Summary highlights

Contributions of the Paper: This paper provides novel evidence on the effects of global sourcing strategies that involve spatial relocation activities on firm performance. Indicators of firm innovativeness, including the propensity to introduce/implement product and process innovations, are chosen as performance measures. The focus on innovation activities is motivated by their key role for firm competitiveness in global markets. With regard to the spatial dimension of business relocations, we distinguish between domestic and international (offshore) activities. From an organizational perspective, we classify relocation activities as being captive (in-house) or relying on an outsourcing arrangement. The core focal point is to test if treatment effects of relocation activities on innovation differ by firm size and organizational configuration.

Research Questions/Purpose: Although there is a constantly growing literature in the field of business and entrepreneurship research that studies the business implications of specific relocation modes, only very few studies have yet comprehensively assessed the performance impact of different relocation activities at the individual firm level. Prior studies have also predominately focused on an analysis of large multinationals—thereby, neglecting the fact that global sourcing strategies have also become increasingly popular for small and medium-sized enterprises (SMEs). This study aims at closing this research gap by answering the following research questions: Does the impact of alternative relocation activities on firm innovativeness differ by firm size? Does it further depend on the organizational mode of the relocation activity? And, if so, what are the key strategic lessons for the innovation management of SMEs?

Database / Information: We employ micro data for German firms from two complementary data sets that cover a particularly large share of SMEs. This allows us to conduct empirical analyses for subgroups of SMEs (i.e., micro, small, and medium-sized firms). As primary data source, we use annual data from the IAB Establishment Panel between 1999 and 2014. While the IAB Establishment Panel offers favorable conditions for robust micro-econometric analyses, namely, (1) a large number of firms, (2) a sufficiently long sample period, and (3) a rich set of firm-specific control variables, it lacks information on the specific organizational model of the underlying relocation activity. To overcome this limitation, we have designed an own supplementary survey for firms located in the Lower Rhine region in Germany, which addresses the role of the organizational setup of the relocation activity. The survey was conducted in 2017/18 and covers information for a 5-year interval ranging from 2013 to 2017 in a cross-sectional setup.

Methodology: An empirical identification approach is applied here that relies on micro-econometric tools for treatment effect estimation. Given the different features of the two datasets utilized in the empirical analysis, we run regressions for panel and cross-sectional data. As main empirical specification, we apply inverse-probability-of-treatment weighted (IPTW) and doubly robust conditional difference-in-difference (CDiD) estimation for multiple treatments. Estimation is organized as a two-stage approach: We first use propensity score estimation (multinomial logit regression) to match treated and comparison firms. We then use the obtained propensity score values to weigh sample observations. In the second stage, we run several fixed effect (FE) regressions to quantify average treatment effects on the treated for firm innovativeness as a function of relocation activity (treatment variable). Effect heterogeneity by firm size is estimated based on a sub-sampling strategy and an interaction term approach. In addition, we run alternative cross-sectional regressions for cases in which panel data are not available.

Results/Findings: The empirical results from panel regressions for the IAB Establishment Panel indicate that relocations only translate into an increased innovation performance for large firms. Particularly for international relocations, average treatment effects are negative and more pronounced the smaller the firm is. Size thresholds for significantly negative relocation effects vary between below 100 (product innovation) and approximately below 300 employees (process innovation). In comparison, for domestically relocating firms, we generally do not find significant differences in the post-relocation innovation performance relative to similar non-relocating sample firms. The regression results from the supplementary firm survey for firms located in the Lower Rhine region in Germany underline negative, size-specific effects of international relocations on the rate of process innovation. These effects are mainly attributable to offshore outsourcing.

Practical Implications and Recommendations: Managers of micro and small firms should be aware that an offshoring activity may—at least temporary—slow down firm innovativeness. This slowdown is particularly significant for the introduction of process innovations, which may indicate that international relocations in small firms increase project completion time, bear hidden costs, absorb significant internal resources, or reflect a lack of internal resources to sufficiently mitigate the challenges of offshoring and the interruption of intra-organizational channels for cooperation and learning. Although offshoring itself constitutes an (organizational) innovation, the slowdown in other dimensions of innovativeness may have adverse effects for the competitive position of offshoring firms in the short to mid run, for instance, related to product quality, lead time, and production flexibility. Another important implication from our empirical analysis is that firms should carefully evaluate the “make-or-buy” decision associated with a relocation activity. Our empirical survey results point to a relatively stronger innovation activity of firms using captive relocation strategies.

Limitations and Future Research Directions: Estimation results presented here are limited as only short- to mid-run effects of spatial relocation activities can be identified. It can thus expect that the identified negative effects constitute upper bounds for the overall negative impact as relocation costs typically occur immediately, while benefits may potentially unfold gradually over time. With longer time series available, a reassessment of the relocation-innovation nexus should be made. Another limitation is that our data only distinguish between domestic and international relocations without further information about geographical choices. Previous research has shown that this two-way categorization may be too simplistic in order to reflect strategic location drivers that determine the location choice of firms and their subsequent performance. Future research should use spatially disaggregated data on international relocation activities to better understand the implications of the geographical choice of destination regions for firm performance. Future research should also focus on the intensity of relocation activities and its innovation impact at the firm level as prior research has indicated the presence of tipping points, i.e., relocation intensity levels from which positive effects become negative.

Introduction

Whereas the last decades witnessed an almost unbowed trend towards business model globalization (Lee et al., 2012), trade-political tensions, the COVID-19 crisis, and particularly the Russo-Ukrainian war have recently shown the vulnerability of global sourcing strategies to sudden exogenous shocks (Raza et al., 2021; Ruta, 2022; Korn and Stemmler, 2022). These developments mark a new phase of re-globalization (Benedikter, 2022) that potentially challenge the view that international sourcing strategies within global value chain (GVC) production are a key source for international competitiveness for small and large firms (De Gregorio, 2009). With growing uncertainty about GVC robustness and resilience in times of crisis, also the future of global sourcing strategies, such as captive offshoring and offshore outsourcing, is more controversially debated: While some scholars advocate alternatives to GVC production such as nearshoring, reshoring, and even the full repatriation of value chains as part of a re-nationalization process (see, e.g., Javorcik, 2020, and Miroudot, 2020, for a critical summary), others argue that high expectations on the potentials of reshoring are exaggerated (Raza et al., 2021). Instead, the future of GVC production should be based on more efficient and diversified global sourcing strategies supported by an intensified use of ICT (Baldwin and Tomiura, 2020; Gurbuz et al., 2023). This unsolved debate about the future of (re-)globalization poses the question whether global or domestic sourcing strategies are generally better equipped to support the strategic goals of firms, particularly small- and medium-sized enterprises (SMEs).

We contribute to this current debate by investigating the role that business relocations at national and international scales have on the consecutive innovation performance of firms. Relocations as part of a complex sourcing strategy are thereby treated as the spatial and organizational reconfiguration of a firm’s business activities. Although there is a constantly growing literature in the field of business and entrepreneurship research that studies the business implications of specific relocation modes such as captive offshoring or offshore outsourcing (see, e.g., Tate et al., 2009, Larsen et al., 2013, da Silveira, 2014, Stentoft et a., 2015, Caniato et al., 2015, Mazzola et al., 2019), only very few studies have yet comprehensively assessed the impact of different relocation activities at the individual firm level (exceptions are Görg and Hanley 2011, Kreutzer and Mitze 2017). Moreover, most prior studies have placed the focus of their analyses on large multinationals—thereby, neglecting the fact that global sourcing strategies have also become increasingly popular among small and medium-sized enterprises (SMEs).

We aim at closing this research gap by answering the following research questions: Does the impact of alternative relocation activities on firm innovativeness differ by firm size? Does it further differ by the organizational mode of the relocation activity? And, if so, what are the key strategic lessons for the innovation management of SMEs?

Different facets of firm innovativeness have been chosen as key performance measures here as they essentially determine firm competitiveness in global markets (Cantwell, 2005, Nkongolo-Bakenda et al., 2010, Asemokha et al., 2019). With regard to the spatial dimension of business relocations, we distinguish between domestic and international (offshore) activities. From an organizational perspective, we classify relocation activities as being captive (in-house) or are part of an outsourcing arrangement. To provide answers to the above research questions, we utilize two complementary micro data sets for German firms and apply micro-econometric tools to shed light on the relocation-innovation-firm size nexus.

Both data sets share the feature that they cover a particularly large share of SMEs, which allowed us to conduct empirical analyses for sub-groups of SMEs (i.e., micro, small, and medium-sized firms). As primary data source, we used annual data from the IAB Establishment Panel between 1999 and 2014. A limitation of the IAB Establishment Panel is that we could not differ between the organizational mode of relocation activities (Larsen et al., 2013, Steinberg et al., 2017, Mazzola et al., 2019). To bypass this limitation, we designed an own supplementary survey for firms located in the Lower Rhine region in Germany, which addresses the role of the organizational setup of the relocation activity. The survey was conducted in 2017/2018 and covers information for a 5-year interval ranging from 2013 to 2017 in a cross-sectional setup.

Our main estimation results based on the IAB Establishment Panel indicate that relocations (both domestic and international) only translate into an increased innovation performance for large firms. Particularly for international relocations, the average treatment effects are negative and more pronounced the smaller the firm is. Size thresholds for significantly negative relocation effects vary between below 100 (product innovation) and approximately below 300 employees (process innovation). In comparison, for domestically relocating firms, we generally do not find significant differences in the innovation performance over time relative to similar non-relocating sample firms. The regression results from our supplementary firm survey underline negative, size-specific effects of international relocations on the rate of process innovation. Additionally, we found that this effect is mainly attributable to offshore outsourcing arrangements. These results have relevant implications for strategic management decisions in SMEs and support earlier work that point to hidden costs of international relocations that negatively affect firm performance—at least in the short to mid run (e.g., Larsen et al., 2013).

The remainder of this paper is organized as follows: Section 2 provides definitions and a conceptual framework to study the relationship between firm size, relocation activities, and firm innovativeness. While Section 3 briefly reviews the empirical literature on this issue, Section 4 covers data and estimation issues. Section 5 presents the empirical results for the treatment effects of relocation on firm innovativeness, and Section 6 finally discusses managerial implications together with limitations and future research directions.

Definitions and theoretical considerations

Relocations are here defined as spatial location changes of originally co-located business activities that may differ by their organization configuration as a captive (in-house) or outsourced arrangement (Olsen, 2006).Footnote 1 Relocating firms mainly seek to realize efficiency gains from specialization (Kedia and Mukherjee, 2009). While domestic and international relocation strategies have thus parallel routes and are intertwined (Ali-Yrkkö and Deschryvere, 2008),Footnote 2 the decision to relocate internationally (offshore) rather than domestically offers several distinct opportunities for firms such as (i) exploiting a larger cost reduction potential due to lower (marginal) factor costs, (ii) making use of a business environment and labor pool not available in the domestic market, or (iii) tapping into foreign knowledge resources (Quinn 2000, Lewin et al. 2009, Manning, 2014, Lundberg and Rehnfors, 2018). These factors may positively affect economies of scale and scope experienced by the individual firm and may subsequently alter the firm’s competitive advantage and innovativeness (Nieto and Rodríguez 2011, Mihalache et al. 2012, Johnson et al., 2013, Rodríguez and Nieto 2016, Mazzola et al., 2019). Whether firms are able to fully exploit these scale effects or not depends partly on the organizational complexity associated with the relocation decision (da Silveira, 2014).

Despite these opportunities, offshoring activities are typically also associated with higher organizational costs than domestic relocations. This may be due to complex search costs in foreign markets or coordination costs to link domestic and offshore business activities. Furthermore, cultural differences, information asymmetries, confidentiality, and intellectual property rights issues may lead to problems in partner selection and the conclusion of contracts (Clampit et al., 2015). Similarly, policing, enforcement, and monitoring cost rise with strategic importance and the frequency of partnership interactions (Shenkar, 2001, Ambos and Ambos, 2009). Finally, as Larsen et al. (2013) and Manning (2014) point to hidden implementation cost of complex relocation arrangements that negatively affect firm performance. Hidden costs may stem from unpredicted task complexities such as a high degree of tacit knowledge flows and the presence of inexact and unknown means-ends connections. Moreover, such costs may also be associated with the disruption to the cohesion and consistency of the firm’s internal activity configuration (Lampel and Bhalla, 2011).

We use these cost-benefit considerations as input for a stylized conceptual model to formulate propositions about theoretically expected effects of relocation activities on firm performance as a function of firm size. The focus is on “second-order selection effects”, which arise if firms have more than one option to pursue the same goal, i.e., profit maximization (Mrázová and Neary, 2019). Point of departure is a simple model framework that distinguishes between international and domestic relocations but ignores the associated organizational decision. In line with the related literature on heterogeneous firms in international business studies (e.g., Helpman et al., 2004, Mukherjee, 2010, Lileeva and Trefler, 2010), our interest is in comparing the behavior of firms whose cost parameters differ by a finite amount under different ways of operating in a market. The model is subsequently extended to cover differences in the organizational arrangements of the relocation activities.

We start by considering the objective function of a representative firm that produces a knowledge output y (e.g., product innovations) on the basis of a knowledge production function (KPF). Output is created by using the specialized knowledge input q such as research and development (R&D) employment (Fritsch and Franke, 2004). The KPF can then be written as

$$y={q}^{\alpha },$$
(1)

where the parameter α measures the return to scale. The use of input q is associated with a specific cost structure. Without relocation (status quo), the firm’s total cost function (cN) is assumed to have the following form

$${c}_N= mq,$$
(2)

where m is the marginal cost parameter of production using input q, which is assumed to be constant and can take a value between 1 ≥ m ≥ 0; i.e., it is defined as a fraction of the final price of the knowledge output being the numeraire. In comparison to the status quo, relocation activities involve fixed costs f related to search, installation, and contracting costs as outlined above but also reduce marginal production costs by a discount factor θ, for instance, due to the exploitation of spatial factor price differences.

If we assume that (i) fixed costs of international relocations (fI) and (ii) the discount on marginal costs (ϑ) are strictly larger compared to domestic relocations,Footnote 3 we can compute threshold values indicating which business strategy is associated with minimizing overall costs for increasing levels of input use (firm size). That is, for firm size up to size \(q<\frac{f_D}{\left(1-\theta \right)m}\), the status quo of no relocation has the lowest unit costs of (knowledge) production. For \(q\ge \frac{f_D}{\left(1-\theta \right)m}\), domestic relocation minimizes costs up to \(q\ge \frac{\left({f}_I-{f}_D\right)}{\left(\theta m-{\theta}^{\vartheta}\right)m}\) from which on international relocation yields the lowest unit costs of (knowledge) production. Hence, we observe that larger firms select into activities with lower marginal; this is what Mrázová and Neary (2019) label “conventional second-order selection effects”. The basic selection mechanism can be used to formulate a first proposition as

Proposition 1:

If relocation activities reduce marginal costs of knowledge production but involve strictly higher fixed costs, the relative advantage of relocation activities vis-à-vis no relocation increases with firm size. For sufficiently small firms, relocation effects can become negative compared to the reference case of no relocation.

Table 1 provides a summary of the knowledge production, cost and profit functions for the cases of (i) no relocation, (ii) domestic, and (iii) international (offshore) relocation activities. A numerical illustration for the sequential switch in minimum unit costs is given in Appendix A in the online supplementary materials.

Table 1 Cost-benefit comparison of spatial relocation strategies as a function of firm size
Full size table

As one extension, we can draw a link to the organizational mode of the relocation activity. This can be done by endogenizing the two cost parameters (f and m) as a function of the underlying organizational mode f = f(Orga) and m = m(Orga). If more complex organizational arrangements associated with relocation activities involve higher fixed costs, then under ceteris paribus conditions, larger firms should be better equipped to cover these organizational costs. If the relocation strategy simultaneously lowers marginal costs, the size effect should be more pronounced; i.e., the degree of complexity of the organizational arrangement of relocation activities acts as a leverage factor to the mechanism outlined in Proposition 1. For instance, if captive offshoring implies higher fixed installation costs than outsourcing arrangements but, at the same time, reduces the relative marginal cost parameter, e.g., through lower monitoring and adjustment costs, than the relative advantage of captive vis-à-vis outsourcing-based relocation activities should increase with firm size.

A second extension relates to differences in the return to scale parameter (α) of the KPF as shown in Eq. (1). If we assume that this parameter varies by relocation mode (both in a spatial and organizational context), this affects the threshold value for profit maximization as a function of firm size. Typically, higher returns to scale (or scope) are assumed to hold for international compared to domestic relocations (e.g., den Butter and Pattipeilohy 2007). Such differences may reflect external increasing returns to scale through foreign knowledge sourcing or be associated with leveraging core competencies and fostering specialization through offshoring (Mihalache et al., 2012, Steinberg et al., 2017). Bagchi et al. (2015) provide evidence for larger economics of scale of offshore outsourcing activities compared to domestic in-house production in the software industry. A numerical example that highlights changes in the optimal relocation mode for differences in the returns to scale parameter (α) is given in Appendix A of the online supplementary materials.

There are, however, also arguments that an international relocation may be associated with a fall in the returns to scale parameter of knowledge production if (domestically established) collaboration networks are weakened or dissolved (research findings by Karpaty and Gustavsson-Tingvall, 2015, and Fritsch and Görg, 2015, may point in this direction). A more careful theoretical analysis of exponential returns to R&D is, for instance, given in Mrázová and Neary (2019). The authors also discuss other cases, e.g., endogenous fixed costs, for which the above-described outcome of “conventional second-order selection effects” is less robust. Although a full theoretical treatment goes beyond the scope of this paper, we can use the insights from the model extensions to formulate a second proposition as:

Proposition 2:

The degree of organizational complexity operates as a leverage factor for the effects of relocation activities. The relative advantage of more complex organizational configurations increases with firm size if these involve strictly higher fixed costs, lower marginal costs, and larger returns to scale of production through knowledge sourcing or specialization.

As the relative size and significance of cost and returns to scale effects remain a priori unclear, we approach the issue empirically in the following.

Prior empirical literature

Complementing the theoretical considerations, we briefly review the related strands of the empirical literature on the link between firm size and (i) relocation, (ii) innovation, and (iii) most notably the relocation-innovation-firm size nexus. This shall guide our empirical estimation approach and serve as a reference point for our empirical findings.

Firm size and relocation activity

There is no general consensus among empirical scholars if larger firms are more likely to relocate than smaller ones. Evidence pointing to a positive correlation between firm size and relocation activity are reported in Pennings and Sleuwaegen (2000), da Silveira (2014), and Murphy and Siedschlag (2018), whereas Pellenbarg et al. (2002) and Brouwer et al. (2004) find that smaller firms are more likely to relocate. This ambivalence in the empirical literature links to our theoretical discussion of relocation activities being potentially associated with additional organizational costs and returns, i.e., scale effects from specialization or the chance to compensate for limited resources—particularly among SMEs (Erramilli and D’Souza 1993, Zacharakis 1997, Manning, 2014).

On the one hand, SMEs may thus have a strong incentive to relocate internationally as offshoring allows them to focus on their core competencies when facing resource constraints (financial, physical, and human).Footnote 4 Lewin et al. (2009) and Roza et al. (2011) report evidence that smaller firms have a higher probability of offshoring innovative parts of the value chain (e.g., R&D, product design, and engineering services) and that entrepreneurial motives and resource needs are the most important factors for small firms, while for larger firms, cost factors are the main driver of offshoring decisions. By offshoring certain parts of the value chain, SMEs can reap substantial benefits and enhance their business model (Musteen et al., 2017).

On the other hand, offshoring activities also require certain resources, which may be scarce in SMEs. For instance, binding credit constraints may hamper the internationalization activities of SMEs (Brutscher et al., 2012, Kranzusch and Holz, 2013). Survey results indicate that approximately half of the offshoring SMEs encounter substantial difficulties in the financing stage of their (planned) relocation processes (IfM Bonn, 2015). Attracting human capital, particularly high-skilled labor, also appears to be more difficult for SMEs compared to large companies (Cardon and Stevens, 2004, Desouza and Awazu, 2006) and limits SMEs’ ability to strongly commit to a foreign market (Cieślik et al., 2012, Ripollés et al., 2012).

Firm size and innovation performance

As for the link between firm size and relocation, no consensus has emerged from the prior empirical literature with regard to whether smaller or larger firms are more innovative, i.e., introduce more product and/or process innovations to the market. One reason for this lack of comparative evidence is that SME-based knowledge production functions substantially vary from innovation systems driven by large firms (Conte and Vivarelli, 2005). While some studies find that SMEs are on average more innovative and innovate faster due to their greater flexibility compared to larger enterprises (Prater and Gosh, 2006, Gusenbauer et al., 2015), other authors report a burden of smallness resulting from a lack of a critical mass of resources necessary for successful innovation projects (Brüderl and Schüssler, 1990). Additionally, large firms have scale advantages when it comes to innovation costs (Vaona and Pianta, 2008).

In contrast, small firms that rely more on tacit and external knowledge are more likely to depend on technologies directed towards the most intensive use of locally abundant production factors (Antonelli and Scellato, 2015). Piergiovanni et al. (1997) point out that large firms depend on R&D inputs, while the exploiting of spillovers is important for SMEs. Baumann and Kritikos (2016) find that, although firm size is negatively correlated with R&D intensity, micro firms are similarly successful as larger firms in terms of translating R&D inputs into innovations and also in terms of raising productivity levels through innovation.

Firm size and the relocation-innovation nexus

There is a rapidly growing empirical literature that analyzes the effects of (international) relocation activities on firm performance in general and firm innovativeness in particular (see Mitze and Kreutzer, 2017, for an overview). One strand of the literature thereby distinguishes between different types of offshoring, i.e., offshore outsourcing versus captive offshoring (Steinberg et al., 2017, Mazzola et al., 2019). Another strand of the literature compares the effects of offshoring vs. domestic relocations (Görg and Hanley, 2011, Fritsch and Görg, 2015) and offshoring, backshoring, and staying at home manufacturing (Stentoft et al., 2015).

Görg and Hanley (2011) find a positive link between relocation and R&D expenditures, which is stronger for offshoring compared to domestic relocation activities. Similarly, Fritsch and Görg (2015) report evidence that offshoring activities are associated with greater investments in R&D and upgrading of existing products compared to domestic relocations. However, only domestic relocations are found to be positively correlated firm innovativeness. Kreutzer and Mitze (2017) find a positive link between domestic relocation activities and the firms’ ability to introduce a product innovation but a statistically significant negative impact from offshoring on the firms’ innovation activity and productivity growth in the first years after relocation. Stentoft et al. (2015) do not find differences in productivity among manufacturing firms that have offshored, backshored, or maintained production at home.

Despite this growing evidence on the relationship between relocation and firm performance, only very few studies focus on the role of firm size therein: Görg and Hanley (2004) find for the Irish electronics sector during the period 1990–1995 that only those firms gain from offshoring in terms of profitability who are considerably larger than the average sample firm. Mitze and Kreutzer (2017) report evidence that small firms experience a significantly larger reduction in the propensity to introduce a process innovation than large firms. However, no evidence is available yet that compares the innovation effects across different relocation types for firms of different size.

Summarizing the arguments made in the empirical literature, prior findings give some support to our propositions that the observed effects of alternative relocation strategies depend on firm size. The findings that international relocations are only beneficial for large firms and may reduce the innovation performance of (sufficiently) small firms may thereby point to the fact underlying, potentially hidden, cost structures matter. As the empirical literature has also shown, local knowledge networks and knowledge spillovers may be crucial for the innovation performance of small firms. Associated with our second proposition, prior empirical studies point to the fact that a slowdown in performance is more pronounced for certain relocation activities, such as (captive) offshoring, if these activities lower the firm’s external knowledge sourcing capabilities and limit intra-organizational learning effects.

Data, variables, and estimation strategy

Panel data and difference-in-difference estimation

Our empirical analysis is built on two complementary data sets. As main data source, we used rich micro data for German firms obtained from the IAB Establishment Panel (see Ellgut et al., 2014, for data documentation, as well as Mitze and Kreutzer, 2017, for an earlier application). Our sample utilizes information for roughly 17,500 firms during the period 1999–2014; we excluded all firms facing a full or partial shutdown during this period as this may confound the effect of relocation activities. With regard to the set of indicators proxying firm innovativeness as key outcome variables, we made use of binary dummies measuring

  • Product innovation (i.e., the dummy takes a value of one if the firm has introduced a new product or service to the market in a given year and is zero otherwise).Footnote 5

  • Product improvement (i.e., the dummy takes a value of one if the firm has improved an already existing product or service in a given year and is zero otherwise).Footnote 6

  • Process innovation (i.e., the dummy takes a value of one if the firm has developed and/or introduced a new process to firm operations in a given year and is zero otherwise).Footnote 7

Dummies are codified in such a way that they are in accordance with the OECD Oslo Manual (see OECD, 2005). As treatment indicator, we created a multinomial variable (relocatei, t) that measures the relocation activity of firm i in time period t asFootnote 8

\({relocate}_{i,t}=\left\{\begin{array}{l}0\kern0.5em \textrm{if}\ \textrm{firm}\ i\ \textrm{does}\ \textrm{not}\ \textrm{relocate},\\ {}\begin{array}{cc}\;&\;\\ {}1& \textrm{if}\ \textrm{firm}\ i\ \textrm{relocates}\ \textrm{domestically},\end{array}\\ {}\begin{array}{cc}\;&\;\\ {}2& \textrm{if}\ \textrm{firm}\ i\ \textrm{relocates}\ \textrm{internationally}\ \left(\textrm{offshoring}\right).\end{array}\end{array}\right.\) 

The IAB Establishment Panel allows us to precisely identify these two different relocation modes in the sample waves 2007, 2008, and 2010. Unfortunately, for the other years, the survey does not distinguish between domestic and international relocation (the survey only asks for relocations in general). To avoid any bias in the estimation procedure due to measurement errors related to this latent information, we therefore excluded all firms that have reported to conduct any relocation (e.g., outsourcing or spin-off) activity in 1999–2006, 2009, and 2011–2014 since we cannot distinguish between domestic or international relocations. This leaves us with a comparison group that only includes non-relocating firms. One should note, though, that the IAB Establishment Panel does not allow us to distinguish between alternative organizational models of relocation activities as being captive or outsourcing.

We are particularly interested in studying the link between relocation and innovation as a function of firm size going beyond the standard “small-versus-large”-type firm categorization. Therefore, SMEs were divided into three distinct sub-groups along the lines of the EU recommendation 2003/361 as: (1) micro firms with less than 10 employees, (2) small firms which employ between 10 and 49 employees, and (3) medium-sized firms with 50 to 249 employees. Firms with 250 or more employees are defined as large firms. As shown in Table 2, the innovation performance varies considerably among firms in different size classes and with different treatment status.

Table 2 Number of sample firms and share of innovators by treatment status and size classes (data set 1: IAB Establishment Panel)
Full size table

Given these heterogeneities in the innovation performance of firms across size groups, it becomes apparent that simple mean outcome comparisons between relocating and non-relocating firms (grouped by firm size) would likely result in biased estimates on the causal impact of relocation on innovativeness. We included a broad set of firm-specific covariates in our estimation approach to control for potential drivers of a firm’s innovation activity apart from its relocation activity. Summary statistics for outcome, treatment, and control variables are given in Appendix B in the online supplementary materials.

To estimate treatment effects of relocation, we applied inverse-probability-of-treatment weighted (IPTW) and doubly robust conditional difference-in-difference (CDiD) estimation for multiple treatments (see, e.g., Robins et al., 2000, McCaffrey et al., 2013). Both econometric specifications build on a two-stage estimation approach, which first uses propensity score estimation (multinomial logit regression) to match treated and comparison firms. We then use this information to run weighted, firm-fixed effect regressions for firm innovativeness as a function of relocation activity (treatment variable) to quantify the average treatment effect on the treated (ATT). Given the limited time length of our sample, we focus on the short- to mid-run relocation effects on firm innovativeness in the first 3 years after the relocation activity took place (see Appendix C in the supplementary research materials for further details).

As it is a core interest of this paper to test for heterogeneous treatment effects linked to firm size, we applied the CDiD estimator in different regression setups: First, as default estimation approach, we partitioned the overall sample into four sub-samples and estimated separate ATTs for the classes of micro-, small-, medium-sized, and large firms. The advantage of this sub-sampling approach is that specific sample weights using the first-stage treatment probabilities can be estimated separately for each size class, which is likely to increase the quality of matches between firms in the treatment and comparison group.

However, the sub-sampling approach has the disadvantage of a limited number of sample observations to estimate the underlying ATTs for some size classes (see Table 2). Therefore, an alternative estimation approach is applied, which uses the full sample of firms covered in the IAB data and augments the standard CDiD estimation framework by including a multiplicative interaction term between the treatment variable and firm size in order to estimate conditional treatment effects along the entire distribution of firm size measured in terms of firm employment (as continuous variable). A technical description of the regression approach for pre-treatment propensity score estimation and sub-sample CDiD estimation and the multiplicative interaction term approach are given in Appendix C in the online supplementary materials.

Supplementary firm survey on the organizational configuration of relocations

While the use of data from the IAB Establishment Panel offers favorable conditions for robust micro-econometric analyses using a large sample of firms, unfortunately, it lacks information on the specific organizational model of the relocation activity and the underlying motives for relocations. This would allow us to tell a richer story on the conditional effects on innovativeness as a function of firm size. We therefore designed and conducted a supplementary survey covering firms located in the Lower Rhine region (“Mittlerer Niederrhein” and “Niederrheinische IHK Duisburg/Wesel”).Footnote 9 The survey focuses on questions on the type of relocation activity (domestic, international), motives for relocation, the organizational choice (captive offshoring versus offshore outsourcing), and information on relocated business units. In addition, firms were asked to report information on their past innovation performance for a 5-year interval until 2017 (along the innovation categories defined in the OECD Oslo Manual) together with a set of firm-specific control variables (including firm size). Six hundred ninety-eight firms registered in the local Chambers of Industry and Commerce (IHK) in the Lower Rhine region were contacted, and 166 firms completed the survey (response rate: 24%). The distribution of responses across size classes is 67% micro and small firms, 26% medium-sized firms, and 7% large firms.

With regard to the relocation activity of surveyed firms, 44 firms reported at least one spatial relocation activity within Germany, while 41 firms relocated internationally. The relocation intensity as share of relocated business operation relative to total operations varies between modes with an average intensity of 22% for domestic relocations and 31% for international relocations. With regard to the organizational setup of relocation activities, Table 3 indicates that domestic relocations are dominated by outsourcing activities (micro and small firms: 82%, medium-sized and large firms: 75%). The share of offshore outsourcing is somewhat lower but still the most frequently used international relocation mode (micro and small firms: 70%, medium-sized and large firms: 57%).

Table 3 Sample firms and share of innovators by treatment status (data set 2: supplementary firm survey)
Full size table

As further shown in Table 3, relocating firms are generally more innovative than non-relocating firms. Although this is generally in line with our IAB Establishment Panel sample (see Table 2),Footnote 10 the supplementary survey only has a cross-sectional setup, and empirical estimates from linear probability models (using ordinary least squares, OLS) cannot be strictly interpreted as causal given that they may suffer from an endogeneity bias (i.e., more innovative firms self-select into relocation activities). We account for selection effects in two ways: First, we controlled for firm-specific characteristics including firm age, employment growth in the last 5 years, and binary dummies indicating the organizational status (individual firm, head office of a national company, head office of an international company, subsidiary) together with sector dummies. In addition, we reduced the sample for estimation to only those relocating firms, which have reported relocation activities prior to our measures of innovativeness (i.e., we asked if the firm has introduced/implemented an innovation in the last 5 years; hence, only firms that have relocated prior to this date are included). To detect heterogeneity across firm size groups, interaction terms are included in the OLS regression model.

Empirical results

Difference-in-difference estimation on firm size heterogeneity of relocation effects

Table 4 shows the results of first-stage multinomial logit regressions for our sample from the IAB Establishment Panel. The logit model specifies the firm’s propensity to conduct a relocation activity as a function of firm characteristics in the pre-treatment period 1999–2006. As Boter and Holmquist (1996) previously pointed out, relocation and internationalization processes must be understood in the context of the industry, company, and people involved. The reported estimation results underline this argument in the light of different sample settings (overall plus sub-samples of micro, small, medium-sized, and large firms).

Table 4 Estimation results for first stage multinomial logit specification (data set 1: IAB Establishment Panel)
Full size table

By analyzing the key determinants for the firms’ relocation decision, some striking differences can be observed. As such, Table 4 highlights that only few features such as firm age matter uniformly across the different sub-sample specifications. Focusing on the group of micro firms, we see that higher turnover and physical capital investment levels, a larger share of high skilled employees and the availability of an in-house R&D department increase the probability that a micro firm with less than 10 employees relocate domestically. The availability of an in-house R&D department shows to be strongly correlated with the firm’s propensity to conduct offshoring activities, which may hint at the fact that at least some of the offshoring activities captured in our IAB data set are related to R&D and innovation offshoring. Additionally, in the case of offshoring, higher percentage shares of turnover in West Germany and in foreign countries as well as higher overall capital investment levels are significant influencing factors for the relocation decision (see da Silveira, 2014, for similar results). In particular, the positive correlation between the share of foreign turnover and the propensity to relocate internationally (offshore) underlines the importance of prior experience in foreign markets and the availability of international networks for offshoring firms (Levin et al., 2009).

While the observed first-stage results for small firms are similar for the group of micro firms, the results for medium-sized firms point to the existence of considerable heterogeneities between groups of firms with different size. Here, a higher overall turnover level and a higher percentage share of foreign turnover are found to reduce the probability that a firm pursues a domestic relocation, while a higher share of skilled employees and a higher share of intermediates are positively correlated with domestic relocation activities. The latter correlation may hint at the role played by underlying input-output linkages between the firm and its suppliers, which offer the scope for gains from specialization through domestic relocation.

With regard to offshoring activities of medium-sized firms, the results in Table 4 indicate that organizationally independent firms with high shares of turnover in West Germany and abroad as well as higher employment levels have a higher propensity to relocate internationally. The latter results are more in line with the observed determinants for offshoring activities by micro and small firms; however, the availability of an in-house R&D department and a higher share of high-skilled employees do not significantly correlate with offshoring activities of medium-sized firms. Finally, for large firms, the availability of an in-house R&D department and a higher foreign turnover share increase the probability to conduct an offshoring activity, while higher turnover levels reduce the latter probability.

Taken together, the first-stage estimation results on the firm’s propensity to conduct a specific type of relocation activity already provide indirect support to our conceptual model. They indicate that alternative relocation modes are driven by different firm-specific determinants (reflecting cost-benefit considerations) and they significant differ by firm size.

Table 5 reports the second-stage regression on the causal effects of relocation with regard to product and process innovation. The table reports the empirical results for IPTW and doubly robust CDiD estimations for the overall sample of all firms as well as the four size-related sub-samples. While the upper part of Table 5 reports the estimated ATTs when comparing domestically relocating and similar non-relocating firms, the lower part of Table 5 compares internationally relocating (offshoring) firms with the comparison group of similar non-relocating firms. As the table shows, we do not find significant differences between the innovativeness of domestically relocating and non-relocating firms. The only exception relates to the ability of micro firms to introduce a product innovation, which is reduced by roughly 5% per year in the first 3 years after the relocation activity took place. However, the effect is only found to be significant for the IPTW estimator and at the 10% significance level.

Table 5 CDiD regression results for causal effect of relocation activities on firm innovativeness (data set 1: IAB Establishment Panel)
Full size table

Different from the insignificant findings for domestic relocations, the empirical results for the causal effect of offshoring on firm innovativeness show that firm-specific heterogeneity strongly matters and that the short- to mid-run effects for different innovation types are negative. The most pronounced effects are found for process innovation, and estimates by size classes show a clear tendency of decreasing innovation performance for smaller sized firms. Our estimated ATTs report a roughly 30% reduction in the firm’s (average annual) propensity to introduce a process innovation in the first 3 years after the offshoring activity took place. The negative marginal effect gradually decreases with firm size but remains statistically significant for all sub-groups of SMEs. Only for large firms we do not observe a statistically significant effect, which provides evidence for Proposition 1.

The link between firm size, relocation, and innovativeness according to Proposition 1 is also confirmed when we plot the estimation results from CDiD specifications with interaction terms. Figure 1 thereby plots the estimated ATTs as a function of firm size (proxied log-transformed employment levels, see equation (A.4) in Appendix C in the online supplementary materials for details).

Fig. 1
figure 1

Conditional average treatment effects on the treated (ATTs) for relocation activities (data set 1: IAB Establishment Panel). Notes: Estimated treatment effects are calculated with a maximum lag length of three years; log(employees) measures the log-transformed employment level of firms. Dashed lines indicate 90% confidence intervals. Regressions based on IPTW specification. See Appendix C in the online supplementary research materials for further details on the estimation setup. Underlying regression tables can be obtained from the authors upon request

Full size image

As the plotted ATTs in panel C of Fig. 1 show for the case of process innovation, the smaller the firm in terms of employment levels, the larger is the estimated negative impact of offshoring on firm innovativeness. Similar ATT plots are also obtained for product innovation (panel A) and product improvements (panel B)—albeit the effects for the latter two variables are of lower magnitude. In the case of product innovation, the graph of plotted ATTs further shows that for firms with more than 50 employees, the negative effect turns into a positive effect for product innovation (as indicated by the intersection of lower-level confidence band with the zero line).

With regard to domestic relocations, the plotted ATTs in Fig. 1 mainly support the sub-sample estimates from Table 5 and report insignificant ATTs in most cases. Exceptions are positive treatment effects for large firms with regard to product innovation (in line with international relocations), while the effect for process innovation turns negative with increasing firm size. All in all, the plotted ATTs provide further strong support for heterogeneous effects for the relocation-innovation nexus with regard to the spatial scale of relocation activity and the moderating role of firm size.

The role of the organizational configuration for relocation effects

Table 6 reports the cross-sectional regression results for a series of linear probability models (LPM) covering the three outcome variables (product innovation, product improvement, and process innovation).Footnote 11 Regressions have been executed in a pairwise manner for the two samples “domestic versus no relocation” and “international (offshoring) versus no relocation”. The reported coefficient for relocate measures the marginal effect of a relocation on the innovation rate (in %) of micro and small firms. The reported coefficient of the interaction term (relocate × size) measures the additional effect of a relocation activity on the innovativeness of medium-sized and large firms. A key result from Table 6 is that international relocations are associated with a significant decline in the rate of process innovation for micro and small firms.

Table 6 LPM estimates for relocation and innovation (data set 2: supplementary firm survey)
Full size table

We also find a significant heterogeneity across firm size categories as the additional effect for medium-sized and large firms is significant and positive (the overall effect for the latter group can be calculated as − 0.267 + 0.353 = 0.086).Footnote 12 If we further decompose the negative overall effect on process innovation into effects stemming captive offshoring or offshore outsourcing activities, the results in Table 6 show that the effect is only statistically significant for offshore outsourcing. This result is in line with findings in the related literature (e.g., Larsen et al., 2013) and lends support to our second proposition that relocation complexity may negatively affect the consecutive firm performance. With regard to product innovation and product improvements, the estimation results indicate that domestic relocation activities within the boundaries of the firm (e.g., captive relocation activities) are associated with the highest product innovation rate. With regard to product improvements, captive (in-company) activities for both domestic and international relocations are positively correlated with the innovation rate. We do not observe a significant firm heterogeneity here.

Summary, managerial implications, and future research

By relocating some of their business activities, firms seek to increase efficiency through cost reductions and focus on core competencies and the exploitation of external (knowledge) resources. From a spatial sourcing perspective, firms can either relocate domestically or choose to offshore to a foreign location. In addition to this location decision, firms face an organizational choice whether they want to keep the relocated business activities in-house or outsource these activities to external contractors. In this paper, we have compared the impact of alternative relocation strategies on firm innovativeness—thereby focusing on the moderating role of firm size therein. Given that larger and smaller firms differ significantly with regard to internal (innovation) resources, management experience, and other capabilities, it is likely that this structural heterogeneity also filters through to the relocation-innovation nexus.

To identify the performance effects of alternative relocation activities, we have employed an empirical estimation approach based on two complementary data sets. As main vehicle for effect identification, we have utilized rich micro data from the IAB Establishment Panel to analyze the innovation activity of relocating German firms relative to a tailored comparison group comprising non-relocating firms with similar characteristics as relocating firms. Our micro-econometric results from IPTW and doubly robust conditional difference-in-difference estimation for the period 1999–2014 indicate that offshoring activities have a significantly negative effect on firm innovativeness in the short to mid run. Particularly, micro firms are found to lower their rate of process innovations in the first 3 years after the offshoring activity took place (compared to non-relocating or domestically relocating firms).

The key result of firm size being an important moderator for the link between offshoring and firm innovativeness is also supported by a multiplicative interaction term approach between the treatment variable and firm size. The results show for all three dimensions of innovativeness (product innovation, product improvement, and process innovation) that smaller firms reduce their innovation activities in the aftermath of the offshoring activity relative to non-relocating comparison firms. However, we also find that this negative effect diminishes with increasing firm size and, in the case of product innovation, turns into positive treatment effects for larger firms. Different from the varied effects for international relocation activities, our estimates generally do not indicate that innovation activities of domestically relocating firms change significantly in comparison to non-relocating firms in the first 3 years after the relocation activity.

As a robustness test and to gain insights on the role played by the organizational arrangement of the relocation activity, we had additionally designed and processed a supplementary firm survey on the relocation activities of German firms located in the Lower Rhine region. The findings from this data set add to our micro-econometric evidence on the basis of the IAB data: First, we also found a negative correlation between international relocations and process innovation. Second, this effect is subject to size class heterogeneity with significant relocation effects only for micro- and small firms. Third, the survey setup also allows us to distinguish between the effects of in- and out-sourcing activities: Here, we found that the negative effect on process innovation is mainly attributable to offshore outsourcing activities. Only captive relocation activities are found to be positively correlated with higher rates of product innovation (domestic relocation) and product improvement (domestic and international relocation). This underlines that next to the spatial dimension of the relocation activity also the organizational context of the relocation arrangement matters for firm performance.

Several noteworthy managerial implications for the innovation management of SMEs emerge from our results. First, and most importantly, managers of micro and small firms should be aware that an offshoring activity may—at least temporary—slow down firm innovativeness. This slowdown is particularly significant for the introduction of process innovations, which may point to the fact that international relocations in small firms increase project completion time (Bagchi et al., 2015), bear hidden costs (Larsen et al., 2013), absorb significant internal resources or reflect a lack of internal resources to sufficiently mitigate the challenges of offshoring (Manning et al., 2014), and cut off intra-organizational channels for cooperation and learning. Although offshoring itself constitutes an (organizational) innovation, the slowdown in other dimensions of innovativeness may have adverse effects for the competitive position of offshoring firms in the mid to short run, for instance, related to product quality, lead time, and production flexibility. As Stentoft et al. (2015) have similarly shown, offshoring firms face a decrease in their operational performance vis-à-vis non-offshoring firms.

Another important implication from our empirical analysis is that firms should carefully evaluate the “make-or-buy” decision associated with a relocation activity. Our empirical survey results point to a relatively higher innovation activity of firms using captive relocation strategies: For domestic relocations, this correlates significantly positive with higher rates of product innovation and improvement. For the case of international relocations, the relative decline in terms of process innovation vis-à-vis non-relocating firms is found to be insignificant for captive (in-sourcing) strategies, while it is significantly negative for offshore outsourcing. This result lends support to earlier findings in Mazzola et al. (2019) indicating that continuing offshore outsourcing efforts bear the risk of losing long-term R&D competitiveness. Managers, particularly in micro and small firms, should thus be aware that (offshore) outsourcing activities may especially reduce the innovative potential of firms through lacking intra-organizational cooperation and learning. This latter argument stresses the role of firms as living laboratories for innovation (Kline and Rosenberg, 1986).

In sum, what our theoretical considerations and empirical results have shown is that the relocation-innovation nexus is complex and that particularly small firms need to carefully estimate the direct and indirect costs of relocation decisions over time on the basis of proper (innovation) cost-accounting systems in order to avoid a negative surprise looming around the (foreign) corner when adopting a global sourcing strategy. Our results point to these challenges in “normal” times, which can be expected to be even more amplified in times of crisis (e.g., the COVID-19 pandemic and growing geo-political tensions, most notably the Russo-Ukrainian war since February 2022).

Finally, three main challenges for future research work can be addressed: First, one has to keep in mind that the estimation results presented here are limited to the identification of short- to mid-run effects of spatial relocation activities. It can thus expect that the identified negative effects constitute upper bounds for the negative impact as relocation costs typically occur immediately, while benefits may potentially unfold gradually over time (Mazzola et al., 2019). Hence, with longer time series available, a reassessment of the relocation-innovation nexus should be made. Previous research on the productivity effects of offshoring has already indicated that negative short-run effects may turn into positive effects in the long run (Fixler and Siegler, 1999, Egger and Egger, 2006).

Second, our main panel data estimation approach only distinguishes between domestic and international relocation. Previous research has shown that this two-way categorization may be too simplistic in order to reflect strategic location drivers that determine the (re-)location choice of firms and their subsequent performance (Caniato et al., 2015). Thus, future research should disaggregate international relocations to better understand the implications of the geographical choice of destination regions on firm performance. If data allow, this may even go that far as to recognize inter-regional differences in the origin and host countries of relocating firms (Hutzschenreuter et al., 2020). Third, future research should also focus on the intensity of relocation activities and its innovation impact at the firm level as prior research indicated the presence of tipping points, i.e., intensity levels from which positive effects become negative. Mazzola et al. (2019), for instance, found evidence that over-shoring, i.e., persisting with offshoring for too long, poses a risk of reducing innovation outputs. Likely reasons are that outsourcing firms lose the ability to learn through intra-organization cooperation and do not get in touch with new technological developments.