Going offshore or better staying in? Spatial relocation strategies and their impact on firm innovativeness

ABSTRACT

We analyse the short- to mid-run effects of spatial relocation strategies on firm innovativeness and productivity growth. Using conditional difference-in-difference estimation with multiple treatments, we find for a sample of German firms in 1999–2013 that offshoring has a statistically significant negative impact on the firms’ innovation activity and productivity growth vis-à-vis nonrelocating comparison firms. In contrast, we find a positive link between domestic relocation activities and the firms’ propensity to introduce a product innovation. Firms should thus carefully account for potentially distorting performance effects when deciding about the spatial scale of relocation strategies in the short to mid-run.

KEYWORDS:

• Relocation
• offshoring
• innovation
• difference-in-difference
• multiple treatments

JEL CLASSIFICATION:

• F23
• O31
• C23

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplemental data for this article can be accessed here.

Notes

¹ We refer to offshoring as all kinds of strategic relocations of business functions across national borders. That is, both locational shifts to subsidiaries and spin-offs within the internal organization of the firm, typically classified as ‘offshore in-house sourcing’, as well as relocations of business functions to external suppliers and contractors, i.e. ‘offshore outsourcing’, are included (see Olsen 2006). Accordingly, domestic relocations comprise in-house sourcing and outsourcing activities as well.

² See Ellguth, Kohaut, and Möller (2014) for details. Data was accessed with controlled data teleprocessing by the FDZ.

³ The third category (offshoring) also comprises simultaneous domestic and international relocations, which cannot be further disentangled on the basis of the IAB data. This mixed strategy, however, only covers very few cases (confidentiality issues prevent us from providing exact case information).

⁴ See McCaffrey et al. (2013) and the supplemental data and underlying research materials for details on the underlying assumptions of propensity score weighting with multiple treatments.

⁵ An overview of CDiD estimation is, e.g., given in Heckman et al. (1998).

⁶ See the supplemental data and underlying research materials for the first-step multinomial regression results and balancing tests for covariates.