees relates to labor productivity in crea- and wages). Given that most firms hire people from the tive industries in Sweden. The idea is that when chang - region, these results could also be analyzed as suggestion ing jobs, workers bring their expertise and knowledge for creative, knowledge intensive firms to locate in areas with them. While a large literature argues about the posi- where there is a large pool of people with related skills to tive effects of labor mobility, the type of knowledge and one another. skills that are brought into the firm is not largely studied. The study creates possibilities for further research. Some studies show however that what mostly contrib- Given the importance of occupational-specific skills utes to firm performance depends on the type of knowl - showed in the results, it would be interesting to dig edge that is brought in and how that matches the existing deeper into what type of occupations are the ones that knowledge base (Boschma et al. 2009; Timmermans and when combined productivity is enhanced. Previous lit- Boschma 2014; Östbring et al. 2018). Others have shown erature has shown how skills should not overlap for new the importance of knowledge diversity for innovation or knowledge to be created (Uzzi et al. 2013), but the litera- productivity growth in a firm (Parrotta et  al. 2014a , b). ture on occupational combinations is scarce. Moreover, it Yet, to my knowledge, no study has looked at the diver- would be interesting to look at this through an innova- sity of the previous experience of the workers, both in tion perspective. Third, while the purpose of this paper terms of occupations and industries, and how that relates has been to look at diversity and relatedness, it would to labor productivity. be insightful to expand the discussion by looking at skill The results of this paper show that diversity of occu - complementarity and firm productivity. Skill comple - pational experience is positive for labor productiv- mentarity is not captured in the diversity or relatedness ity, but this the diversity of industrial experience shows measures, but it would be a great avenue to expand the either insignificant or negative relationship. When the current analysis. Further, as previously mentioned, the distinction between relatedness and unrelatedness of results should be analyzed with caution, given the lack of experience is made, the results indicate that the positive a suitable instrumental variable or any other exogenous relationship is mostly driven by relatedness, which is in shock, which would have made possible causal results. line with similar existing studies on relatedness and per- Moving into the direction of causality is another avenue formance (Boschma et  al. 2009; Martynovich and Hen- where this work can be extended into. ning 2018; Östbring et al. 2018). This relationship is even stronger when experience relatedness is measured as Appendix See Tables 8, 9, 10, 11, 12, 13, and 14. 18 Page 16 of 21 O. Kekezi Table 8 Industries included in the analysis NACE Description 58 Publishing activities 59 Motion picture, video and television programme production, sound recording and music publishing activities 60 Programming and broadcasting activities 62 Computer programming, consultancy and related activities 71 Architectural and engineering activities; technical testing and analysis 72 Scientific research and development 73 Advertising and market research 74 Other professional, scientific and technical activities 90 Creative, arts and entertainment activities 91 Libraries, archives, museums and other cultural activities 93 Sports activities and amusement and recreation activities Table 9 Characteristics of the creative industries and the plants in the rest of the economy Non-creative industries Creative industries 2007 2016 Growth 2007 2016 Growth Employment 3,779,542 4,128,471 9.2% 361,781 424,406 17.3% Number of plants 419,993 498,471 18.7% 72,528 94,576 30.4% Average Wages 2111 2747 30.1% 2455 3104 26.4% Average Productivity 5495 6874 25.1% 5555 6938 24.9% Average sales 1814 1937 6.8% 1169 1288 10.2% Table 10 The 2-digit educational types Group Education type 1 General education 14 Pedagogics and teaching 21 Arts and media 22 The humanities 31 Social and behavioral science 32 Journalism and information 34 Business 38 Law and legal science 42 Biology and environmental science 44 Physics, chemistry, and geoscience 46 Mathematics and natural science 48 Computer science 52 Engineering: technical, mechanical, chemical, and electronics 54 Engineering: manufacturing 58 Engineering: construction 62 Agriculture 64 Animal healthcare 72 Healthcare 76 Social work 81 Personal services 84 Transport services 85 Environmental care 86 Security Diversity of experience and labor productivity in creative industries Page 17 of 21 18 Table 11 Correlation matrix 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 (1) Productivity 1.000 (2) Wages 0.605 1.000 (3) Capital 0.181 0.098 1.000 (4) Labor 0.129 0.224 0.444 1.000 (5) FRACT_occu − 0.022 − 0.038 0.216 0.347 1.000 (6) FRACT_ind − 0.055 − 0.049 0.102 0.309 0.468 1.000 (7) Occ_R 0.102 0.155 0.061 0.092 0.465 0.142 1.000 (8) Occ_U − 0.198 − 0.308 0.014 − 0.060 0.443 0.217 − 0.498 1.000 (9) Ind_R 0.093 0.199 − 0.011 0.148 0.133 0.514 0.174 − 0.087 1.000 (10) Ind_U − 0.189 − 0.301 0.020 − 0.021 0.296 0.589 − 0.017 0.346 − 0.340 1.000 (11) Occ_Ind_R 0.112 0.216 0.013 0.130 0.251 0.385 0.524 − 0.291 0.718 − 0.234 1.000 (12) Occ_Ind_U − 0.167 − 0.261 0.003 − 0.034 0.309 0.383 − 0.327 0.669 − 0.245 0.680 − 0.236 1.000 (13) FRACT_Edu − 0.062 − 0.112 0.172 0.243 0.313 0.186 0.060 0.156 0.001 0.126 0.002 0.104 1.000 (14) Edu 0.191 0.353 − 0.019 0.184 − 0.006 0.083 0.009 − 0.081 0.170 − 0.102 0.116 − 0.044 − 0.188 1.000 (15) Age 0.027 0.021 0.225 0.241 0.021 − 0.177 − 0.054 − 0.033 − 0.175 − 0.119 − 0.130 − 0.077 0.072 − 0.061 1.000 (16) Multiplant 0.154 0.139 0.263 0.297 − 0.019 0.020 0.026 − 0.153 0.026 − 0.069 0.041 − 0.095 − 0.017 0.068 0.122 1.000 (17) Change_LA − 0.026 − 0.010 − 0.023 0.031 0.078 0.260 0.027 0.051 0.137 0.172 0.099 0.117 − 0.041 0.114 − 0.228 0.163 1.000 (18) Den 0.141 0.244 − 0.049 0.129 0.091 0.111 0.090 − 0.037 0.161 − 0.054 0.136 − 0.049 0.087 0.224 − 0.095 − 0.164 − 0.165 1.000 18 Page 18 of 21 O. Kekezi Table 12 Akaike information criteria for the different estimations Average value added Average wages K,L 50,799 − 93,790 Control variables 49,958 − 96,290 Full model 1(a) 49,796 Full model 1(b) 49,809 Full model 1(c) 49,824 Full model 3(a) − 96,408 Full model 3(b) − 96,733 Full model 3(c) − 96,564 In the first columns the AIC is calculated only when including capital and labor in the estimations. In the second columns, all control variables are included besides the variables of interest. In the last three columns, the full models are estimated. 1(a)–1(c) and 3(a)–3(c) correspond to the estimations in Table 3 Table 13 Regression results when the sample ends in 2014 to ensure robustness from changes in SSYK codes Average value added – FE Average value added – OP average wages – FE FRACT_occu 0.073*** 0.061*** 0.032*** (0.011) (0.020) (0.006) FRACT_ind 0.013 − 0.177*** − 0.024*** (0.009) (0.017) (0.004) Occ_R 0.030*** 0.052*** 0.031*** (0.010) (0.018) (0.005) Occ_U − 0.006 − 0.175*** − 0.002 (0.011) (0.017) (0.005) Ind_R 0.023** − 0.004 − 0.002 (0.009) (0.017) (0.004) Ind_U − 0.016* − 0.240*** − 0.035*** (0.009) (0.015) (0.004) Occ_Ind_R 0.041*** 0.135*** 0.028*** (0.011) (0.020) (0.005) Occ_Ind_Un − 0.023** − 0.302*** − 0.035*** (0.011) (0.016) (0.005) Observations 66,748 66,748 66,748 66,748 66,748 66,748 66,748 66,748 66,748 Plants 14,786 14,786 14,786 14,786 14,786 14,786 14,786 14,786 14,786 R− squared 0.793 0.793 0.793 0.900 0.900 0.900 Robust standard errors in parentheses for columns 1 and 3. For the OP estimations, bootstrapped standard errors are presented with 250 replications. ***p < 0.01, **p < 0.05, * < 0.1. The constant term is not reported. Control variables and year fixed effects are included in all estimations Diversity of experience and labor productivity in creative industries Page 19 of 21 18 Table 14 Labor productivity in firms that have not experienced any change in the workforce Value added – OLS Value added – OP Wages – OLS 1(a) 1(b) 1(c) 2(a) 2(b) 2(c) 3(a) 3(b) 3(c) FRACT_occu − 0.160** − 0.189* − 0.129** (0.063) (0.107) (0.054) FRACT_ind − 0.198* − 0.153 − 0.147* (0.104) (0.136) (0.079) Occ_R − 0.033 − 0.069 0.031 (0.055) (0.093) (0.039) Occ_U − 0.268*** − 0.267*** − 0.287*** (0.056) (0.091) (0.058) Ind_R 0.038 0.100 0.077 (0.088) (0.126) (0.067) Ind_U − 0.271*** − 0.266** − 0.219*** (0.093) (0.121) (0.078) Occ_Ind_R 0.246** 0.358** 0.253*** (0.121) (0.166) (0.084) Occ_Ind_U − 0.271*** − 0.288** − 0.315*** (0.098) (0.134) (0.117) Observations 1,175 1,175 1,175 1,175 1,175 1,175 1,175 1,175 1,175 R-squared 0.275 0.294 0.274 0.349 0.385 0.356 Robust standard errors in parentheses for columns 1 and 3. For the OP estimations, bootstrapped standard errors are presented with 250 replications. *** p < 0.01, ** p < 0.05, * < 0.1. The constant term is not reported. Since the diversity measures are constant, the results in columns 1 and 3 do not include firm fixed effects, but instead industry, region, and year fixed effects. Control variables are included in all estimations Acknowledgements Received: 25 August 2020 Accepted: 22 June 2021 I want to thank the three anonymous referees, Martin Henning, Ron Boschma, Rikard Eriksson, Charlotta Mellander, Johan Klaesson, Sandy Dall’erba, Geoffrey Hewings, Jonna Rickardsson, and Emma Lappi for helpful comments and sug- gestions during different stages of this paper. References Authors’ contributions Alesina, A., Devleeschauwer, A., Easterly, W., Kurlat, S., Wacziarg, R.: Fractionali- I am the sole author of the paper, thus responsible for the whole manuscript. zation. J. Econ. Growth 8(2), 155–194 (2003) Almeida, P., Kogut, B.: Localization of knowledge and the mobility of engineers Funding in regional networks. Manag. Sci. 45(7), 905–917 (1999) Not applicable. And, H.W.A., Isaksen, A.: Knowledge intensive business services and urban industrial development. Serv. Ind. J. 27(3), 321–338 (2007) Availability of data and materials Arcidiacono, P., Kinsler, J., Price, J.: Productivity spillovers in team production: The micro-data used in this paper are protected by the Swedish Secrecy Act evidence from professional basketball. J. Law Econ. 35(1), 191–225 (2017) and cannot be freely accessed or shared. The data can however be ordered Autor, D.H., Levy, F., Murnane, R.J.: The skill content of recent technological through Statistics Sweden, through the following link: https:// scb. se/ en/ servi change: an empirical exploration. Q. J. Econ. 118(4), 1279–1333 (2003) ces/ guida nce- for- resea rchers- and- unive rsiti es/. Bacolod, M., Blum, B.S., Strange, W.C.: Skills in the city. J. Urban Econ. 65(2), 136–153 (2009) Bantel, K.A., Jackson, S.E.: Top management and innovations in banking: Does Declarations the composition of the top team make a difference? Strateg. Manag. J. 10(S1), 107–124 (1989) Ethics approval and consent to participate Barbour, E., Markusen, A.: Regional occupational and industrial structure: does Not applicable. one imply the other? Int. Reg. Sci. Rev. 30(1), 72–90 (2007) Bassett-Jones, N.: The paradox of diversity management, creativity and innova- Consent for publication tion. Creat. Innov. Manag. 14(2), 169–175 (2005) Not applicable. Becker, G.S.: Investment in human capital: a theoretical analysis. J. Polit. Econ. 70(5, Part 2), 9–49 (1962) Competing interests Becker, G.S.: Human Capital: A Theoretical Analysis with Special Reference to The author has no competing interests to disclose. Education. Columbia University Press, New York (1964) Bercovitz, J., Feldman, M.: The mechanisms of collaboration in inventive teams: Author details composition, social networks, and geography. Res. Policy 40(1), 81–93 1 2 SOFI, Stockholm University, Stockholm, Sweden. Centre for Entrepreneur- (2011) ship and Spatial Economics (CEnSE), Jönköping International Business School, Berliant, M., Fujita, M.: The dynamics of knowledge diversity and economic Jönköping, Sweden. growth. South. Econ. J. 77(4), 856–884 (2011) 18 Page 20 of 21 O. Kekezi Boschma, R.: Proximity and innovation: a critical assessment. Reg. Stud. 39(1), Manning, A., Petrongolo, B.: How local are labor markets? Evidence from a 61–74 (2005) spatial job search model. Am. Econ. Rev. 107(10), 2877–2907 (2017) Boschma, R., Eriksson, R., Lindgren, U.: How does labour mobility affect the Maroto-Sánchez, A.: Productivity in the services sector: conventional and performance of plants? The importance of relatedness and geographical current explanations. Serv. Ind. J. 32(5), 719–746 (2012) proximity. J. Econ. Geogr. 9(2), 169–190 (2009) Martynovich, M., Henning, M.: Labour force building in a rapidly expanding Card, D., Heining, J., Kline, P.: Workplace heterogeneity and the rise of West Ger- sector. Ind. Innov. 25(2), 199–227 (2018) man wage inequality. Q. J. Econ. 128(3), 967–1015 (2013) Mas, A., Moretti, E.: Peers at work. Am. Econ. Rev. 99(1), 112–145 (2009) Castañer, X., Campos, L.: The determinants of artistic innovation: bringing in Mellander, C.: Creative and knowledge industries: an occupational distribu- the role of organizations. J. Cult. Econ. 26(1), 29–52 (2002) tion approach. Econ. Dev. q. 23(4), 294–305 (2009) Caves, R.E.: Creative Industries: Contracts Between Art and Commerce. Harvard Miguel-Molina, B., Hervas-Oliver, J.-L., Boix, R., De-Miguel-Molina, M.: The University Press, Cambridge (2000) importance of creative industry agglomerations in explaining the Christopherson, S.: The divergent worlds of new media: how policy shapes wealth of European regions. Eur. Plan. Stud. 20(8), 1263–1280 (2012) work in the creative economy. Rev. Policy Res. 21(4), 543–558 (2004) Mincer, J.: Schooling, Experience, and Earnings. National Bureau of Econom- Cohen, W.M., Levinthal, D.A.: Absorptive capacity: a new perspective on ics Research, New York (1974) learning and innovation. Adm. Sci. q. 35(1), 128–152 (1990) Müller, K., Rammer, C., Trüby, J.: The role of creative industries in industrial Delgado-Verde, M., Castro, G.-d, Amores-Salvadó, J.: Intellectual capital and innovation. Innovation 11(2), 148–168 (2009) radical innovation: Exploring the quadratic effects in technology-based Mulligan, C.B., Sala-I-Martin, X.: Measuring aggregate human capital. J. Econ. manufacturing firms. Technovation 54, 35–47 (2016) Growth 5(3), 215–252 (2000) Essletzbichler, J.: Relatedness, Industrial Branching and Technological Cohe- Neal, D.: Industry-specific human capital: evidence from displaced workers. sion in US Metropolitan Areas. Reg. Stud. 49(5), 752–766 (2015) J. Law Econ. 13(4), 653–677 (1995) Florida, R.: The rise of the creative class: and how it’s transforming work, Neffke, F.: Coworker complementarity. SSRN J. (2017). https:// doi. org/ 10. leisure, and everyday life. Basic Books, New York (2002)2139/ ssrn. 29293 39 Florida, R., Mellander, C., Stolarick, K.: Inside the black box of regional Neffke, F., Henning, M.: Skill relatedness and firm diversification. Strateg. development—human capital, the creative class and tolerance. J. Econ. Manag. J. 34(3), 297–316 (2013) Geogr. 8(5), 615–649 (2008) Nooteboom, B.: Learning and Innovation in Organizations and Economies. Frederiksen, L., Sedita, S.R.: Embodied knowledge transfer for innovation: Oxford University Press, Oxford (2000) comparing interfirm labor mobility between music and manufacturing Nooteboom, B., Van Haverbeke, W., Duysters, G., Gilsing, V., van den Oord, industries. In: Belussi, F., Staber, U. (eds.) Managing Networks of Creativ- Ad.: Optimal cognitive distance and absorptive capacity. Res. Policy ity. Routledge, Abingdon (2011) 36(7), 1016–1034 (2007) Frenken, K., Van Oort, F., Verburg, T.: Related variety, unrelated variety and Olley, G.S., Pakes, A.: The dynamics of productivity in the telecommunica- regional economic growth. Reg. Stud. 41(5), 685–697 (2007) tions equipment industry. Econometrica 64(6), 1263–1297 (1996) Gathmann, C., Schönberg, U.: How general is human capital? A task-based Östbring, L., Lindgren, U.: Labor mobility and plant performance: on the approach. J. Law Econ. 28(1), 1–49 (2010) (dis)similarity between labor- and capital-intensive sectors for knowl- Glaeser, E.L., Mare, D.C.: Cities and skills. J. Law Econ. 19(2), 316–342 (2001) edge diffusion and productivity. Geogr. Ann. Ser. B Hum. Geogr. 95(4), Goos, M., Manning, A.: Lousy and lovely jobs: the rising polarization of work 287–305 (2013) in Britain. Rev. Econ. Stat. 89(1), 118–133 (2007) Östbring, L., Eriksson, R., Lindgren, U.: Labour mobility and organisational Horwitz, S.K., Horwitz, I.B.: The effects of team diversity on team outcomes: a proximity: routines as supporting mechanisms for variety, skill integra- meta-analytic review of team demography. J. Manag. 33(6), 987–1015 tion and productivity. Ind. Innov. 24(8), 775–794 (2017) (2007) Östbring, L., Eriksson, R., Lindgren, U.: Relatedness through experience: on Hotho, S., Champion, K.: Small businesses in the new creative industries: the importance of collected worker experiences for plant performance. innovation as a people management challenge. Manag. Decis. 49(1), Pap. Reg. Sci. 97(3), 501–518 (2018) 29–54 (2011) Østergaard, C.R., Timmermans, B., Kristinsson, K.: Does a different view create Howkins, J.: The Creative Economy: How People Make Money from Ideas. something new? The effect of employee diversity on innovation. Res. Penguin, UK (2002) Policy 40(3), 500–509 (2011) Ingram, B.F., Neumann, G.R.: The returns to skill. Labour Econ. 13(1), 35–59 Parent, D.: Industry-specific capital and the wage profile: evidence from the (2006) national longitudinal survey of youth and the panel study of income Jara-Figueroa, C., Jun, B., Glaeser, E.L., Hidalgo, C.A.: The role of industry- dynamics. J. Law Econ. 18(2), 306–323 (2000) specific, occupation-specific, and location-specific knowledge in Parrotta, P., Pozzoli, D., Pytlikova, M.: Labor diversity and firm productivity. the growth and survival of new firms. Proc. Natl. Acad. Sci. 115(50), Eur. Econ. Rev. 66, 144–179 (2014a) 12646–12653 (2018) Parrotta, P., Pozzoli, D., Pytlikova, M.: The nexus between labor diversity and Jarvis, H., Pratt, A.C.: Bringing it all back home: The extensification and ‘over - firm’s innovation. J. Popul. Econ. 27(2), 303–364 (2014b) flowing’ of work. Geoforum 37(3), 331–339 (2006) Penrose, E.: The Theory of the Growth of the Firm. Oxford University Press, Jehn, K.A., Northcraft, G.B., Neale, M.A.: Why differences make a difference: a Oxford (1959) field study of diversity, conflict and performance in workgroups. Adm. Perretti, F., Negro, G.: Mixing genres and matching people: a study in Sci. q. 44(4), 741–763 (1999) innovation and team composition in Hollywood. J. Organ. Behav. 28(5), Kambourov, G., Manovskii, I.: Occupational specificity of human capital. Int. 563–586 (2007) Econ. Rev. 50(1), 63–115 (2009) Pitcher, P., Smith, A.D.: Top management team heterogeneity: personality, Koster, S., Andersson, M.: When is your experience valuable? Occupation- power, and proxies. Organ. Sci. 12(1), 1–18 (2001) industry transitions and self-employment success. J. Evol. Econ. 28(2), Power, D.: Behind the music. Profiting from sound: a systems approach to the 265–286 (2018). https:// doi. org/ 10. 1007/ s00191- 017- 0528-2 dynamics of the nordic music industry. Oslo: STEP/Nordic Innovation Kremer, M.: The O-ring theory of economic development. Q. J. Econ. 108(3), Centre (2003) 551–575 (1993). https:// doi. org/ 10. 2307/ 21184 00 Protogerou, A., Kontolaimou, A., Caloghirou, Y.: Innovation in the European Larsen, J.N.: Knowledge, human resources and social practice: the creative industries: a firm-level empirical approach. Ind. Innov. 24(6), knowledge-intensive business service firm as a distributed knowledge 587–612 (2017) system. Serv. Ind. J. 21(1), 81–102 (2001) Savino, T., Petruzzelli, A.M., Albino, V.: Teams and lead creators in cultural and Lazear, E.P.: Globalisation and the market for team-mates. Econ. J. 109(454), creative industries: evidence from the Italian haute cuisine. J. Knowl. 15–40 (1999) Manag. 21(3), 607–622 (2017) Madsen, T.L., Mosakowski, E., Zaheer, S.: Knowledge retention and personnel Schumpeter, J.A.: The Theory of Economic Development: An Inquiry Into mobility: the nondisruptive effects of inflows of experience. Organ. Sci. Profits, Capital, Credit, Interest, and the Business Cycle, vol. 55. Transaction 14(2), 173–191 (2003) Publishers, New Jersey (1934) Diversity of experience and labor productivity in creative industries Page 21 of 21 18 Scott, A.J.: Human capital resources and requirements across the metropolitan UNESCO.: Creative economy report 2013: special edition: widening local devel- hierarchy of the USA. J. Econ. Geogr. 9(2), 207–226 (2008) opment pathways. UNCTAD New York: United Nations (2013) Serafinelli, M.: “Good” firms, worker flows, and local productivity. J. Law Econ. Uzzi, B., Spiro, J.: Collaboration and creativity: the small world problem. Am. J. 37(3), 747–792 (2019) Sociol. 111(2), 447–504 (2005) Siepel, J., Cowling, M., Coad, A.: Non-founder human capital and the long-run Uzzi, B., Mukherjee, S., Stringer, M., Jones, B.: Atypical combinations and scien- growth and survival of high-tech ventures. Technovation 59, 34–43 tific impact. Science 342(6157), 468–472 (2013) (2017) Van Ark, B.: Measuring the new economy: an international comparative per- Solheim, M.C.W., Boschma, R., Herstad, S.J.: Collected worker experiences and spective. Rev. Income Wealth 48(1), 1–14 (2002) the novelty content of innovation. Res. Policy 49(1), 103856 (2020) Visintin, F., Pittino, D.: Founding team composition and early performance of Söllner R.: Human capital diversity and product innovation: a micro-level university—based spin-off companies. Technovation 34(1), 31–43 (2014) analysis. Jena Econ. Res. Papers 2010(027) (2010) Williams, K., O’Reilly, C.: The complexity of diversity: a review of forty years of Stinchcombe, A.L.: Social structure and organizations. Handb. Organ. 7, research. Res. Organ. Behav. 21, 77–140 (1998) 142–193 (1965) Wixe, S.: The impact of spatial externalities: skills, education and plant produc- Sullivan, P.: Empirical evidence on occupation and industry specific human tivity. Reg. Stud. 49(12), 2053–2069 (2015). https:// doi. org/ 10. 1080/ 00343 capital. Labour Econ. 17(3), 567–580 (2010)404. 2014. 891729 Tao, J., Ho, C.-Y., Luo, S., Sheng, Y.: Agglomeration economies in creative indus- Wixe, S., Andersson, M.: Which types of relatedness matter in regional growth? tries. Reg. Sci. Urban Econ. 77, 141–154 (2019) Industry, occupation and education. Reg. Stud. 51(4), 523–536 (2016) Taylor, A., Greve, H.R.: Superman or the fantastic four? Knowledge combina- Young, A.: Consistency without Inference: Instrumental Variables in Practical tion and experience in innovative teams. Acad. Manag. J. 49(4), 723–740 Application. Unpublished manuscript, London School of Economics (2006) (2019) Thompson, W., Thompson, P.: From industries to occupations: rethinking local economic development. Econ. Dev. Comment. 9(3), 12–18 (1985) Publisher’s Note Timmermans, B., Boschma, R.: The effect of intra- and inter-regional labour Springer Nature remains neutral with regard to jurisdictional claims in pub- mobility on plant performance in Denmark: the significance of related lished maps and institutional affiliations. labour inflows. J. Econ. Geogr. 14(2), 289–311 (2014) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal for Labour Market Research Springer Journals

Diversity of experience and labor productivity in creative industries

Journal for Labour Market Research , Volume 55 (1) – Jul 6, 2021

Loading next page...
 
/lp/springer-journals/diversity-of-experience-and-labor-productivity-in-creative-industries-Z2qS9BG3ct

References (80)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2021
ISSN
1614-3485
eISSN
2510-5027
DOI
10.1186/s12651-021-00302-3
Publisher site
See Article on Publisher Site

Abstract

This paper studies how the previous experience among workers relates to the labor productivity of the creative indus- tries in Sweden. Eec ff tive knowledge transfers are dependent on the cognitive distance among employees. Using longitudinal matched employer-employee data, I measure the portfolio of the skills within a workplace through (i) the workers’ previous occupation, and (ii) the industry they have been working in previously. Estimates show that diversity of occupational experience is positive for labor productivity, but the diversity of industry experience is not. When dis- tinguishing between related and unrelated diversity, the relatedness of occupational experience is positive for labor productivity, while unrelated occupational experience instead shows negative relationship with productivity. These results point towards the importance of occupational skills that workers bring with them to a new employment, for labor productivity. Keywords: Diversity, Skill relatedness, Previous experience, Labor mobility, Knowledge spillovers JEL classifications: J24, L25 earnings and productivity (Parent 2000; Gathmann and 1 Introduction Schönberg 2010; Sullivan 2010). As people move across Research has often focused on the importance of differ - jobs they bring some knowledge which was specific to ent forms of human capital and firm performance (Del - what they were previously doing to the new employment gado-Verde et  al. 2016; Siepel et  al. 2017). However, the (Almeida and Kogut 1999). From a theoretical stand- productivity of workers within a firm also depends on point, the diversity of the workforce could foster creativ- who they work with (Mas and Moretti 2009; Card et  al. ity and innovation, where new knowledge is created from 2013; Arcidiacono et al. 2017; Neffke 2017). The question the recombination of differentiated skills (Schumpeter that then arises is how the composition of skills relates to 1934; Penrose 1959). However, if skills are too different, firm performance. The purpose of this paper is to exam - misunderstandings and conflicts can arise, which would ine how the diversity of skills which come from previous lead to negative effects on performance. experience within a plant matters for labor productivity. Moreover, for knowledge spillovers and learning to I specifically focus on the diversity of skills which arises happen, workers in a firm, need to have some sort of cog - from previous work experience and labor productivity in nitive proximity among each other (Nooteboom 2000). terms of (i) their previous occupation, and (ii) the indus- Along these lines, I further define diversity by distin - try they have been working in. Since the work of Becker guishing between the relatedness and unrelatedness of (1962), researchers have argued about the importance of experience. While previous literature in these lines meas- industry-specific and occupational-specific human capi - ures the relatedness of skills through either educational tal that people accumulate during their working life on background (Boschma et  al. 2009), previous industry experience (Timmermans and Boschma 2014), or previ- *Correspondence: orsa.kekezi@sofi.su.se ous occupational experience (Östbring et  al. 2017), it is SOFI, Stockholm University, Stockholm, Sweden Full list of author information is available at the end of the article Becker (1962) initially discussed firm-specific human capital, but that is not the focus of this paper. © The Author(s) 2021. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. 18 Page 2 of 21 O. Kekezi all of them which make up the skills of individuals. To my performance (Östbring et al. 2018). To sum up, combin- knowledge, the only previous study which considers mul- ing the project-based type of work, with labor-intensity tiple measures of skills is the one of Östbring et al. (2018) in production, as well as a high innovative potential, who use both education and previous industry experi- creative industries become a good case of study for ence. However, occupations are crucial to add as they are issues regarding the diversity of human capital and labor proxies of skills and abilities of the workforce beyond the productivity. It is also important to note that with the educational background (Bacolod et al. 2009). What peo- increasing focus on the knowledge economy, creative ple work with is sometimes argued to be more important industries are an important sector for regional develop- than their educational degree (Florida 2002). Hence, this ment (Florida 2002; UNESCO 2013). Thus, understand - paper contributes to the existing literature by proxying ing how these sectors become more productive and grow the diversity of skills within a workplace through their has implications for the economy at large. previous work experience, since people can bring with To answer the research questions, I use longitudinal them both industry-specific knowledge but also occu - matched employer-employee yearly data from 2007 to pational-specific one. By disentangling the type of skills 2016 for all the firms and individuals employed in crea - and experience brought into the firm, we can get a deeper tive industries in Sweden. I track the current employees understanding of the micro-mechanisms of knowledge 5 years back to see what type of experience they had. The transfer, knowledge spillovers, and labor productivity. diversity of skills is measured through a fractionaliza- This paper also contributes to the literature by apply - tion index. To disentangle whether diversity is related or ing this research question on creative industries, which unrelated, I use the relatedness index proposed by Nef- are the focus for several reasons. First, as knowledge- fke and Henning (2013), which is based on labor flows. intensive industries, they rely heavily on knowledge as Results show that the diversity of occupational experi- an input. Labor creativity is the main factor of produc- ence is positive for labor productivity, but the diversity of tion (Florida 2002), and they are characterized by tal- industry experience is not. Second, the unrelatedness of ented and high-ability individuals and firms which create industry and occupational experience are both negative new knowledge (Larsen 2001; And and Isaksen 2007). for labor productivity. On the other hand, the related- Employment in such industries is also inherently auton- ness of occupational experience within the workplace is omous and more self-expressive than more traditional positive for firm performance. Third, when experience is workplaces (Howkins 2002; Florida 2002). Second, crea- measured as a combination of industry and occupation, tive industries are characterized by a project-based pro- the relatedness of the two is positive and strongly related duction system and the production is dependent on the to productivity. These results point towards the impor - interaction of multiple agents (Caves 2000), who work tance of occupational specific skills for labor productivity in teams which are put to work together for a short time and indicate that the positive relation between the diver- (Jarvis and Pratt 2006). Interactions among employees sity of the workforce and productivity is mostly driven by are a crucial assumption when studying diversity within relatedness. a firm, because for productivity to be affected work - The paper is organized as follows. Section  2 describes ers need to work together or to interact with each other the theoretical framework and previous literature on for learning to happen. Given the high probability that skills, knowledge spillovers and growth. Section  3 pre- workers within firms in creative industries work together sents the data and variables. In Sect. 4 empirical findings to produce something, they become an interesting case and analysis of the results are shown, and in Sect.  5 the of study. Yet, their skill decomposition has not been stability of the results is checked. Section 6 concludes. extensively studied, with a few exceptions (Taylor and Greve 2006). Third, creative industries are widely seen in the literature as being innovative and the within-firm decomposition is an important determinant of innova- tion (Castañer and Campos 2002; Protogerou et al. 2017). Table  9 in the appendix shows the characteristics of plants that belong to creative industries (as defined on the paper) and the plants that do not for Last, by focusing on a similar set of industries, I am also 2007 and 2016 which is the time studied empirically. The data show that crea - able to mitigate issues arising from sectoral heterogene- tive industries have experienced a much larger growth in terms of employ- ity, which has been shown to give different results regard - ment, number of plants, as well as sales. Productivity growth does not differ between the two groupings, but the growth of wages is lower for creative ing the importance of diversity and relatedness on firm industries. The growth rate of the creative industries during this time period shows that they are an important segment of the Swedish economy, which is growing fast, and it employed about 9 percent of the workforce in 2016.More- over, they also indirectly support the economy by for example facilitating and Table  8 in the Appendix presents the list of industries included, adapted supporting innovation for other sectors in the economy (Müller et al. 2009). from Miguel-Molina et al. (2012). Diversity of experience and labor productivity in creative industries Page 3 of 21 18 et  al. 2014b, a). Boschma et al. (2009) look deeper at the 2 Diversity, relatedness, and firm performance type of educational diversity within firms and find evi - Research on workforce diversity and firm performance, dence that firms with higher education relatedness show broadly defined, is extensive. Some researchers have used higher productivity growth. Similar results are found in case studies and focused on team diversity (Horwitz and Östbring and Lindgren (2013), and the effect is stronger Horwitz 2007) as well as the composition of the top man- for labor-intensive industries than capital-intensive ones. agement and founding team (Bantel and Jackson 1989; However, proxying skills of the workforce through edu- Pitcher and Smith 2001; Visintin and Pittino 2014). Oth- cation has not come without critique in the literature, ers have used linked employer-employee data to examine since the quality of education is heterogenous, not only the within-firm diversity (Söllner 2010; Østergaard et  al. across countries but also across regions within a country 2011; Parrotta et al. 2014a, b; Solheim et al. 2020). On the (Mulligan and Sala-I-Martin 2000; Ingram and Neumann one side, the diversity of skills contributes to the creation 2006). Moreover, skills and human capital are to a large of new ideas and thus better performance (Bantel and extent collected from the working-life experience, some- Jackson 1989; Lazear 1999; Taylor and Greve 2006; Ber- thing that education does not capture. Becker (1962) dis- liant and Fujita 2011). Firms with more diverse knowl- cussed that human capital can be general which increases edge bases also have higher “absorptive capacity”, i.e. productivity no matter the job people have, but it can accumulated knowledge to understand and use the new, also be specific to the firm people are working. Specific incoming one, which is crucial for innovation and growth human capital can therefore not be transferred across (Cohen and Levinthal 1990). On the other side, for cer- jobs. Extending Becker’s work, literature has discussed tain tasks, Kremer’s O-ring predicts that workers with that human capital is also industry (Neal 1995), or occu- similar skills should work together to see higher produc- pation-specific (Kambourov and Manovskii 2009). u Th s, tivity returns (Kremer 1993). Moreover, people might as people change industries or occupations, there are prefer working with others whom they see as similar. If skills which cannot be transferrable. This indicates that diversity leads to misunderstandings, conflicts, or unco - if workers within a firm have very different skills, work - operativeness across workers, negative effects of diver - ing together would not necessarily be beneficial as they sity are observed (Bassett-Jones 2005; Jehn et  al. 1999; would not understand each other, which goes back to the Madsen et  al. 2003; Williams and O’Reilly 1998). Thus, cognitive proximity argument (Nooteboom 2000). how diversity impacts firm performance is an empirical A complementary measure of human capital often question. used in the literature is through different occupations In a theoretical contribution, Lazear (1999) argues that individuals had (Thompson and Thompson 1985; however that for diversity to have a positive effect on per - Florida 2002; Florida et  al. 2008; Scott 2008). Occupa- formance, the skills of the workforce should be disjoint tions measure the practical skills of people, beyond their but still relevant to one another. Moreover, they should formal education (Bacolod et al. 2009; Wixe and Anders- be learnt by the other groups at a not too high cost. Thus, son 2016). The diversity of occupations within a firm has for learning to happen, some level of cognitive proxim- not been extensively studied, but the existing literature ity or complementarity is required (Nooteboom 2000). If suggest a positive effect on innovation (Söllner 2010; Par - the knowledge bases of the firm are too different, people rotta et  al. 2014b). Östbring et  al. (2017) further suggest do not understand each other. Yet, too much cognitive that the positive effect of occupational diversity on pro - proximity might create a lock-in problem that disables ductivity is driven by relatedness because the unrelated- the capability of companies to adopt new technologies or ness of occupations in a firm either displays insignificant market possibilities (Boschma 2005). Nooteboom et  al. or negative effect. Besides education and occupation, (2007) find for instance an inverted U-shaped impact of human capital can also come from industry experience the cognitive distance and innovation of firms, indicating (Neal 1995). Östbring et al. (2018) have studied how the that knowledge shouldn’t be too similar or too different relatedness of industry experience in knowledge-inten- for innovation to happen. To take the cognitive distance sive business services impacts firm performance. Their into account, the notion of relatedness has emerged in results show that for single-plant firms, the variety of the literature, where several studies, stemming from the knowledge and previous industrial experience affect firm work of Frenken et al. (2007), have distinguished between performance positively. related and unrelated diversity (Boschma et al. 2009; Öst- To sum up, the literature has previously investigated bring and Lindgren 2013; Östbring et al. 2017, 2018). the importance of educational diversity, occupational When examining the effect of skill diversity on firm diversity, or diversity of industrial experience on firm performance, most existing studies focus on the diver- performance. Their results point toward a positive impact sity of educational background, where the results often of diversity, but these effects seem to be stronger in the show a positive effect (Østergaard et  al. 2011; Parrotta 18 Page 4 of 21 O. Kekezi case of related diversity. Yet, Timmermans and Boschma particularly challenging in the smaller firms (Christo - (2014) find that it is the unrelatedness which matters for pherson 2004; Hotho and Champion 2011). When it productivity growth of firms in the region of Copenha - comes to the decomposition of the team, Taylor and gen in Denmark. They speculate that it could be because Greve (2006) and Perretti and Negro (2007) find evidence Copenhagen is mostly characterized by service industries that creative industries especially benefit from teams with compared to the rest of Denmark, which might benefit diverse skills. Thus, the literature on firm diversity and mostly from unrelatedness. Therefore, we do not know firm performance discussed at the beginning of Sect.  2, a priori what type of diversity matter most for creative is highly relevant and applicable to the creative indus- industries. tries. Moreover, because the probability of teamwork is Moreover, these studies primarily study the diversity of higher in such industries, the results obtained would give the current occupation individuals have, and not at their a clearer and more accurate picture on the importance of occupational and industrial history. From a theoretical diversity for knowledge spillovers and productivity. perspective, the knowledge of workers is also shaped by Moreover, creative industries are characterized by their previous experiences and job tasks. When people high labor mobility (Florida 2002; Frederiksen and Sed- change jobs, the skills that they have accumulated are ita 2011). Florida (2002) also identifies creative work - not necessarily left behind but instead brought into their ers as mobile in their career choices, since they have the new workplace (Almeida and Kogut 1999). While labor skills and education to change jobs or careers. This can mobility has been extensively studied, we do not know be directly connected to the structure of such industries enough on the type of knowledge and skills are brought which are characterized by a lot of small firms with high into the firm and how that affects performance (Boschma entry and exit rates (Power 2003). Thus, the probability et al. 2009; Timmermans and Boschma 2014). Therefore, that workers have previous experience from other indus- the skills that people bring can come from their previ- tries and occupations is higher. Furthermore, they are ous industry experience, from previous occupations they labor intensive and they usually employ high-skilled indi- might have had, or from a combination of the two. Sul- viduals who create new knowledge (Larsen 2001; Wiig livan (2010) argues that human capital is both connected Aslesen and Isaksen 2007). to the industry and occupation. Similarly, the literature What is also important to note is that skills obtained on job polarization treats a “job” as an occupation-indus- from occupational experience are especially important try interaction (Autor et  al. 2003; Goos and Manning for people working in creative industries. By definition, 2007). The reason for using a combination of the two is creative industries, are characterized by a high concen- that there are industry effects on wages, after controlling tration of creative workers. Florida’s (2002) creative class for the occupation. is based on the occupations people have and what they do in their everyday tasks, rather than the industries 2.1 W hy employee diversity in creative industries? where they are employed. Moreover, the occupational The literature covered so far does not specifically focus distribution across industries can be heterogenous. For on creative industries, raising the questions on how it instance, a high-tech firm employs accountants, engi - relates to them, as well as what can we learn from study- neers, manufacturing jobs, as well as service jobs at the ing the diversity of skills in such sectors. Creative indus- food court (Mellander 2009). Along these lines, Barbour tries are a good case of study for this research question and Markusen (2007) discuss that the occupational struc- for several reasons. ture of high-tech industries in California is different from Researchers have increasingly argued that workers in the rest of the US. Thus, these results hint towards the creative industries are likely to collaborate and work in idea that industry-specific skills might not be equally teams (Caves 2000; Jarvis and Pratt 2006; Uzzi and Spiro important for creative industries. 2005; Savino et  al. 2017). Moreover, due to the project- based character of these industries, the workforce if 3 Data, variables, and method constantly required to readjust and form new teams To examine the relatedness of the previously acquired since projects are often short-term, which can become skills among workers on labor productivity in the crea- tive industries, I use register longitudinal matched employer-employee yearly data, collected from Sta- tistics Sweden, during 2007–2016. To allow plants to In a related strand of literature, studies have indeed looked at the impor- reach some skill diversity, similar studies drop plants tance of industry or occupational experience (not combined) in a firm, for with less than 10 employees (Parrotta et  al. 2014a, b). wages, firm survival as well as productivity (Timmermans and Boschma 2014, Martynovich and Henning 2018, Jara-Figueroa et al. 2018). However, the focus However, creative industries in are characterized by of these studies is on relatedness to the current job rather than relatedness small firms which is clearly shown in Fig.  1 below. To across workers within the workplace. Diversity of experience and labor productivity in creative industries Page 5 of 21 18 Fig. 1 Distribution of plant size Table 1 The most common occupations and industries where workers employed in creative industries come from Occupation Industry Computing professionals Computer programming activities Physical and engineering science technicians Computer consultancy activities Architects, engineers and related professionals Construction and civil engineering activities and related technical consultancy Writers and creative or performing artists Advertising agency activities Finance and sales associate professionals Industrial engineering activities and related technical consultancy Managers of small enterprises Business and other management consultancy activities Shop, stall and market salespersons and demonstrators Engineering activities and related technical consultancy in energy, environ- ment, plumbing, heat and air-conditioning Business professionals Architectural activities Computer associate professionals Other software publishing Artistic, entertainment and sports associate professionals Technical testing and analysis make the visualization clearer, all plant with more than 3.1 Variables and method 50 employees are put together in the last bar.3.1.1 Dependent variable The figure shows the distribution of firm size, where The outcome variable is labor productivity, measured about 62 percent of the firms only have one employee as value-added per employee, in its logged form. Previ- and an additional 12 percent have only two employees. ous research usually measures the effect of relatedness To not exclude too many of the firms in the creative of skills on productivity growth with a time lag of more industries, and to be able to give a representative pic- than one year due to the time it may take for the knowl- ture of the creative industries, I keep firms that have at edge spillovers to influence growth. However, due to the least 3 employees, where at least some level of diversity project-based characteristics of some creative industries, is reached. the short-term effects of such spillovers are of interest. I track the current employees five years back in time u Th s, a one-year lag is implemented. Following Tim - to see what type of experience they had. If they have mermans and Boschma (2014), for multi-plant firms, the changed industries or occupation several times, the most value-added across plants is distributed according to the recent is considered. If they have been working in the distribution of wages. same industry and occupation in the past 5 years, the cur- While measuring productivity through value-added rent job is considered. During the time of the study, the is common, using value added for creative industries experience of the workers in creative industries comes might be cumbersome (Maroto-Sánchez 2012). In broad from 113 different occupations and approximately 700 terms, productivity refers at the ability of a firm to gen - industries. Table  1 shows the 10 most common occupa- erate outputs from a set of inputs. Service sectors, in tions and industries that the workers currently employed general, do not have the same inputs or outputs as the in creative industries have experience on. traditional manufacturing firms, creating so difficulties 18 Page 6 of 21 O. Kekezi in measuring labor productivity (Van Ark 2002). There - (2013), industry changes of individuals who earn less fore, besides value added, results are also estimated using than the industry median wage as well as managers are wages. Assuming that wages also reflect labor productiv - excluded since these are individuals who are not very ity (Becker 1964; Mincer 1974), a more efficient flow of likely to have industry-specific skills. The intuition is knowledge across employees would indicate higher pro- that we want to capture industries that require similar ductivity and thus higher earnings. skill sets. Inter-industrial moves of individuals who do not have industry specific skills, would not give us that 3.1.2 Measuring diversity and relatedness of skills information. I then run a zero-inflated negative binomial In the first step, following Parrotta et  al. (2014b), the regression with pairwise industrial flows as the depend - diversity of skills is measured through a fractionalization ent variable. The independent variables are the employ - index (Alesina et al. 2003) which is computed at the plant ment size, average wage, as well as wage growth in the level as one minus the Herfindahl index: origin and destination industries. Using the point esti- mates obtained, the predicted labor flows are calculated for each industry pair. The SR measure is: Fract = 1 − p wt (1) wst s=1 obs ij SR = (2) ij where w denotes the workplace, s is the variable for which ij the diversity is computed, and t is time. p is square of the obs share of workers within each category s, each year. The where F and F are the observed and predicted flows ij ij index takes the minimum value of zero if there is only respectively. A value of larger than 1 indicates that the one category present in the workplace and its maximum observed flows are larger than predicted, making the value occurs when all categories are distributed equally: industries related. A ratio of lower than 1 shows that the (1 − ) . The index is measured for the diversity of occu - industries are skill dissimilar. In the last step, arguing that pational and industry experience. the probability for an individual to move from industry Besides diversity itself, following the discussion pre- i to j is the following, it is possible to statistically test sented in the literature review, it is also interesting to whether the observed flows are exceptionally large: look at whether the degree of diversity matters. Frenken ij et  al. (2007) proposed the entropy measures of related pˆ = (3) ij emp and unrelated variety, which have been often used in the literature to measure the degree of diversity within a SR is significant and higher than 1 in 4 percent of all firm (Boschma et  al. 2009; Östbring and Lindgren 2013; industry combinations. The NACE industrial classifica - Östbring et al. 2017, 2018). However, these measures are tion changed in 2007, where the industries were split and dependent on industry or occupational classifications aggregated differently, creating difficulties into translat - which do not fully capture the degree of relatedness or ing the old industrial codes to the new ones. Thus, the cognitive proximity since they are arbitrarily decided skill-relatedness index is constructed in the same way (Essletzbichler 2015). for the new codes for labor mobility during the years Therefore, to define related and unrelated industries 2010–2013. and occupations, I rely on the revealed skill-relatedness 5 However, human capital is also dependent on the type (SR) measure proposed by Neffke and Henning (2013). of job workers have in the firm. Gathmann and Schön - The main assumption behind SR is that individuals are berg (2010) find that people are more likely to switch more likely to switch jobs across industries where their occupations across those jobs where they can use their skills can partly be used. The steps described below fol - skills more. Thus, the skill relatedness matrix is also con - low the original paper and are based on the labor flows structed for the 3-digit occupational codes in the same of the working population in Sweden. First, a matrix with way as explained above. The main difference between pairwise labor flows for all 5-digit industry codes during this calculation and the industrial relatedness one is that 2004–2007 is constructed. Like in Neffke and Henning While the Neffke and Henning (2013) skills relatedness index is well-estab - The empirical estimations are however relatively stable even when managers lished in the literature, the index does not consider the geography of labor and people who earn less than median wage are included in the SR. They are mobility. People are more likely to switch jobs in the areas where they live or available upon request. work (Manning and Petrongolo 2017), thus industrial mobility is partly con- strained to the industries available in the region. This concern does however Since the period studied is 2007–2016, for 2007–2010, the relatedness of not change the findings of the paper, neither the suitability of the skill-related - experience is calculated through the old classification and for 2011—2016 ness measure for the research question. with the new one. Diversity of experience and labor productivity in creative industries Page 7 of 21 18 labor flows are not measured each year, but rather every calculated with a time lag of one year to allow for the second year. The reason is that Statistics Sweden does not knowledge spillovers to take place. Ŵ represents a vector collect data regarding occupations for the full population of the plant specific control variables, and Z represents each year. After two years approximately 80 percent of the vector of the region-specific characteristics, D and D f t the population is covered, which makes the occupational are fixed effects on the firm, and time. switches more reliable. About 13 percent of the combina- One problem that the literature has pinpointed how- tions are statistically related to each other. ever, is that the error term consists of ω which is a it Following the steps above, the industry- and occupa- productivity shock observed by all firms but not by econ - tion-pairs which are skill-related to each other are iden- ometricians, while η is observed by both firms and econ - it tified. To aggregate this to a plant level, I first identify ometricians as shown below: all possible industry (and occupation) combinations of u = ω + η it it it experience between workers within a plant. Then, the number of all combinations which are statistically signifi - For that reason, the estimates observed by linear cant with a SR above one is divided by the total number regressions usually show upward biases in the coeffi - of combinations to calculate the share of relatedness in cient of labor and the coefficients for capital are down - a firm. In the same way, the number of people with the ward bias. u Th s, the methodology developed by Olley same industry experience is divided by the total number and Pakes (1996) (henceforth, OP). The OP estimation of combinations. The rest is the share of skill unrelated - is a semi-parametric method which is calculated on the ness in a firm. Since the three shares add up to one, the identification of a proxy variable which is assumed to similarity of skills is not included in the estimations. be a function of ω productivity shocks. The proxy vari - it able is often investments which firm make, which are 3.1.3 Method assumed to increase productivity. Therefore, they sug - To study how relatedness of skills relates to on average gest the use of a control function approach, which con- labor productivity, a linear regression model with fixed trols for the endogeneity of labor, where investments are effects is used. The panel is not balanced, since firms used to replace the unobserved productivity shock. Fol- can enter and exit during the time studied. As in many lowing Tao et al. (2019), investments are measured as the studies where the dependent is productivity, the starting change of fixed assets. Similar two-step approaches have point is often the Cobb–Douglas production function, often been used in the literature to infer productivity by where productivity of plant i at time t is a function of observing the input choices of the firms (Parrotta et  al. technology (A), capital (K), and labor (L): 2014a; Serafinelli 2019; Tao et al. 2019). Y = AL K (4) it it it3.1.4 Control variables Following the Cobb–Douglas production function, labor However, since I am interested in productivity per and capital are included in the estimations. Besides, the employee, we can divide everything by L, allowing the diversity of educational background is also controlled Cobb–Douglas to take the following form: for in the empirical model. The main reason for doing α so is to ensure that our measures of diversity of work Y AL K it it it α−1 (5) = y = = AL K it are not driven by the diversity of the education tracks. it it L L it it Previous literature has mostly found a positive effect between the diversity of education and labor productivity In order to facilitate the empirical estimation, the (Østergaard et  al. 2011; Parrotta et  al. 2014b, a). Share model is estimated in its logarithmic form where all the of workers with high education, plant age, and whether control variables were captured in the A parameter in the the firm is multi-plant or not are further controlled for previous equations: in the model (Östbring and Lindgren 2013; Wixe 2015). ln y =δ ln L + β ln K + ϕ Div + ϕ ln Ŵ it it it 1 it−1 2 it Since knowledge can also be region-specific, to examine + ϕ ln Z + ϕ D + ϕ D + u 3 rt 4 f 5 t it the importance of skills acquired in a different region (6) (Timmermans and Boschma 2014; Boschma et al. 2009), where δ = α − 1 and since α< 1 by definition, the coef - the share of workers who have worked in another labor ficient of labor in this case is expected to be negative. Div are the plant diversity and relatedness which are it−1 8 9 In 2014 the occupational codes changed, and the new codes were manually The education tracks are presented in respectively Table 10 in the Appendix. matched with the old ones. 18 Page 8 of 21 O. Kekezi Table 2 List of Variables and descriptive statistics Variables Measured as Mean SD Min. Max. Outcome variables Avg_Prod (000) Value added per labor 937.217 7396.918 0.094 809,103.1 Wages (00) Average yearly wage in the plant 3865.126 1366.513 3.667 25,718.73 Diversity and relatedness measures FRACT_occu 1 minus the Herfindahl index of the diversity of occupation experience 0.641 0.206 0 0.959 FRACT_ind 1 minus the Herfindahl index of the diversity of industry experience 0.568 0.245 0 0.976 Occ_R Share with related occupation experience 0.44 0.231 0 1 Occ_U Share with unrelated occupation experience 0.313 0.238 0 1 Ind_R Share with related industry experience 0.318 0.229 0 1 Ind_U Share with unrelated industry experience 0.337 0.255 0 1 Occ_Ind_R Share with related occupation and industry experience 0.161 0.165 0 1 Occ_Ind_U Share with unrelated occupation and industry experience 0.149 0.177 0 1 Control variables K (000) Capital 29,533.66 1,137,329 0 1.80E + 08 L Labor–plant size 20.859 57.582 3 3331 FRACT_Edu 1- the Herfindahl index of the diversity of education tracks 0.578 0.215 0 0.91 Edu Share with at least a 3-year university degree 0.352 0.296 0 1 Change_LA Share who have worked in another labor market 0.221 0.221 0 1 Age Years of operation 12.577 8.17 1 30 Multiplant Dummy = 1 if the firm has more than 1 plants 0.297 0.457 0 1 Den Population per square kilometer 1575.774 1943.169 0.2 5496.4 All independent variables are measured at time t, besides the diversity and relatedness which are measured in t-1, to allow time for the knowledge spillovers to take place. All monetary values are in SEK market is included. Last, I include population den- 4 Empirical findings and analysis sity in the municipality to account for the importance Table  3 presents the baseline results. In columns 1(a)– of agglomeration economies on labor productivity and (c) the linear regression results are presented when the wages (Wixe 2015; Glaeser and Mare 2001). Table  11 in dependent variable is average labor productivity. The the appendix presents the correlation matrix. No large Olley–Pakes estimations are presented in columns 2(a)– values are shown from there, indicating that multicollin- (c). The last columns, 3(a)–(c) present linear regression earity is not a problem in this dataset. Table 2 presents models when average wages are instead used as depend- the list of variables used in the estimations. ent variables. The last columns of Table  2 present the descriptive sta- Starting with the diversity variables, measured through tistics, when the variables are in their non-logged form. the fractionalization indices (columns 1(a), 2(a), 3(a)), The fractionalization indices show that individuals have results show that the diversity of previous occupation rather broad backgrounds. The diversity of occupational experience between employees is positively related to experience is on average higher than the industrial one. average labor productivity, as well as wages, the year On average about one third of the employees have a later. However, the fractionalization index of industrial higher education. The table also shows that there are experience does not display a significant relationship for many small workplaces where the mean size is 21, but the labor productivity, but it shows a negative and signifi - median size is 9. Small workplaces are not uncommon for cant relation in the OP estimation as well as for wages. creative industries as shown in Fig. 1 above. 30 percent of These results indicate that having people with differ - the workplaces belong to multi-plant firms. ent occupational backgrounds work together is posi- tive for productivity while having individuals who come from many different industries is not. The findings about occupations are in line with Parrotta et  al. (2014b) and Sweden has 81 labor market regions which consist of several municipalities. Söllner (2010), but their outcome is innovation and not Multicollinearity is also tested through the VIF value in the regressions productivity. Regarding the industrial experience, one and the VIF values are very low, indicating that multicollinearity is not an can speculate that the results might be driven by the issue. Diversity of experience and labor productivity in creative industries Page 9 of 21 18 Table 3 Baseline results Value added – FE Value added – OP Wages – FE 1(a) 1(b) 1(c) 2(a) 2(b) 2(c) 3(a) 3(b) 3(c) FRACT_occu 0.063*** 0.038** 0.021*** (0.009) (0.016) (0.005) FRACT_ind 0.007 − 0.174*** − 0.025*** (0.008) (0.016) (0.004) Occ_R 0.026*** 0.042*** 0.024*** (0.008) (0.014) (0.004) Occ_U − 0.010 − 0.184*** − 0.016*** (0.009) (0.015) (0.004) Ind_R 0.014* − 0.010 − 0.003 (0.008) (0.014) (0.004) Ind_U − 0.022*** − 0.235*** − 0.036*** (0.008) (0.014) (0.004) Occ_I 0.031*** 0.126*** 0.027*** (0.009) (0.018) (0.005) Occ_Ind_U − 0.035*** − 0.296*** − 0.040*** (0.009) (0.016) (0.005) Capital 0.021*** 0.021*** 0.021*** 0.020*** 0.022*** 0.021*** 0.004*** 0.004*** 0.004*** (0.002) (0.002) (0.002) (0.005) (0.005) (0.005) (0.000) (0.000) (0.000) Labor − 0.077*** − 0.072*** − 0.072*** − 0.060*** − 0.073*** − 0.076*** − 0.022*** − 0.022*** − 0.022*** (0.003) (0.003) (0.003) (0.006) (0.006) (0.006) (0.002) (0.002) (0.002) FRACT_Edu 0.079*** 0.097*** 0.097*** − 0.073*** − 0.021 − 0.057*** 0.026*** 0.032*** 0.029*** (0.011) (0.011) (0.011) (0.018) (0.017) (0.017) (0.006) (0.005) (0.005) Edu 0.044*** 0.050*** 0.049*** 0.191*** 0.181*** 0.188*** 0.045*** 0.047*** 0.045*** (0.011) (0.011) (0.011) (0.014) (0.013) (0.014) (0.006) (0.006) (0.006) Age 0.002*** 0.002*** 0.002*** − 0.005 − 0.004 − 0.004 0.001*** 0.001*** 0.001*** (0.000) (0.000) (0.000) (0.197) (0.191) (0.189) (0.000) (0.000) (0.000) Multiplant 0.036** 0.035** 0.035** 0.169*** 0.146*** 0.159*** 0.005 0.005 0.005 (0.015) (0.015) (0.015) (0.009) (0.009) (0.009) (0.004) (0.004) (0.004) Change_LA − 0.088*** − 0.083*** − 0.084*** − 0.036** − 0.020 − 0.059*** − 0.050*** − 0.048*** − 0.051*** (0.009) (0.009) (0.008) (0.015) (0.015) (0.015) (0.004) (0.004) (0.004) Den 0.028*** 0.027*** 0.027*** 0.051*** 0.045*** 0.047*** 0.022*** 0.022*** 0.022*** (0.001) (0.001) (0.001) (0.005) (0.005) (0.005) (0.001) (0.001) (0.001) Obs. 88,078 88,078 88,078 88,078 88,078 88,078 88,078 88,078 88,078 Plants 15,983 15,983 15,983 15,983 15,983 15,983 15,983 15,983 15,983 R-squared 0.785 0.785 0.785 0.894 0.894 0.894 Year FE ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ Robust standard errors in parentheses for columns 1 and 3. For the OP estimations, bootstrapped standard errors are presented with 250 replications. *** p < 0.01, ** p < 0.05, * < 0.1. Estimates of the constant term is not reported non-transferability of industry human capital. Having Given the differences in the results for diversity, it is individuals with different backgrounds working together interesting to see whether the results differ for the degree can come with difficulties in communicating due to for of diversity (columns 1(b), 2(b), 3(b)). Previous findings example different routines they are used to. However, this are inconsistent because these skills are obtained through can also have to do with the nature of creative industries different mechanisms and measure different types of where the main focus is put on the creativity of individu- knowledge. Results show that the relatedness of occupa- als which is more connected to the occupational tasks tional experience is positively related to labor productiv- rather than the industry specific ones. ity and wages, but the relatedness of industry experience is statistically insignificant (besides in column 1(b) at a 18 Page 10 of 21 O. Kekezi 10% significance level). The importance of related occu - can be that the firms do not always have control over the pational experience is also supported by Östbring et  al. diversity of the workforce (Parrotta et  al. 2014a). This (2017). However, the negative sign regarding industrial would especially be true in Sweden, where firms are relatedness does not support the positive results found increasingly experiencing difficulties to find the right by Östbring et  al. (2018) for KIBS industries. Unrelated- person for the right job. This problem of job matching is ness of both industry and occupational experience are even more pronounced in knowledge intensive sectors. negatively related to firm performance. These results Another reason can be that even is some skill combina- suggest that the positive sign of the fractionalization tions are unrelated, they might be complementary to index on the diversity of occupational experience is most each other, which can also matter for productivity and likely driven by the related occupational diversity rather wages, as shown by Neffke (2017). Moreover, firms might than the unrelated one. The negative sign for the rela - not have full information on the type of skill mixture tion observed with the unrelated experience, for occupa- would create the highest productivity. Given that creative tions and industry experience, is not surprising given the industries are characterized by mostly smaller plants, this importance of cognitive proximity (Nooteboom 2000). If assumption would not be unrealistic. the skills are too different, no knowledge spillovers would However, it can also be the case that the composition be possible. of the workforce within a plant is endogenous. One issue In the last columns for the three specifications, I now that could be problematic when looking at firm produc - estimate the relationship between the relatedness and tivity, is the endogeneity of hiring where the more pro- unrelatedness of previous experience measured through ductive firms would hire the more productive workers. If the combination of occupation and industry experience. the diversity of previous experience is endogenously cho- While, to my knowledge, this has not been previously sen by the firm to enhance productivity, the regressions estimated in the firm relatedness literature, a branch of presented in Table  3 should be analysed with caution. A labor economics argues that the skills of the individuals common way to deal with this issue is through the use of come from the tasks they perform which is connected to instrumental variables (IV). However, after trying a few both industry and occupation. Results, once more, show different IVs, they showed to be weak ones. Given that that the relatedness of experience is positively related to weak instruments are biased and uninformative (Young productivity and wages, but the unrelatedness of experi- 2019), they were not included in the paper. It is impor- ence is negatively related to the firm performance. The tant to point out that even if there would be endogeneity fact that the magnitude of the coefficients is also much in hiring, relying on economic theory and previous litera- higher than in columns 1(b), 2(b), and 3(b), hints towards ture, the direction of the relationship should be going the the idea that a combination of skills obtained from direction tested in the paper. Diversity is important for industry and occupation is specifically important for the creation of new ideas, and the results shown for relat- knowledge spillovers which are then mirrored in higher edness support the existing research they are based upon. productivity or higher wages. Unrelatedness of experi- Moving on to the control variables, the diversity of ence, in this case, is negatively related to the outcomes. educational experience is positive and significant in the The results hint towards the idea that the diversity of linear specifications, supporting the existing research occupational experience is important for labor produc- (Østergaard et  al. 2011; Parrotta et  al. 2014b, a). How- tivity, but the diversity of industrial experience shows no ever, the OP estimation shows a negative and significant significance. When separating between the type of diver - relationship between education diversity and productiv- sity, the positive results seem to be solely driven from the ity. Thus, the interpretation of that results should be with relatedness of previous occupational experience. These care. The rest of the control variables take the expected results are in general in line with the existing literature on signs. Note that the negative sign taken by the plant knowledge flows and relatedness (Boschma et  al. 2009; size comes because the dependent variables are divided Östbring and Lindgren 2013; Östbring et al. 2017, 2018). with labor (as shown in Eqs. 4–6 above). Labor elasticity Unrelatedness of industry and occupation experience is about 0.93 in all estimations, which is relatively large shows an either insignificant or negative relationship to compared to the norm in the production function litera- productivity and wages. ture. This can however be driven by the that that creative The question that could arise is why firms would build a workforce with unrelated work experience. One reason Table 11 in the appendix shows the Akaike Information Criterion (AIC) for different model specifications to ensure that the inclusion of the diversity and Footnote 12 (continued) relatedness variables improve the model fit. As shown in the table, the inclu - the model improves. This is further suggestive evidence regarding the impor - sion of the variables decreases the AIC value, indicating indeed that the fit of tance of workforce diversity for labor productivity. Diversity of experience and labor productivity in creative industries Page 11 of 21 18 industries are heavily dependent on labor as an input and to the baseline model. Diversity of occupation is how- much less on capital (with an elasticity of about 0.02). ever only significant in the first estimation (column Capital is positive and significant to firm performance. 1(a)). Occupational relatedness is positive for productiv- The share of highly educated and workplace age are ity while unrelatedness is negative. On the other hand, positively related to labor productivity and wages. Multi- industry relatedness shows a positive relation to value plant firms and older firms which show higher labor pro - added in the OP estimation but a negative in the wages ductivity. However, hiring individuals who have worked model. As in Table  3, unrelatedness of industry experi- in other regions negatively relates to productivity. This ences is negatively related to productivity. The last col - can have to do with knowledge being rather localized umns in the three estimations (1(c), 2(c), 3(c)) confirm the and people working in different regions follow different relationship found above. Interestingly, the share of new routines and other ways of solving problems. It can also employees is negative for both productivity and wages. be the case that there is a need for an adjustment period This means that for the positive spillovers to emerge from which is not considered since this variable is measured the churning of employees, more time is needed. Other at time t. Regarding the regional variables, as expected, research also suggests that teams become more produc- the workplaces located in denser regions, are also the one tive the longer they work together (Bercovitz and Feld- showing higher productivity (Wixe 2015). man 2011). These results have broader implications for policy Moreover, the estimations are also run on a subset of when it comes to workforce building for creative indus- plants that do not experience any change in the work- tries. Rather than focusing on the skills of one individual, force during the time they are in the sample. While there it is important to look at how well it matches the skill sets might be a selection of firms that satisfy this condition, of the people currently employed in the firm. A higher this exercise is still useful for us to understand the under- degree of relatedness is positive for labor productiv- lying mechanisms behind labor productivity. Results are ity (no matter if it is measured through value added or presented in Table  13 in the appendix and they show a wages), which is crucial for firm growth. These results slightly different picture where the diversity measures are of great importance in countries like Sweden where show a negative relationship to labor productivity. The knowledge-intensive firms are constantly struggling to relatedness measures for industry or occupation are now find the right person for the right job. Given that most statistically insignificant in all estimations, while their firms hire from the local labor market region, results sug - unrelatedness is still negatively related to productivity. gest that when deciding on the plant location it might be However, a higher share of related and industry experi- of importance to study the composition of skills in the ence is positively related to labor productivity, which is labor market as well. in line with the previous results. Similarly, unrelated- ness of industry and occupation is negatively related 5 Robustness and stability to productivity. What this table shows is that the posi- To further check the stability of the results, three differ - tive results of relatedness of industry or occupation ent sets of specifications are shown and discussed. experience might be partly driven by the churn in the labor force, where new related knowledge is crucial for 5.1 D iversity of experience or churn of employees? the productivity boost of employees. However, even if Labor mobility in creative industries is generally high there are no changes in the employees, they still benefit (Florida 2002; Frederiksen and Sedita 2011). It is there- from the relatedness of experience in both industry and fore important to ensure that the relationship found occupation. between diversity of experiences and productivity is not only driven by the hiring and firing behavior of the firm. 5.2 Plant size Two different estimations are shown to ensure that this To see if there are any differences between the smaller is indeed the case. First, Table  4 below shows the results and larger workplaces, the sample is separated between when share of new hires as well as the share of people the ones that employ at least 10 employees and the ones who have left the firm are included as control variables. that employ less than 10 employees. Note that previ- Even when the churn of the employees if controlled for, ous literature literature drops firms with less than 5 the diversity and relatedness variables behave similarly This is also confirmed when controlling for how long employees have To ensure that changes in occupational codes in 2014 are not driving the worked together. The results on the main variables of interest are unchanged, results, one more robustness test is shown in Table  12 in the appendix. The and therefore that analysis is not included in the paper. The results when con - model is now estimated for the period 2007–2014. Results are in line with trolling for how long the employees have worked together are available upon what has been presented earlier in the paper. request. 18 Page 12 of 21 O. Kekezi Table 4 Productivity estimations when controlling for new hires and those who have left the firm Value added – FE Value added – OP Wages – FE 1(a) 1(b) 1(c) 2(a) 2(b) 2(c) 3(a) 3(b) 3(c) FRACT_occu 0.031*** 0.024 0.004 (0.009) (0.016) (0.004) FRACT_ind − 0.028*** − 0.121*** − 0.044*** (0.008) (0.015) (0.004) Occ_R 0.021*** 0.050*** 0.022*** (0.008) (0.014) (0.004) Occ_U − 0.016* − 0.166*** − 0.019*** (0.009) (0.015) (0.004) Ind_R − 0.004 0.028** − 0.013*** (0.008) (0.014) (0.004) Ind_U − 0.044*** − 0.182*** − 0.047*** (0.007) (0.014) (0.004) Occ_Ind_R 0.019** 0.151*** 0.021*** (0.009) (0.017) (0.005) Occ_Ind_U − 0.053*** − 0.246*** − 0.049*** (0.009) (0.015) (0.004) Share hires − 0.356*** − 0.359*** − 0.357*** − 0.576*** − 0.544*** − 0.572*** − 0.192*** − 0.189*** − 0.187*** (0.009) (0.009) (0.009) (0.015) (0.015) (0.015) (0.004) (0.004) (0.004) Share left 0.017* 0.018* 0.017* 0.066*** 0.054*** 0.056*** 0.013*** 0.013*** 0.012*** (0.010) (0.010) (0.010) (0.016) (0.016) (0.016) (0.004) (0.004) (0.004) Observations 88,078 88,078 88,078 88,078 88,078 88,078 88,078 88,078 88,078 Plants 15,983 15,983 15,983 15,983 15,983 15,983 15,983 15,983 15,983 R-squared 0.791 0.791 0.791 0.898 0.898 0.898 Robust standard errors in parentheses for columns 1 and 3. For the OP estimations, bootstrapped standard errors are presented with 250 replications. ***p < 0.01, **p < 0.05, * < 0.1. The constant term is not reported. Control variables and year fixed effects are included in all estimations 5.3 Plant age employees (Östbring et  al., 2018) or 10 employees (Par- Following Timmermans and Boschma (2014), Table  6 rotta et  al., 2014a, 2014b). The results are presented in presents the results for plants which are at least 5  years Table 5 below. old, since the younger firms suffer from the liability of They are mostly in line with what has been shown newness (Stinchcombe 1965). In the sample, approxi- before for the larger firms. The main difference is that mately 26% of the plants are now dropped. smaller plants do not seem benefit from the relatedness Results are in line with the baseline estimation and of occupations or industries, but rather from the related- show that the diversity of occupational experience is ness of industry and occupation relatedness. Unrelated- positively related to labor productivity, but the diversity ness is however still negatively related to productivity. of industry experience is not. When looking at related- There can be two different explanations to why we do not ness, the relatedness of occupations, the relatedness of observe significant results for the relatedness of occupa - industry experience, as well as the relatedness in their tions. Small firms might not reach as high levels of relat - combination are positive for labor productivity. However, edness as the larger firms, and there too little variation industry relatedness shows no significance on the wage in the variable to show significant results. It can be the estimation. On the other hand, the unrelatedness of both case that due to the low number of workers people in the industry and occupation experience are negatively related plant need to work with different tasks simultaneously to firm performance. This suggests the importance of and work all together rather than to be separated into cognitive proximity among workers, when it comes to teams. It can also be that firms need to reach a specific knowledge spillovers and productivity advantages. size to benefit from relatedness. However, the relatedness Related to this issue, to understand what facilitates the of both industry and occupation experience is significant success of new firms in creative industries, Table  7 pre- for both categories. sents the results for start-ups instead of the older firms. Diversity of experience and labor productivity in creative industries Page 13 of 21 18 Table 5 Results with plants when a cutoff of 10 employees is made Average value added – FE Average value added – OP Average Wages – FE 1(a) 1(b) 1(c) 2(a) 2(b) 2(c) 3(a) 3(b) 3(c) More than 10 employees FRACT_occu 0.063*** 0.042 0.021*** (0.016) (0.031) (0.007) FRACT_ind − 0.019 − 0.263*** − 0.041*** (0.013) (0.024) (0.005) Occ_R 0.073*** 0.173*** 0.052*** (0.017) (0.031) (0.008) Occ_U − 0.019 − 0.193*** − 0.036*** (0.019) (0.033) (0.008) Ind_R 0.017 − 0.026 − 0.001 (0.014) (0.025) (0.006) Ind_U − 0.057*** − 0.410*** − 0.068*** (0.015) (0.028) (0.006) Occ_Ind_R 0.051*** 0.180*** 0.050*** (0.018) (0.033) (0.008) Occ_Ind_U − 0.090*** − 0.582*** − 0.088*** (0.019) (0.038) (0.008) Observations 31,769 31,769 31,769 31,769 31,769 31,769 31,769 31,769 31,769 Plants 7579 7579 7579 7579 7579 7579 7579 7579 7579 R-squared 0.798 0.799 0.798 0.914 0.915 0.914 Less than 10 employees FRACT_occu 0.042*** 0.019 0.019*** (0.014) (0.023) (0.007) FRACT_ind 0.047*** − 0.167*** 0.011* (0.013) (0.022) (0.006) Occ_R − 0.001 0.016 0.011* (0.011) (0.019) (0.006) Occ_U − 0.028** − 0.155*** − 0.013** (0.012) (0.018) (0.007) Ind_R 0.036*** − 0.028 0.014** (0.012) (0.021) (0.006) Ind_U 0.006 − 0.192*** − 0.005 (0.011) (0.017) (0.006) Occ_Ind_R 0.040*** 0.085*** 0.025*** (0.013) (0.023) (0.007) Occ_Ind_U − 0.018 − 0.217*** − 0.014** (0.014) (0.019) (0.007) Observations 37,430 37,430 37,430 37,430 37,430 37,430 37,430 37,430 37,430 Plants 7979 7979 7979 7979 7979 7979 7979 7979 7979 R-squared 0.797 0.797 0.797 0.895 0.895 0.895 Robust standard errors in parentheses for columns 1 and 3. For the OP estimations, bootstrapped standard errors are presented with 250 replications. ***p < 0.01, ** p < 0.05, * < 0.1. The constant term is not reported. Control variables and year fixed effects are included in all estimations Results here are not as clear cut, but rather similar to value added show a positive sign but the other estima- the results shown for the smaller plants in Table 6 above. tions (columns 2(a) and 3(a)) instead display a negative While the diversity of occupation continues to show relationship. Moreover, the relatedness measures are a positive relationship, the results for industry diver- mainly statistically insignificant or do not show consist - sity are not stable where the fixed effect estimations for ent results across the estimations when taken separately. 18 Page 14 of 21 O. Kekezi Table 6 Regression results when only firms that are at least 5 years old are included Value added – FE Value added – OP Wages – FE 1(a) 1(b) 1(c) 2(a) 2(b) 2(c) 3(a) 3(b) 3(c) FRACT_occu 0.069*** 0.051** 0.026*** (0.010) (0.020) (0.005) FRACT_ind 0.010 − 0.107*** − 0.022*** (0.009) (0.017) (0.004) Occ_R 0.041*** 0.081*** 0.031*** (0.009) (0.018) (0.005) Occ_U − 0.001 − 0.174*** − 0.010* (0.011) (0.019) (0.005) Ind_R 0.025*** 0.041** 0.002 (0.009) (0.018) (0.004) Ind_U − 0.014 − 0.178*** − 0.031*** (0.009) (0.016) (0.004) Occ_Ind_R 0.042*** 0.191*** 0.031*** (0.011) (0.021) (0.005) Occ_Ind_U − 0.040*** − 0.267*** − 0.039*** (0.012) (0.019) (0.005) Observations 65,424 65,424 65,424 65,424 65,424 65,424 65,424 65,424 65,424 Plants 12,020 12,020 12,020 12,020 12,020 12,020 12,020 12,020 12,020 R-squared 0.780 0.780 0.780 0.899 0.899 0.899 Robust standard errors in parentheses for columns 1 and 3. For the OP estimations, bootstrapped standard errors are presented with 250 replications. *** p < 0.01, ** p < 0.05, * < 0.1. The constant term is not reported. Control variables and year fixed effects are included in all estimations Table 7 Labor productivity in startups Value added – FE Value added – OP Wages – FE 1(a) 1(b) 1(c) 2(a) 2(b) 2(c) 3(a) 3(b) 3(c) FRACT_occu 0.092*** 0.066** 0.025*** (0.016) (0.028) (0.008) FRACT_ind 0.032** − 0.239*** − 0.024*** (0.015) (0.027) (0.007) Occ_R 0.017 0.034 0.012* (0.014) (0.024) (0.007) Occ_U − 0.003 − 0.170*** − 0.009 (0.015) (0.025) (0.008) Ind_R 0.017 − 0.059** − 0.015** (0.015) (0.025) (0.007) Ind_U − 0.020 − 0.294*** − 0.035*** (0.014) (0.024) (0.007) Occ_Ind_R 0.032** 0.117*** 0.013* (0.016) (0.029) (0.008) Occ_Ind_U − 0.031** − 0.294*** − 0.027*** (0.015) (0.025) (0.007) Observations 32,259 32,259 32,259 32,259 32,259 32,259 32,259 32,259 32,259 R-squared 0.770 0.770 0.770 0.903 0.903 0.903 Robust standard errors in parentheses for columns 1 and 3. For the OP estimations, bootstrapped standard errors are presented with 250 replications. *** p < 0.01, ** p < 0.05, * < 0.1. The constant term is not reported. Control variables are included in all estimations Diversity of experience and labor productivity in creative industries Page 15 of 21 18 When looking at the combination of industry and occu- a combination of industry and occupation, rather than pation experience results are in line with what has been when they are separated. This suggests that the specific previously shown in the paper; relatedness is positive for human capital of the individuals is connected to both productivity but unrelatedness of experience harms the industry and occupation. productivity growth of plants. What these results suggest Besides contributing to the literature regarding the is that when it comes to startups, the experience of the micro-mechanisms of knowledge spillovers and produc- workers needs to be diverse, but not too diverse. Since tivity which arise from the previous experience, these the teams in this case are smaller, and the probability of results are also important from a policy perspective. working together is larger, the diversity of teams needs Given the importance of creative industries in regional to be related both for industry and for occupation expe- development, understanding how labor productivity is rience, at least in the first years of the startups. These enhanced in these firms benefit the economy at large. results support the findings of Koster and Andersson Moreover, these results reflect the importance of finding (2018) who argue about the importance of occupational the right person for the right job. Knowledge-intensive skills on top of industry skills for the survival of startups. firms in Sweden are continuously having difficulties to Focusing only on one of these dimensions when examin- find the competence for the job. The results shown here ing the previous work experience is not enough to show results suggest that one potential way to look for the right positive results on productivity. competence is to consider the composition of the experi- ence of the people within the plant. Hiring people with related experience in terms of occupation or occupation 6 Conclusions and industry, would benefit the firm in the form of higher The paper studies how the diversity of work experiences labor productivity (which is mirrored in both value added among employees relates to labor productivity in crea- and wages). Given that most firms hire people from the tive industries in Sweden. The idea is that when chang - region, these results could also be analyzed as suggestion ing jobs, workers bring their expertise and knowledge for creative, knowledge intensive firms to locate in areas with them. While a large literature argues about the posi- where there is a large pool of people with related skills to tive effects of labor mobility, the type of knowledge and one another. skills that are brought into the firm is not largely studied. The study creates possibilities for further research. Some studies show however that what mostly contrib- Given the importance of occupational-specific skills utes to firm performance depends on the type of knowl - showed in the results, it would be interesting to dig edge that is brought in and how that matches the existing deeper into what type of occupations are the ones that knowledge base (Boschma et al. 2009; Timmermans and when combined productivity is enhanced. Previous lit- Boschma 2014; Östbring et al. 2018). Others have shown erature has shown how skills should not overlap for new the importance of knowledge diversity for innovation or knowledge to be created (Uzzi et al. 2013), but the litera- productivity growth in a firm (Parrotta et  al. 2014a , b). ture on occupational combinations is scarce. Moreover, it Yet, to my knowledge, no study has looked at the diver- would be interesting to look at this through an innova- sity of the previous experience of the workers, both in tion perspective. Third, while the purpose of this paper terms of occupations and industries, and how that relates has been to look at diversity and relatedness, it would to labor productivity. be insightful to expand the discussion by looking at skill The results of this paper show that diversity of occu - complementarity and firm productivity. Skill comple - pational experience is positive for labor productiv- mentarity is not captured in the diversity or relatedness ity, but this the diversity of industrial experience shows measures, but it would be a great avenue to expand the either insignificant or negative relationship. When the current analysis. Further, as previously mentioned, the distinction between relatedness and unrelatedness of results should be analyzed with caution, given the lack of experience is made, the results indicate that the positive a suitable instrumental variable or any other exogenous relationship is mostly driven by relatedness, which is in shock, which would have made possible causal results. line with similar existing studies on relatedness and per- Moving into the direction of causality is another avenue formance (Boschma et  al. 2009; Martynovich and Hen- where this work can be extended into. ning 2018; Östbring et al. 2018). This relationship is even stronger when experience relatedness is measured as Appendix See Tables 8, 9, 10, 11, 12, 13, and 14. 18 Page 16 of 21 O. Kekezi Table 8 Industries included in the analysis NACE Description 58 Publishing activities 59 Motion picture, video and television programme production, sound recording and music publishing activities 60 Programming and broadcasting activities 62 Computer programming, consultancy and related activities 71 Architectural and engineering activities; technical testing and analysis 72 Scientific research and development 73 Advertising and market research 74 Other professional, scientific and technical activities 90 Creative, arts and entertainment activities 91 Libraries, archives, museums and other cultural activities 93 Sports activities and amusement and recreation activities Table 9 Characteristics of the creative industries and the plants in the rest of the economy Non-creative industries Creative industries 2007 2016 Growth 2007 2016 Growth Employment 3,779,542 4,128,471 9.2% 361,781 424,406 17.3% Number of plants 419,993 498,471 18.7% 72,528 94,576 30.4% Average Wages 2111 2747 30.1% 2455 3104 26.4% Average Productivity 5495 6874 25.1% 5555 6938 24.9% Average sales 1814 1937 6.8% 1169 1288 10.2% Table 10 The 2-digit educational types Group Education type 1 General education 14 Pedagogics and teaching 21 Arts and media 22 The humanities 31 Social and behavioral science 32 Journalism and information 34 Business 38 Law and legal science 42 Biology and environmental science 44 Physics, chemistry, and geoscience 46 Mathematics and natural science 48 Computer science 52 Engineering: technical, mechanical, chemical, and electronics 54 Engineering: manufacturing 58 Engineering: construction 62 Agriculture 64 Animal healthcare 72 Healthcare 76 Social work 81 Personal services 84 Transport services 85 Environmental care 86 Security Diversity of experience and labor productivity in creative industries Page 17 of 21 18 Table 11 Correlation matrix 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 (1) Productivity 1.000 (2) Wages 0.605 1.000 (3) Capital 0.181 0.098 1.000 (4) Labor 0.129 0.224 0.444 1.000 (5) FRACT_occu − 0.022 − 0.038 0.216 0.347 1.000 (6) FRACT_ind − 0.055 − 0.049 0.102 0.309 0.468 1.000 (7) Occ_R 0.102 0.155 0.061 0.092 0.465 0.142 1.000 (8) Occ_U − 0.198 − 0.308 0.014 − 0.060 0.443 0.217 − 0.498 1.000 (9) Ind_R 0.093 0.199 − 0.011 0.148 0.133 0.514 0.174 − 0.087 1.000 (10) Ind_U − 0.189 − 0.301 0.020 − 0.021 0.296 0.589 − 0.017 0.346 − 0.340 1.000 (11) Occ_Ind_R 0.112 0.216 0.013 0.130 0.251 0.385 0.524 − 0.291 0.718 − 0.234 1.000 (12) Occ_Ind_U − 0.167 − 0.261 0.003 − 0.034 0.309 0.383 − 0.327 0.669 − 0.245 0.680 − 0.236 1.000 (13) FRACT_Edu − 0.062 − 0.112 0.172 0.243 0.313 0.186 0.060 0.156 0.001 0.126 0.002 0.104 1.000 (14) Edu 0.191 0.353 − 0.019 0.184 − 0.006 0.083 0.009 − 0.081 0.170 − 0.102 0.116 − 0.044 − 0.188 1.000 (15) Age 0.027 0.021 0.225 0.241 0.021 − 0.177 − 0.054 − 0.033 − 0.175 − 0.119 − 0.130 − 0.077 0.072 − 0.061 1.000 (16) Multiplant 0.154 0.139 0.263 0.297 − 0.019 0.020 0.026 − 0.153 0.026 − 0.069 0.041 − 0.095 − 0.017 0.068 0.122 1.000 (17) Change_LA − 0.026 − 0.010 − 0.023 0.031 0.078 0.260 0.027 0.051 0.137 0.172 0.099 0.117 − 0.041 0.114 − 0.228 0.163 1.000 (18) Den 0.141 0.244 − 0.049 0.129 0.091 0.111 0.090 − 0.037 0.161 − 0.054 0.136 − 0.049 0.087 0.224 − 0.095 − 0.164 − 0.165 1.000 18 Page 18 of 21 O. Kekezi Table 12 Akaike information criteria for the different estimations Average value added Average wages K,L 50,799 − 93,790 Control variables 49,958 − 96,290 Full model 1(a) 49,796 Full model 1(b) 49,809 Full model 1(c) 49,824 Full model 3(a) − 96,408 Full model 3(b) − 96,733 Full model 3(c) − 96,564 In the first columns the AIC is calculated only when including capital and labor in the estimations. In the second columns, all control variables are included besides the variables of interest. In the last three columns, the full models are estimated. 1(a)–1(c) and 3(a)–3(c) correspond to the estimations in Table 3 Table 13 Regression results when the sample ends in 2014 to ensure robustness from changes in SSYK codes Average value added – FE Average value added – OP average wages – FE FRACT_occu 0.073*** 0.061*** 0.032*** (0.011) (0.020) (0.006) FRACT_ind 0.013 − 0.177*** − 0.024*** (0.009) (0.017) (0.004) Occ_R 0.030*** 0.052*** 0.031*** (0.010) (0.018) (0.005) Occ_U − 0.006 − 0.175*** − 0.002 (0.011) (0.017) (0.005) Ind_R 0.023** − 0.004 − 0.002 (0.009) (0.017) (0.004) Ind_U − 0.016* − 0.240*** − 0.035*** (0.009) (0.015) (0.004) Occ_Ind_R 0.041*** 0.135*** 0.028*** (0.011) (0.020) (0.005) Occ_Ind_Un − 0.023** − 0.302*** − 0.035*** (0.011) (0.016) (0.005) Observations 66,748 66,748 66,748 66,748 66,748 66,748 66,748 66,748 66,748 Plants 14,786 14,786 14,786 14,786 14,786 14,786 14,786 14,786 14,786 R− squared 0.793 0.793 0.793 0.900 0.900 0.900 Robust standard errors in parentheses for columns 1 and 3. For the OP estimations, bootstrapped standard errors are presented with 250 replications. ***p < 0.01, **p < 0.05, * < 0.1. The constant term is not reported. Control variables and year fixed effects are included in all estimations Diversity of experience and labor productivity in creative industries Page 19 of 21 18 Table 14 Labor productivity in firms that have not experienced any change in the workforce Value added – OLS Value added – OP Wages – OLS 1(a) 1(b) 1(c) 2(a) 2(b) 2(c) 3(a) 3(b) 3(c) FRACT_occu − 0.160** − 0.189* − 0.129** (0.063) (0.107) (0.054) FRACT_ind − 0.198* − 0.153 − 0.147* (0.104) (0.136) (0.079) Occ_R − 0.033 − 0.069 0.031 (0.055) (0.093) (0.039) Occ_U − 0.268*** − 0.267*** − 0.287*** (0.056) (0.091) (0.058) Ind_R 0.038 0.100 0.077 (0.088) (0.126) (0.067) Ind_U − 0.271*** − 0.266** − 0.219*** (0.093) (0.121) (0.078) Occ_Ind_R 0.246** 0.358** 0.253*** (0.121) (0.166) (0.084) Occ_Ind_U − 0.271*** − 0.288** − 0.315*** (0.098) (0.134) (0.117) Observations 1,175 1,175 1,175 1,175 1,175 1,175 1,175 1,175 1,175 R-squared 0.275 0.294 0.274 0.349 0.385 0.356 Robust standard errors in parentheses for columns 1 and 3. For the OP estimations, bootstrapped standard errors are presented with 250 replications. *** p < 0.01, ** p < 0.05, * < 0.1. The constant term is not reported. Since the diversity measures are constant, the results in columns 1 and 3 do not include firm fixed effects, but instead industry, region, and year fixed effects. Control variables are included in all estimations Acknowledgements Received: 25 August 2020 Accepted: 22 June 2021 I want to thank the three anonymous referees, Martin Henning, Ron Boschma, Rikard Eriksson, Charlotta Mellander, Johan Klaesson, Sandy Dall’erba, Geoffrey Hewings, Jonna Rickardsson, and Emma Lappi for helpful comments and sug- gestions during different stages of this paper. References Authors’ contributions Alesina, A., Devleeschauwer, A., Easterly, W., Kurlat, S., Wacziarg, R.: Fractionali- I am the sole author of the paper, thus responsible for the whole manuscript. zation. J. Econ. Growth 8(2), 155–194 (2003) Almeida, P., Kogut, B.: Localization of knowledge and the mobility of engineers Funding in regional networks. Manag. Sci. 45(7), 905–917 (1999) Not applicable. And, H.W.A., Isaksen, A.: Knowledge intensive business services and urban industrial development. Serv. Ind. J. 27(3), 321–338 (2007) Availability of data and materials Arcidiacono, P., Kinsler, J., Price, J.: Productivity spillovers in team production: The micro-data used in this paper are protected by the Swedish Secrecy Act evidence from professional basketball. J. Law Econ. 35(1), 191–225 (2017) and cannot be freely accessed or shared. The data can however be ordered Autor, D.H., Levy, F., Murnane, R.J.: The skill content of recent technological through Statistics Sweden, through the following link: https:// scb. se/ en/ servi change: an empirical exploration. Q. J. Econ. 118(4), 1279–1333 (2003) ces/ guida nce- for- resea rchers- and- unive rsiti es/. Bacolod, M., Blum, B.S., Strange, W.C.: Skills in the city. J. Urban Econ. 65(2), 136–153 (2009) Bantel, K.A., Jackson, S.E.: Top management and innovations in banking: Does Declarations the composition of the top team make a difference? Strateg. Manag. J. 10(S1), 107–124 (1989) Ethics approval and consent to participate Barbour, E., Markusen, A.: Regional occupational and industrial structure: does Not applicable. one imply the other? Int. Reg. Sci. Rev. 30(1), 72–90 (2007) Bassett-Jones, N.: The paradox of diversity management, creativity and innova- Consent for publication tion. Creat. Innov. Manag. 14(2), 169–175 (2005) Not applicable. Becker, G.S.: Investment in human capital: a theoretical analysis. J. Polit. Econ. 70(5, Part 2), 9–49 (1962) Competing interests Becker, G.S.: Human Capital: A Theoretical Analysis with Special Reference to The author has no competing interests to disclose. Education. Columbia University Press, New York (1964) Bercovitz, J., Feldman, M.: The mechanisms of collaboration in inventive teams: Author details composition, social networks, and geography. Res. Policy 40(1), 81–93 1 2 SOFI, Stockholm University, Stockholm, Sweden. Centre for Entrepreneur- (2011) ship and Spatial Economics (CEnSE), Jönköping International Business School, Berliant, M., Fujita, M.: The dynamics of knowledge diversity and economic Jönköping, Sweden. growth. South. Econ. J. 77(4), 856–884 (2011) 18 Page 20 of 21 O. Kekezi Boschma, R.: Proximity and innovation: a critical assessment. Reg. Stud. 39(1), Manning, A., Petrongolo, B.: How local are labor markets? Evidence from a 61–74 (2005) spatial job search model. Am. Econ. Rev. 107(10), 2877–2907 (2017) Boschma, R., Eriksson, R., Lindgren, U.: How does labour mobility affect the Maroto-Sánchez, A.: Productivity in the services sector: conventional and performance of plants? The importance of relatedness and geographical current explanations. Serv. Ind. J. 32(5), 719–746 (2012) proximity. J. Econ. Geogr. 9(2), 169–190 (2009) Martynovich, M., Henning, M.: Labour force building in a rapidly expanding Card, D., Heining, J., Kline, P.: Workplace heterogeneity and the rise of West Ger- sector. Ind. Innov. 25(2), 199–227 (2018) man wage inequality. Q. J. Econ. 128(3), 967–1015 (2013) Mas, A., Moretti, E.: Peers at work. Am. Econ. Rev. 99(1), 112–145 (2009) Castañer, X., Campos, L.: The determinants of artistic innovation: bringing in Mellander, C.: Creative and knowledge industries: an occupational distribu- the role of organizations. J. Cult. Econ. 26(1), 29–52 (2002) tion approach. Econ. Dev. q. 23(4), 294–305 (2009) Caves, R.E.: Creative Industries: Contracts Between Art and Commerce. Harvard Miguel-Molina, B., Hervas-Oliver, J.-L., Boix, R., De-Miguel-Molina, M.: The University Press, Cambridge (2000) importance of creative industry agglomerations in explaining the Christopherson, S.: The divergent worlds of new media: how policy shapes wealth of European regions. Eur. Plan. Stud. 20(8), 1263–1280 (2012) work in the creative economy. Rev. Policy Res. 21(4), 543–558 (2004) Mincer, J.: Schooling, Experience, and Earnings. National Bureau of Econom- Cohen, W.M., Levinthal, D.A.: Absorptive capacity: a new perspective on ics Research, New York (1974) learning and innovation. Adm. Sci. q. 35(1), 128–152 (1990) Müller, K., Rammer, C., Trüby, J.: The role of creative industries in industrial Delgado-Verde, M., Castro, G.-d, Amores-Salvadó, J.: Intellectual capital and innovation. Innovation 11(2), 148–168 (2009) radical innovation: Exploring the quadratic effects in technology-based Mulligan, C.B., Sala-I-Martin, X.: Measuring aggregate human capital. J. Econ. manufacturing firms. Technovation 54, 35–47 (2016) Growth 5(3), 215–252 (2000) Essletzbichler, J.: Relatedness, Industrial Branching and Technological Cohe- Neal, D.: Industry-specific human capital: evidence from displaced workers. sion in US Metropolitan Areas. Reg. Stud. 49(5), 752–766 (2015) J. Law Econ. 13(4), 653–677 (1995) Florida, R.: The rise of the creative class: and how it’s transforming work, Neffke, F.: Coworker complementarity. SSRN J. (2017). https:// doi. org/ 10. leisure, and everyday life. Basic Books, New York (2002)2139/ ssrn. 29293 39 Florida, R., Mellander, C., Stolarick, K.: Inside the black box of regional Neffke, F., Henning, M.: Skill relatedness and firm diversification. Strateg. development—human capital, the creative class and tolerance. J. Econ. Manag. J. 34(3), 297–316 (2013) Geogr. 8(5), 615–649 (2008) Nooteboom, B.: Learning and Innovation in Organizations and Economies. Frederiksen, L., Sedita, S.R.: Embodied knowledge transfer for innovation: Oxford University Press, Oxford (2000) comparing interfirm labor mobility between music and manufacturing Nooteboom, B., Van Haverbeke, W., Duysters, G., Gilsing, V., van den Oord, industries. In: Belussi, F., Staber, U. (eds.) Managing Networks of Creativ- Ad.: Optimal cognitive distance and absorptive capacity. Res. Policy ity. Routledge, Abingdon (2011) 36(7), 1016–1034 (2007) Frenken, K., Van Oort, F., Verburg, T.: Related variety, unrelated variety and Olley, G.S., Pakes, A.: The dynamics of productivity in the telecommunica- regional economic growth. Reg. Stud. 41(5), 685–697 (2007) tions equipment industry. Econometrica 64(6), 1263–1297 (1996) Gathmann, C., Schönberg, U.: How general is human capital? A task-based Östbring, L., Lindgren, U.: Labor mobility and plant performance: on the approach. J. Law Econ. 28(1), 1–49 (2010) (dis)similarity between labor- and capital-intensive sectors for knowl- Glaeser, E.L., Mare, D.C.: Cities and skills. J. Law Econ. 19(2), 316–342 (2001) edge diffusion and productivity. Geogr. Ann. Ser. B Hum. Geogr. 95(4), Goos, M., Manning, A.: Lousy and lovely jobs: the rising polarization of work 287–305 (2013) in Britain. Rev. Econ. Stat. 89(1), 118–133 (2007) Östbring, L., Eriksson, R., Lindgren, U.: Labour mobility and organisational Horwitz, S.K., Horwitz, I.B.: The effects of team diversity on team outcomes: a proximity: routines as supporting mechanisms for variety, skill integra- meta-analytic review of team demography. J. Manag. 33(6), 987–1015 tion and productivity. Ind. Innov. 24(8), 775–794 (2017) (2007) Östbring, L., Eriksson, R., Lindgren, U.: Relatedness through experience: on Hotho, S., Champion, K.: Small businesses in the new creative industries: the importance of collected worker experiences for plant performance. innovation as a people management challenge. Manag. Decis. 49(1), Pap. Reg. Sci. 97(3), 501–518 (2018) 29–54 (2011) Østergaard, C.R., Timmermans, B., Kristinsson, K.: Does a different view create Howkins, J.: The Creative Economy: How People Make Money from Ideas. something new? The effect of employee diversity on innovation. Res. Penguin, UK (2002) Policy 40(3), 500–509 (2011) Ingram, B.F., Neumann, G.R.: The returns to skill. Labour Econ. 13(1), 35–59 Parent, D.: Industry-specific capital and the wage profile: evidence from the (2006) national longitudinal survey of youth and the panel study of income Jara-Figueroa, C., Jun, B., Glaeser, E.L., Hidalgo, C.A.: The role of industry- dynamics. J. Law Econ. 18(2), 306–323 (2000) specific, occupation-specific, and location-specific knowledge in Parrotta, P., Pozzoli, D., Pytlikova, M.: Labor diversity and firm productivity. the growth and survival of new firms. Proc. Natl. Acad. Sci. 115(50), Eur. Econ. Rev. 66, 144–179 (2014a) 12646–12653 (2018) Parrotta, P., Pozzoli, D., Pytlikova, M.: The nexus between labor diversity and Jarvis, H., Pratt, A.C.: Bringing it all back home: The extensification and ‘over - firm’s innovation. J. Popul. Econ. 27(2), 303–364 (2014b) flowing’ of work. Geoforum 37(3), 331–339 (2006) Penrose, E.: The Theory of the Growth of the Firm. Oxford University Press, Jehn, K.A., Northcraft, G.B., Neale, M.A.: Why differences make a difference: a Oxford (1959) field study of diversity, conflict and performance in workgroups. Adm. Perretti, F., Negro, G.: Mixing genres and matching people: a study in Sci. q. 44(4), 741–763 (1999) innovation and team composition in Hollywood. J. Organ. Behav. 28(5), Kambourov, G., Manovskii, I.: Occupational specificity of human capital. Int. 563–586 (2007) Econ. Rev. 50(1), 63–115 (2009) Pitcher, P., Smith, A.D.: Top management team heterogeneity: personality, Koster, S., Andersson, M.: When is your experience valuable? Occupation- power, and proxies. Organ. Sci. 12(1), 1–18 (2001) industry transitions and self-employment success. J. Evol. Econ. 28(2), Power, D.: Behind the music. Profiting from sound: a systems approach to the 265–286 (2018). https:// doi. org/ 10. 1007/ s00191- 017- 0528-2 dynamics of the nordic music industry. Oslo: STEP/Nordic Innovation Kremer, M.: The O-ring theory of economic development. Q. J. Econ. 108(3), Centre (2003) 551–575 (1993). https:// doi. org/ 10. 2307/ 21184 00 Protogerou, A., Kontolaimou, A., Caloghirou, Y.: Innovation in the European Larsen, J.N.: Knowledge, human resources and social practice: the creative industries: a firm-level empirical approach. Ind. Innov. 24(6), knowledge-intensive business service firm as a distributed knowledge 587–612 (2017) system. Serv. Ind. J. 21(1), 81–102 (2001) Savino, T., Petruzzelli, A.M., Albino, V.: Teams and lead creators in cultural and Lazear, E.P.: Globalisation and the market for team-mates. Econ. J. 109(454), creative industries: evidence from the Italian haute cuisine. J. Knowl. 15–40 (1999) Manag. 21(3), 607–622 (2017) Madsen, T.L., Mosakowski, E., Zaheer, S.: Knowledge retention and personnel Schumpeter, J.A.: The Theory of Economic Development: An Inquiry Into mobility: the nondisruptive effects of inflows of experience. Organ. Sci. Profits, Capital, Credit, Interest, and the Business Cycle, vol. 55. Transaction 14(2), 173–191 (2003) Publishers, New Jersey (1934) Diversity of experience and labor productivity in creative industries Page 21 of 21 18 Scott, A.J.: Human capital resources and requirements across the metropolitan UNESCO.: Creative economy report 2013: special edition: widening local devel- hierarchy of the USA. J. Econ. Geogr. 9(2), 207–226 (2008) opment pathways. UNCTAD New York: United Nations (2013) Serafinelli, M.: “Good” firms, worker flows, and local productivity. J. Law Econ. Uzzi, B., Spiro, J.: Collaboration and creativity: the small world problem. Am. J. 37(3), 747–792 (2019) Sociol. 111(2), 447–504 (2005) Siepel, J., Cowling, M., Coad, A.: Non-founder human capital and the long-run Uzzi, B., Mukherjee, S., Stringer, M., Jones, B.: Atypical combinations and scien- growth and survival of high-tech ventures. Technovation 59, 34–43 tific impact. Science 342(6157), 468–472 (2013) (2017) Van Ark, B.: Measuring the new economy: an international comparative per- Solheim, M.C.W., Boschma, R., Herstad, S.J.: Collected worker experiences and spective. Rev. Income Wealth 48(1), 1–14 (2002) the novelty content of innovation. Res. Policy 49(1), 103856 (2020) Visintin, F., Pittino, D.: Founding team composition and early performance of Söllner R.: Human capital diversity and product innovation: a micro-level university—based spin-off companies. Technovation 34(1), 31–43 (2014) analysis. Jena Econ. Res. Papers 2010(027) (2010) Williams, K., O’Reilly, C.: The complexity of diversity: a review of forty years of Stinchcombe, A.L.: Social structure and organizations. Handb. Organ. 7, research. Res. Organ. Behav. 21, 77–140 (1998) 142–193 (1965) Wixe, S.: The impact of spatial externalities: skills, education and plant produc- Sullivan, P.: Empirical evidence on occupation and industry specific human tivity. Reg. Stud. 49(12), 2053–2069 (2015). https:// doi. org/ 10. 1080/ 00343 capital. Labour Econ. 17(3), 567–580 (2010)404. 2014. 891729 Tao, J., Ho, C.-Y., Luo, S., Sheng, Y.: Agglomeration economies in creative indus- Wixe, S., Andersson, M.: Which types of relatedness matter in regional growth? tries. Reg. Sci. Urban Econ. 77, 141–154 (2019) Industry, occupation and education. Reg. Stud. 51(4), 523–536 (2016) Taylor, A., Greve, H.R.: Superman or the fantastic four? Knowledge combina- Young, A.: Consistency without Inference: Instrumental Variables in Practical tion and experience in innovative teams. Acad. Manag. J. 49(4), 723–740 Application. Unpublished manuscript, London School of Economics (2006) (2019) Thompson, W., Thompson, P.: From industries to occupations: rethinking local economic development. Econ. Dev. Comment. 9(3), 12–18 (1985) Publisher’s Note Timmermans, B., Boschma, R.: The effect of intra- and inter-regional labour Springer Nature remains neutral with regard to jurisdictional claims in pub- mobility on plant performance in Denmark: the significance of related lished maps and institutional affiliations. labour inflows. J. Econ. Geogr. 14(2), 289–311 (2014)

Journal

Journal for Labour Market ResearchSpringer Journals

Published: Jul 6, 2021

Keywords: Diversity; Skill relatedness; Previous experience; Labor mobility; Knowledge spillovers; J24; L25

There are no references for this article.