Labour force building in a rapidly expanding sector

Abstract

Between 1991 and 2010, jobs in the knowledge-intensive IT services sector in Sweden increased from 30,000 to 104,000. Departing from recent theoretical insights suggesting that the skill composition of worker inflows is an indicator of knowledge relevant to employers, we investigate labour inflows into the sector. Who were the people getting jobs in this expanding sector? And, how were their skills valued by employers as the sector evolved? Our findings suggest that sectoral evolution was not reflected in how the skills of incoming workers were valued, but rather in who was hired into the sector. The paper suggests that the analysis of worker inflows is a tool for investigating the evolution of both sectors and their knowledge bases. It provides some lessons for industrial and educational policies regarding technologically turbulent industries, and takes the first step towards developing an approach that integrates industry dynamics with labour force sourcing and evolution.

Keywords:

• Industry dynamics
• evolution
• labour sourcing
• labour mobility
• human capital

JEL Codes:

• J23
• J24
• J44
• L86

Acknowledgements

The authors would like to thank two anonymous reviewers for their constructive comments and suggestions. The authors are grateful for the constructive advice and critical comments on earlier versions of this paper provided by Emelie Hane-Weijman, Karl-Johan Lundquist, David Rigby, Olof Zaring, and Christian Østergaard. The paper has benefited from fruitful discussions at the Workshop on Industry Relatedness and Regional Change held in Umeå, Sweden, on October 29–31, 2014, the Annual Meeting of the American Association of Geographers held in Chicago, IL, USA, on April 21–25, 2015, and the Workshop on Evolutionary Approaches Informing Research on Entrepreneurship and Regional Development held in Gothenburg, Sweden, on December 10–11, 2015.

Notes

¹ We exclude two 5-digit industries also classified as computer and related activities, namely (1) the maintenance and repair of office, accounting and computing machinery, because this industry has a different functional focus than the selected industries, and (2) other computer-related activities, because this industry includes a vast range of heterogeneous activities. Besides, the latter industry changed its classification code when the industrial classification system changed in 2002, making it difficult to ensure consistent observation over the entire period we were studying.

² Our selection variable is the sectoral employment growth in excess of three years after entry. We expect it to be positively linked to the probability of an individual remaining in employment in the sector.

³ Entry into one of the IT service industries is observed at a particular point in time, entailing that individuals can appear several times in the data-set during the period.

⁴ 5-digit industries at t−1 and t should be different.

⁵ Post-upper secondary education is predominantly university education, but could also include other forms. Upper seconday school is the highest level of secondary education in Sweden.

⁶ Regions are local labour markets defined by patterns of commuting between municipalities in a way that maximises the homogeneity of mobility patterns within a region while sustaining cross-regional heterogeneity (Statistics Sweden 2010). This definition of a region makes it possible to distinguish between ‘real’ movers and commuters in the most efficient way.

⁷ We introduce six regional groups (or families) according to the methodology suggested by the Swedish Agency for Economic and Regional Growth (NUTEK 2004): the metropolitan areas (Stockholm, Gothenburg and Malmö), large regional centres, smaller regional centres, and peripheral regions. Criteria involved in the definitions of these groups include population size and density, business dynamics, share of individuals with higher education, and access to higher education institutions.

⁸ Here defined as having at least two years of upper secondary education (including individuals with doctoral degrees).

⁹ An entropy measure is calculated as the joint entropy of four characteristics: the educational level, educational track, type of job switch (job-to-job, unemployment-to-job, non-participation-to-job), and previous background in the sector: , where is the observed probability of each possible combination of background characteristics. Higher joint entropy values imply a greater dispersion of background characteristics.

¹⁰ These models have not been reported, but they are available from the authors upon request.