2.1 The rational choice approach

Confronted with different strategic options, of which business relocation is one, business managers face the challenge of deciding which of them serves their company’s goals and objectives best. The traditional regional science literature (e. g., Isard, 1956), describes strategic decision-making as a rational process, focussing on the strategic fit between a firm and its environment. To this end, business managers are assumed to carefully compile a list of locational requirements of their firms, analyse the features of different regions, and, to maximize profits, design the spatial organization of their production processes accordingly.

As such, it is assumed that managers monitor changes in their business environment regularly, grounding their strategic choices in a deliberate analysis of potential threats and opportunities. In this light, business relocation decisions are seen to be affected by both ‘push factors’ and ‘pull factors’. While push factors are defined as external circumstances that make a company want to leave its current location, pull factors are defined as external conditions that attract companies to a new location (Pellenbarg et al., 2002).

However, while it appears sensible to assume that negative alterations in the external business environment have the potential to push companies away from their current location, it must be kept in mind that companies differ in their need and ability to respond to such changes due to differences in their internal characteristics. When confronted with changes in their external business environment, managers are assumed to deliberately analyse their internal resources and capabilities, evaluating the feasibility of different strategic responses. On this matter, the resource-based view of the firm (Barney, 1991) argues that a firm’s ability to respond to changes derives from the tangible and intangible assets it controls, including physical capital, referring to physical assets, such as plants and equipment; human capital, covering the composition of the workforce; organisational capital, related to the organisational structure and processes; and financial capital, representing a company’s financial assets.

2.2 Towards more flexible, behavioural, and relational approaches

While the rational choice approach to strategy offers a useful structure for analysing relocation decisions, it has been criticized as simplistic (Whittington, 1993), as it assumes that i) all business managers continuously and effectively assess their business environment and internal resources, ii) accurate information about changes in the business environment is available to all managers; iii) all managers have the same cognitive ability to process such information; iv) the decision-making of managers is not affected by subjective considerations. Most of these assumptions are unrealistic. While some managers analyse their external business environment and internal resources very carefully, others simply ignore them, as they are pre-occupied with the daily business of survival (Porter, 1996). Moreover, information about the business environment is often imperfect and changes dynamically, particularly in times of high uncertainties (Clark, 2010). Finally, business managers differ in the capacity to process and apply information and are never completely free from being affected by subjective factors (e. g., Pred, 1967; Dahl and Sorenson, 2007). Rooted in this critique, alternative approaches have evolved not just in economic geography, but also in related fields, such as business management and economics.

In the business management literature, advocates of flexible approaches to strategy (e. g., Phelps and Kapsalis, 2001) argue that, given the complexity and volatility of the modern world, business managers would do better by planning for a range of different future scenarios rather than founding their strategic decisions on the characteristics of their current business environment. They argue that by doing so uncertainty can be reduced significantly, even though it can never be eliminated. Such flexible approaches to strategy, however, still do not take into account that business managers are likely to differ in their capacity and willingness to plan for different scenarios, and that subjective factors are likely to affect their decision-making. Hence, advocates of behavioural approaches to strategy (e. g., Cyert and March, 1963; Mintzberg and Quinn, 1991), argue that strategic decision-making, far from being a rational endeavour, is influenced by business managers’ personalities, experiences, ideologies, and values. Depending on such personal characteristics, business managers vary in their perceptions and judgements of changes in the business environment. Moreover, advocates of the extended resource-based view of the firm (e. g., Arya and Lin, 2007) consider relational resources, such as territorial embeddedness, as being essential in the strategic decision-making of business managers.

The nature of human behaviour is discussed not only in the business management literature, but also in behavioural economics. Behavioural economists examine how the characteristics of economic actors affect their decisions under conditions of risk and uncertainty (Winter, 1986). It is now widely acknowledged that the behaviour of economic actors is not only tied to rational decision-making (Kahneman and Tversky, 1979). Instead, it is assumed to be influenced by a dual system: behavioural processes that are reflective, controlled, deliberative, analytical, and governed by rules; and those that are affective, intuitive, emotional, and relatively unconscious (Kahneman, 2003). The role of affective behaviour is also evident in the “risk-as-feelings” model (Loewenstein et al., 2001), showing that in difficult situations decision-making is often driven by emotional reactions.

Such discussions have also found their way into economic geography. Closely related to behavioural economics, behavioural economic geography analyses economic activity in space at the level of individual economic actors (Strauss, 2008). It acknowledges that economic actors differ in their individual characteristics and argues that theoretical models of economic activity can be enhanced by integrating more realistic assumptions about human behaviour (e. g., Pred, 1967; North, 1974; Pen, 2000; Pellenbarg et al., 2002; Strauss, 2008). Based on Pred (1967) who argues that most (re)location decisions are not optimal, since accurate information required to make an optimal (re)location decision is not always readily available, while the cognitive ability of individuals to use information optimally might differ significantly, Pen (2000) calls for economic geographers to incorporate a stronger behavioural perspective when analysing location choices. In response, relational economic geography seeks to provide a deeper understanding of the spatial organization of production processes by drawing attention to the strategies and objectives of economic agents and regarding them as positioned in contexts of intertwined social and institutional relations, and as open-ended (Bathelt and Glückler, 2003). By regarding economic processes as being integrated into networks that are reaching beyond the locational structures of firms (Bathelt and Glückler, 2018), they set the foundation for analysing the importance of territorial embeddedness for strategic decision-making.

2.3 Conceptual framework

We strongly agree with Bathelt and Glückler (2003) and Strauss (2008) that analysing context-specific strategic decision-making of economic agents is core to economic geography. As such, we treat business managers and their relocation intentions and behaviours as the research object of this study. By grounding this paper in the research philosophy of critical realism, we engage with the literature to develop a holistic conceptual framework (see Figure 1). We take an eclectic approach, aiming to consider a wide array of factors that are discussed in the literature as affecting relocation decisions of firms. By doing so, we aim at analysing to which degree the strategic decision-making process of economic actors is open-ended and unpredictable.

To this end, we regard the flexible, behavioural, and relational approaches on the one hand and the rational choice approach on the other as not being mutually exclusive. Rather, we consider the strategic decision-making of business managers as being positioned on a spectrum between one and the other. As such, we follow Lerner et al. (2015) who propose a model of decision-making that attempts to account for both rational and affective elements. Their model shares similarities with the rational choice approach by assuming that decision-makers evaluate different options available to them (e. g., the intention to relocate or not to relocate to the European Union (EU)) by assessing the outcome utility for each option. In line with the flexible choice approach, they assume that the outcome of this assessment is influenced by the characteristics of the options themselves (e. g., the benefits and costs of relocating in a specific Brexit scenario). In addition, they account for the personal characteristics of the decision-maker (e. g., having family and friends in the United Kingdom (UK)). Finally, they also consider the current emotional state of the decision-makers (e. g., feeling moody as it’s rainy in London), and the expected emotional outcome of their decision (e. g., expecting to feel happy after relocating to sunny Barcelona). While we do not account for emotions and affective reactions directly, we assume that if they played a predominate role for business managers’ relocation intentions, the effects of the more tangible factors on business managers’ relocation intentions would be low and insignificant.

Conceptualising territorial embeddedness

Engaging with the relational approach, we regard business managers’ relocation decisions as being embedded in context-specific structures of social and economic relations (Bathelt and Glückler, 2003). In this regard, we consider social and economic relations as the glue that binds enterprises within local economies (Grabher, 1993), leading to increased territorial embeddedness. We define territorial embeddedness as the degree of which organizations are geographically anchored in a given location, based on the nature of their relationships with other local economic actors and institutions. As such, we expect that high territorial embeddedness in a local economy limits the relocation options available to business managers and, as such, reduces managers’ intentions to relocate. Consequently, we apply Arya and Lin’s (2007) extended resource-based view, focusing not only on the effect of internal resources, such as physical, human, organisational, and financial capital, but also on the effect of territorial embeddedness. Referring to Carrincazeaux and Coris (2015), we do not only consider relationships with other firms, but also relationships with employees, customers, and financers as integral part of territorial embeddedness and an important driver of firm relocation. In doing so, we follow Knoben (2011) who shows that the distance firms are willing to move is positively associated with low (high) level of territorial embeddedness in their home (host) country, as well as with high internal resources endowment.

First, as established companies have a larger stock of physical capital than younger companies, we assume that the costs for an established company to move tend to be greater than for a younger company (Brouwer et al., 2004). Consequently, we expect startups to be more likely than established companies to consider international integral relocation (Storper and Walker, 1989). However, at the same time, we assume established companies to be more likely to consider international partial relocation than startups, as they are more likely to operate business units abroad already. Due to their specific organisational structure, established firms have greater experience in launching and running business units abroad, and exhibit a greater degree of territorial embeddedness in alternative locations, enhancing their capability and confidence to partially relocate (Stam, 2007).

Second, following Carrincazeaux and Coris (2015), we consider buyer-supplier relationships as an integral part of organisational capital, affecting managers’ international relocation intentions. From a relational perspective, we expect managers of companies that depend on international buyer-supplier relationships to be more likely to consider international relocation than managers of companies that do not depend on such relationships, as they there are less strongly embedded in the UK (Knoben, 2011). Moreover, we assume buyer-supplier relationships with financial services in the EU to amplify the business managers’ relocation intentions, due to the passporting-rights in the realm of financial services being put at risk by Brexit.

Figure 1:

Conceptual Framework

Source: Own illustration.

Third, we consider the composition of the workforce as an important part of human capital, affecting managers’ relocation intentions. Employing an international workforce would likely increase managers’ ability to consider international relocation, particularly if their employees are nationals of the considered relocation destination (Thite et al., 2009), due to their reduced embeddedness in the UK. Moreover, we expect a strong dependence on EU nationals to amplify business manager’s relocation intentions, due to their staff’s right-to-work being put at risk by Brexit.

Fourth, we see financial capital as an important factor affecting managers’ ability to consider international relocation, as it provides resources needed for such relocation. In this regard, we consider dependencies on different sources of finance, such as bank loans and venture capital, assuming that dependency on the former is a sign of high embeddedness in the UK, while the latter is a sign of low embeddedness in the UK (Knoben, 2011). Moreover, we assume a strong dependency on funding from the European Investment Fund to amplify business managers’ relocation intentions, due to access to EU funding being threatened by Brexit.

3.1 Empirical Framework

Survey participants were asked how likely the three Brexit scenarios would lead to the integral (partial) relocation of (some functions of) the UK business unit to the EU, or the US. The likelihood of choosing a specific relocation strategy was measured on an ordinal scale, ranging from “very unlikely” (1), through “unlikely” (2) and “likely” (3), to “very likely” (4). This led to the creation of four dependent variables, based on the combinations of integral and partial relocations and the two relocation destinations.

As shown in the conceptual framework, we distinguish between two groups of independent variables: those at the scenario level and those at the company level (see Table 1 for descriptive statistics). At the scenario level, we included two dichotomous variables in the regression model, representing the “Deal” and “No-Deal” Brexit scenarios, with the “No Brexit” scenario used as the reference category. We also included five dichotomous variables representing the participants’ perception of the scenarios, focusing on five domains of the entrepreneurial ecosystem: domestic demand, access to foreign markets, talent, and finance, as well as government support. Participants were asked whether they believe that these domains would be positively or negatively affected by a specific Brexit scenario or not at all. A value of 1 captures belief in a negative impact; while a value of 0 captures belief in a positive impact or no impact. At the company level, we included company age, distinguishing between start-ups (younger than 3.5 years represented by value 1) and established companies (older than 3.5 years – value 0). 36 % of the companies in our sample were start-ups and 64 % established companies. We also considered the share of non-UK nationals employed by the company, distinguishing between the % of EU nationals and the % of non-EU nationals. 55 % of the employees working in the UK establishments of the surveyed companies were British nationals, with 24 % from the EU, and 21 % from non-EU countries. In addition, the perceived importance of B2B relationships with foreign technology companies and foreign financial services companies was captured by four independent variables: the importance of the technology industry located in the EU, the technology industry located outside the EU, the financial services industry located in the EU, and the financial services industry located outside the EU. These variables were measured on an ordinal scale, ranging from “very unimportant” (1), through “unimportant” (2) and “important” (3), to “very important” (4). On average, the EU financial services industry (3.13) and the EU technology industry (3.11) were seen as slightly more important than the technology industry (3.03) and financial services industry (3.00) located outside the EU. The perceived importance of different sources of finance was captured by five independent variables: the importance of equity capital, bank loans, venture capital, UK government funding, and EU funding (also measured on an ordinal scale, ranging from “very unimportant” to “very important”). On average, venture capital was seen as the most important funding source (2.67), followed by equity capital (2.52), funding from the UK government (2.13), bank loans (2.12), and funding from the EU (2.08). Finally, we considered the origin of the company representatives, a value of 0 representing people born in the UK, and 1 representing those born outside the UK. 45 % of them were born outside the UK.

3.2 The Regression Model

Due to the structure of the data, multilevel modelling was used to simultaneously estimate the effect of different Brexit scenarios as well as company characteristics on the likelihood of integral (partial) relocation of (some functions of) the UK business unit. In this approach, the Brexit scenarios are treated as the lower level of analysis, while the 38 surveyed company representatives are treated as the higher level (Oll et al. 2016). As each business manager was asked to judge three different Brexit scenarios (one business manager only judged two of the three scenarios), the dataset consists of 113 observations. As our dependent variables are categorical and ordered, two-level mixed effects ordered logistic regressions were used to estimate the effects of the independent variables on dependent variables:

(1) 

represents the estimated cumulative logarithmic probability (log odds) of scenario i in company j of being less than or equal to a specific category k. Here, k represents the four ordinal outcomes of the dependent variable, representing the likelihood of anticipating to fully (partially) relocate (some functions of) the UK business unit to the EU or the US. In terms of coefficients, β_k0 represents the constant term of the regression; β_p the coefficients of the independent variables representing the scenario characteristics (X_pij); and β_q the coefficients of the company characteristics (X_qij). The error terms are n_j (company level), and ε_ij (scenario level). To ensure that the assumptions of ordinal logistic regressions are met, we tested for multicollinearity between the independent variables. With an VIF of 2.86, this was ruled out.

4.1 Analysing Relocation Intentions

While the descriptive results suggest a significant impact of the Brexit scenarios on the anticipated likelihood of integral (partial) relocation of (some functions of) UK business units to the EU, they are not sufficient to conclude that this effect remains significant when taking additional company characteristics into account. To examine such effects, we developed a range of regression models following a stepwise approach, in which variables are introduced in succession.

In the first step, we estimated the intercept models (m0), which contain no explanatory variables and divide the variance into the two independent components related to the scenario and the company levels:

(2) 

Then we calculated the intra-cluster correlations (ICC), which indicate the estimated proportion of the total variance attributed to the two levels, using the following equations with σ2_i representing the residual variance at the scenario level, and σ2_j at the company level:

(3) ;

The estimated ICCs yield two interesting results. Firstly, they show that 34 % of the variation in the likelihood of anticipating partially relocating the UK business unit to the EU can be explained by differences among companies, while the remaining 66 % can be attributed to difference among Brexit scenarios. In contrast, 63 % of the variation in the likelihood of anticipating partially relocating the UK business unit to the US can be explained by differences among companies, while the remaining 37 % can be attributed to differences among Brexit scenarios. Secondly, our results show that 51 % of the variation in the likelihood of anticipating fully relocating the UK business unit to the EU can be explained by differences among companies, while the remaining 49 % can be attributed to differences among Brexit scenarios. In contrast, 82 % of the variation in the likelihood of anticipating fully relocating the UK business unit to the US can be explained by differences among companies, while only 18 % can be attributed to differences among Brexit scenarios.

Key Finding 2: Anticipated relocation to the EU is more strongly driven by the Brexit outcome than anticipated relocation to the US. Anticipated partial relocation is more strongly driven by the Brexit outcome than anticipated integral relocation.

In the second step, we added two dichotomous variables to the regression models, representing the “No Deal” and “Deal” Brexit scenarios, while treating the “No Brexit” scenario as the reference scenario (m1), aiming to compare the specific effects of the different Brexit outcomes:

(4) 

The results confirm that both the “No Deal” and the “Deal” Brexit scenario had a significant positive effect on the likelihood of anticipating to fully and partially relocate the UK business unit to the EU and the US, with anticipated relocation to the EU being more strongly driven by the “No Deal” Brexit scenario than by the “Deal” Brexit scenario.

In the third step, we added company characteristics to the regression models (m2) to analyse whether factors commonly known as driving actual relocation behaviour also have a significant effect on relocation intentions of business managers:

(5) 

The results suggest that the effects of the “No Deal” Brexit and “Deal” Brexit scenario remain roughly the same, indicating the robustness of the estimated effects. Moreover, the results offer valuable insights into the general (scenario-independent) effects of different company characteristics on business managers’ relocation intentions. Unfortunately, due to the relatively small number of participants that anticipated to fully relocate the UK business unit to the US, it was not possible to estimate reliable effects for the anticipated likelihood of integral relocation to the US (prob>chi2=0.4168). Therefore, we cannot rule out that the estimated effects, despite being significant, are random. To guarantee reliability and comparability, the following discussion focuses solely on anticipated partial relocation to the EU and US (see Tables 2 and 3). The regression tables for anticipated integral relocation can be found in the appendix (see Appendix a and Appendix b). 

In line with the extended resource-based view of the firm (Arya and Lin, 2007), our results suggest that company characteristics such as the age of the company, the origin of the workforce, buyer/supplier relationships with companies located abroad, and sources of finance have a significant effect on business managers’ intentions to partially relocate the UK business unit to the EU and the US. For instance, our results suggest that managers of start-ups are significantly less likely to anticipate to partially relocate the UK business unit to the US than managers of established companies, presumably due to their company’s weaker endowment with internal resources. In addition, our results indicate that business managers that perceive B2B relationships with the financial services industry outside the UK as important are significantly more likely to anticipate to partially relocate the UK business unit to the US. Moreover, while business managers that perceive UK-based bank loans as an important source of finance are significantly less likely to anticipate to partially relocate the UK business unit to the EU and the US, business managers that perceive global venture capital and EU funding as important sources of funding are significantly more likely to anticipate to partially relocate the UK business unit to the EU and the US.

In the fourth step, we extended the analysis by adding five dichotomous variables, representing business managers’ perception of the potential consequences of different Brexit scenarios (m3). This allows us to respond to Anderson and Wilson (2018) by analysing whether the perception of potential future consequences of different Brexit scenarios drive business managers’ anticipated likelihood to partially relocate their UK business unit to the EU and the US. First, the results validate our decision to focus on anticipated partial relocation, as the regression models estimating anticipated integral relocation do not lead to reliable results (see Appendix a and Appendix b: integral relocation to the EU: prob>chi2=0.1955; integral relocation to the US: prob>chi2=0.8687). Second, the results indicate that anticipated partial relocation to the EU is driven solely by concerns about worsening domestic demand, while anticipated partial relocation to the US is driven by concerns about worsening domestic demand, worsening access to international markets, and worsening government support. Interestingly, the Brexit scenario effects on anticipated relocation to the US lose their significance when taking participants’ perception of the potential consequences of different Brexit scenarios into account, while the Brexit scenario effects on anticipated relocation to the EU remain significant despite adding the perception of potential consequences.

Key Finding 3: The impacts of the Brexit scenarios on anticipated partial relocation to the US can be clearly attributed to participants’ perception of the potential consequences of different Brexit scenarios. In contrast, while the Brexit scenarios have a strong effect on anticipated partial relocation to the EU, most of the potential underlying causes do not, suggesting that there might be other, less tangible, reasons, as to why anticipated partial relocation to the EU seems so strongly driven by the Brexit outcome.

In the fifth step, we added cross-level interaction effects to the regression models to analyse whether the Brexit scenario effects on anticipated partial relocation to the EU (m4) and US (m5) vary among companies with different internal characteristics. To limit the complexity of the models, we decided to add each interaction effect separately:

(6) 

The results in Table 4 indicate significant interaction effects, suggesting that certain company characteristics amplify the effects of the Brexit scenarios on anticipated partial relocation to the EU. First, the positive effect of the “No Deal” Brexit scenario appears to be significantly stronger for established companies than for start-ups. Second, it seems that the positive effect of the “No Deal” Brexit scenario is significantly stronger for companies employing a larger share of EU nationals. Third, the results suggest that the positive effects of the “No Deal” Brexit and “Deal” Brexit scenarios are significantly stronger for companies that perceive relationships with the technology industry outside the UK (both inside EU and outside the EU) as more important than for companies that perceive such relationships as less important. In addition, it appears that the positive effect of the “No Deal” Brexit scenario is significantly stronger for companies that perceive relationships with financial services located in the EU as more important than for companies that perceive such relationships as less important. Fourth, the results indicate that the positive effects of the “No Deal” Brexit and “Deal” Brexit scenario are significantly stronger for companies that perceive venture capital as more important than for companies that perceive it as less important. In addition, it appears that the positive effect of the “No Deal” Brexit scenario is significantly stronger for companies that perceive EU funding as more important than for companies that perceive it as less important. Fifth, the positive effects of the “No Deal” Brexit and “Deal” Brexit scenario appears to be stronger for companies run by non-UK natives than for companies run by UK natives.

Finally, the results reported in Appendix c. suggest that some of the interaction effects (such as the importance of relationships with the technology industry outside the UK, the importance of relationships with financial services in the EU, the importance of EU funding, and the origin of the company representative) might also influence anticipated relocation to the US. However, the insignificant “prob>chi” values indicate the low robustness of the model, making us treat these effects with caution.

Key Finding 4: Being embedded in social and economic relations that are reaching beyond the UK amplifies business managers’ reactions to Brexit, increasing their likelihood to consider moving their company (or parts thereof) abroad.

4.2 Comparing Relocation Intentions and Relocation Behaviour

The findings presented above are in line with previous research on actual relocation behaviour. As such, at the first glance, one could question the added value of such results. However, our results highlight that business managers’ relocation intentions are not arbitrary but driven by factors similar to those known as driving actual relocation behaviour. Nevertheless, despite being driven by similar factors, relocation intentions do not necessarily have to translate into actual moves, particularly in times of high uncertainty. To further evaluate the feasibility of using factorial surveys to anticipate strategic behaviour under such conditions, we compare business managers’ relocation intentions with their with their companies’ actual relocation outcomes.

According to information publicly available on LinkedIn, Companies House, Crunchbase, and the webpages of the companies under consideration, five (13 %) of the 38 companies under consideration had fully dissolved their UK establishments by the end of February 2022. Three out of the five establishments that were fully dissolved where small sales offices of foreign multinational companies headquartered in Israel and the US. Since closing their branches in the UK, they have shifted their focus back to their home markets. Back in 2018, business representatives of all three anticipated domestic demand in the UK to worsen due to Brexit. As such, it can be assumed that they had established a branch in the UK to explore the market and decided to end this endeavour. In contrast, two out of the five establishments that were fully dissolved were UK-based companies founded and managed by Europeans who went back to their home country – Belgium and Romania – after Brexit. Interestingly, the Romanian founder started a similar company in Romania, offering the same services under a different company name. Moreover, the Belgian founder still describes himself as the “active CEO” of the company on LinkedIn. As such it can be assumed that, while the company is currently dormant, it might be re-opened in Belgium in the future. In line with the results of the factorial survey analysis, these findings indicate that the national background of the founders matters, with founders of European origin being more likely to relocate to the EU. All the UK establishments considered thus far were small, employing fewer than five employees in the UK in 2018. As such, the findings suggest that size matters as well with regards to full relocation, with smaller establishments being more likely to be closed than larger ones.

Furthermore, 15 (39 %) of the 38 companies under consideration had opened new offices overseas by February 2022 without closing their UK establishment. While the publicly available data does not allow for a systematic distinction between partial relocation and international expansion, as well as a systematic identification of the motives behind opening a new office abroad, the data provides, at least, anecdotal evidence. For instance, one of the companies under consideration stated on their webpage that they moved their headquarters from London to Sofia as a response to declining growth opportunities in the UK. Another stated that they had to open an additional office in Malta to keep the EEA passport. In line with the results of the factorial survey analysis, these findings indicate the importance of concerns about worsening domestic demand in the aftermath of Brexit, as well as the importance of maintaining links with the financial services industry in the EU. Interestingly, the data indicates that companies under consideration opened offices in the EU’s financial centres, such as Paris, Milan, Amsterdam, Madrid, and Brussels or moved to European cities more strongly regarded as technology hubs, such as Barcelona, Berlin, Kraków, Bucharest, Sofia, and Kyiv. As such, instead of significant concentration in specific European cities, our results suggest a more dispersed relocation pattern. In contrast, partial relocation/expansion to the US mainly led to New York. In addition, some of the companies under consideration opened offices outside of the EU and the US in cities including Auckland, Melbourne, and Singapore.

To analyse the intention-behaviour gap by comparing business managers’ relocation intentions with their companies’ actual relocation outcomes, we use two distinctive classification approaches to judge the accuracy of business managers’ predictions: a conservative and a more permissive one. In the conservative approach, we categorize a business manager’s prediction as incorrect if their companies’ actual relocation outcome deviates from their relocation intention in any, even the slightest, respect. For instance, we would categorize business managers’ predictions as incorrect if they anticipated to partially relocate to the EU, but their company had fully relocated to the EU. In contrast, in the permissive approach, we categorize a business manager’s prediction as correct if they correctly anticipated to relocate irrespective of the extent and destination of the relocation. For instance, we would categorize business managers’ predictions as correct if they anticipated to partially relocate to the US, but their company fully relocated to the EU.

Depending on the classification approach, between 66 % (conservative approach) and 84 % (permissive approach) of the 38 companies under consideration behaved as their business managers anticipated, suggesting a moderate to high degree of correlation between relocation intentions and actual relocation outcomes. Based on the more permissive classification, we had a closer look at the six companies that did not behave as their business manager anticipated. Five of them are companies whose business managers did not indicate that they would relocate/expand at all but did so in the end. While one of these companies was dissolved in the UK and re-opened in Romania, the majority opened additional offices abroad while maintaining their UK establishments. Given the dynamic development of the global FinTech industry, the latter group may have been encouraged by their growth and expanded earlier than their business representative expected in 2018. However, it is also possible that the business representatives’ predictions were not based on adequate information or arbitrary. The latter might, indeed, be the case, since the share of business representatives without managerial responsibility is significantly larger in this specific sub-sample (60 %) than in the overall sample (16 %). Interestingly, only one company did not relocate/expand despite its CEO anticipating that it would be partially relocated in a “Deal” Brexit and “No Deal” Brexit scenario. This company is headquartered in Barcelona and, in 2018, had offices in London and Minsk. Given their presence in the EU, there wasn’t a pressing need to open another office in response to Brexit. However, while not publicly announced, it is possible that some business functions have been shifted from London to Barcelona.

Key Finding 5: Although factorial surveys are only moderately accurate when predicting the exact extent and destination of actual relocations, they are highly accurate when predicting whether a company relocates or not.