Relatedness and the Resource Curse: Is There a Liability of Relatedness?

abstract

Literature in evolutionary economic geography has emphasized knowledge spillover benefits of co-location with related industries. We draw on resource curse literature to demonstrate that relatedness also comes with costs in the form of labor market competition. Using a case study of a growth period in the Norwegian petroleum industry, we show that this had positive as well as negative implications for related industries. Industries related to petroleum grew faster than unrelated industries over the period. However, they also suffered from high labor costs and loss of human capital. Related industries had to pay higher wages than unrelated industries, even after controlling for worker characteristics. Furthermore, several of their employees, in particular the most productive ones, left for the petroleum industry. The relationship between petroleum and related industries is asymmetric insofar as workers tend to leave related industries for petroleum at higher rates than vice versa. Furthermore, the petroleum industry recruits the most productive workers from related industries and returns its least productive workers. Overall, this could potentially lead to de-skilling in related industries, which could more than outweigh any potential knowledge spillover benefits from their relatedness to the petroleum industry. Consequently, we argue that relatedness is not an even playing field: there may be losers, as well as winners, from relatedness.

Key words:

• relatedness
• resource curse
• petroleum
• labor mobility
• Norway

Acknowledgments

The authors are grateful to Frank Neffke, Ron Boschma, and Torfinn Harding as well as participants at the RSA Annual Conference 2015 in Piacenza, the Research Workshop on Entrepreneurship and Regional Development 2015 in Gothenburg, and the Geography of Innovation Conference 2018 in Barcelona. Silje Haus-Reve and Ole Bergesen provided valuable data assistance. This work is supported by the Research Council of Norway under the programs Demosreg, grant no. 209761, and VRI, grant no. 233737. The relatedness matrix and regional industrial statistics were analyzed in the former project based on data from Statistics Norway, and the relatedness matrix developed in this project is available in Timmermans and Fitjar (2015). The regional skill relatedness method was further developed in the latter project. Errors and omissions are the responsibility of the authors.

Notes

¹ Although on political and institutional implications of oil extraction at the regional level, see, for example, Caselli and Michaels (2013), Dube and Vargas (2013), and Fitjar (2010, 2013).

² Consistent with this story, Jacobsen and Parker (2016) identify a boom-and-bust cycle where petroleum regions have more positive employment and income growth in the short term but weaker long-term growth.

³ We selected this period because of data availability restrictions; however, as Figure 1 illustrates, this period closely overlaps a period of strong growth in petroleum prices and hence in the Norwegian petroleum industry.

⁴ We exclude employees in NACE rev.1.1 codes 75–91 (public administration, education, and health and social services) and NACE 74.50 (recruitment agencies).

⁵ Ekeland (2014) uses a similar definition, although omitting manufacturing of refined petroleum products, and tugboats and supply vessels. We follow Blomgren et al. (2013) in including these as petroleum industries.

⁶ An alternative approach that is used to obtain an exogenous relatedness matrix is to calculate relatedness using labor mobility in a comparable country. This turns out not to be a viable strategy in this case due to the uniqueness of the Norwegian petroleum industry. Other comparable countries either do not have a petroleum industry (e.g., Sweden and Germany) or have an industry with different skill requirements (e.g., Denmark). Thus, using data from a period preceding that of the analysis is the best option for obtaining an exogenous measure of relatedness. As we measure labor mobility on the basis of yearly changes, the last period used to calculate the relatedness matrix is 2003–4, while the first period used in the study is 2004–5. Hence, there is no overlap between the periods.

⁷ The relatedness matrix is a directed network. However, in the analysis, we include industries with high outbound and/or inbound relatedness to petroleum in our definition of petroleum-related industries. In effect, we therefore transform the matrix into an undirected network. Therefore, Figure 2 also shows undirected edges in the network graph.

⁸ In the transition from NACE rev. 1.1 to NACE rev. 2, the majority of industries are reassigned new industry codes. However, some industries are split in more detailed industry classes, and a few change sectors, for example, from manufacturing to services (such as publishing). We therefore see some growth in nontradables (and corresponding decline in tradables) between 2007 and 2008. This pertains mainly to unrelated tradables, while the vast majority of petroleum-related tradables translate well between the systems, and employment numbers are consistent. In making the transition from NACE rev. 1.1 to NACE rev. 2, we rely on the correspondence table provided by RAMON (http://ec.europa.eu/eurostat/ramon/index.cfm).

⁹ Admittedly, wages are not the only reason for moving into petroleum. Opportunities for development and training, the possibility to work with state-of-the-art technology, opportunities for job-related travel, the on- and off-duty work schedule while working offshore, and the expected job security that coincides with periods of growth, might all contribute to making the sector attractive. However, research has repeatedly demonstrated wages are the most important factor in accepting a job offer (Rynes, Gerhart, and Minette 2004). Since wage differences are so substantial, this is surely also a large driver of the observed mobility into petroleum.

¹⁰ The differences within the petroleum industry are also substantial, in particular between oil and gas extraction firms and other subcategories. The average wage in oil and gas extraction was NOK 1,037,112 (USD 184,954) in 2011, while in other petroleum industries, it varied from NOK 586,950 (USD 104,674) to NOK 925,800 (USD 165,103).

¹¹ The counties merged are Oslo and Akershus, Hedmark and Oppland, Aust-Agder and Vest-Agder, and Sør-Trøndelag and Nord-Trøndelag.

¹² Appendix Table A3 shows the results for analyses conducted separately for the 2004–7 and 2008–11 periods, corresponding to before and after the transition to NACE rev. 2. The results are highly consistent across both periods, indicating that the change in the NACE classification does not have a major impact on the results. Separate analyses for the other regression models are available from the authors upon request.

¹³ This is calculated by exponentiating the coefficients.

Additional information

Funding

This work was supported by the Norges Forskningsråd: [Grant Number Demosreg, 209761, VRI, 233737].