Global cities as a location: The effect on firm financial performance

(1)

1

Global cities as a location: The effect on firm financial performance

Paula Boneschansker 13415948

28 January 2022

Master Business Administration – International Business Track University of Amsterdam

20220112020109

Supervisor: Katiuscia Lavoratori

(2)

2

Statement of originality

This document is written by Paula Boneschansker (13415948) who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original

and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and

Business is responsible solely for the supervision of completion of the

work, not for the contents

(3)

3

Tables

Table 1 Origin of company – continent ... 26

Table 2 Overview of variables ... 30

Table 3 Independent Sample t-test ... 35

Table 4 Model summary AV_ROA (Gc only) ... 36

Table 5 Coefficient AV_ROA (Gc only) ... 36

Table 6 Model summary AV_ROA ... 36

Table 7 Coefficient AV_ROA ... 37

Table 8 Model summary AV_ROS (Gc only) ... 37

Table 9 Coefficients AV_ROS (Gc only) ... 38

Table 10 Model Summary AV_ROS ... 38

Table 11 Coefficients AV_ROS ... 39

Table 12 Cross-tabulation D_contin ... 40

Table 13 Pearson Chi Square D_contin ... 41

Table 14 Interaction model AV_ROA ... 42

Table 15 Specification interaction model AV_ROA ... 42

Table 16 Interaction model AV_ROS ... 43

Table 17 Interaction model AV_ROA ... 44

Table 18 Interaction model Av_ROS ... 45

Table 19 Specification interaction model AV_ROS ... 45

Table 20 Descriptive statistics before computing ... 62

Table 21 Descriptive statistics after computing ... 62

Table 22 Global cities ... 63

Table 23 Continent difference ... 63

Table 24 Food vs non-food ... 63

Table 25 Digital nature ... 63

Table 26 Correlations ... 64

Figures

Figure 1 Conceptual Model ... 24

Figure 2 Global cities ... 27

(6)

6

Figure 3 Global cities / Region ... 41

Figure 4 Histogram AV_ROA ... 61

Figure 5 Histogram AV_ROS ... 61

Figure 6 Histogram control variables ... 61

(7)

7

Abstract

Multinational enterprises (MNEs) constantly search for possibilities to maximize their profits and location choice seems to have an impact on eventual performance. This research focuses on the choice of MNEs to locate within global cities and how this has an effect on firm financial performance. Moreover, the moderating effect of being located within the home region and the effect of being a food (vs. non-food) retailer was tested. Using a sample of 421 subsidiaries over a period of three years (2019, 2018, 2017) findings show a non-significant effect of global cities on firm financial performance. However, two moderating effects were identified, namely a negative moderating effect of being located outside the home region of the parent company and a positive moderating effect of being a non-food retailer. This work helps managers to better understand different factors related to location choices and their effect on firm financial performance and thus make more informed decisions.

(8)

8

Chapter 1: Introduction

‘’The three most important things in retail are location, location, location.’’ – Jeff Bezos (2003)

The location choices of multinational enterprises (MNEs) have been a hot topic in recent years among international business (IB) scholars. Location choice can be based on the motivation to seek resources, market, strategic-assets or efficiency (Dunning, 1998) or can differ per function of the subsidiary. Whereas retailers are often seeking markets (Goerzen et al., 2013), R&D facilities are often seeking for knowledge to play catch-up or diversify their offer (Demirbag &

Glaiser, 2010). In essence, the decision about where to locate is decided based on the benefits outweighing the costs of doing business. MNEs aim to maximize their profits by benefiting from the imperfect markets around the world, which makes location choice very important (Buckley & Casson, 1976).

Location choice often takes a country-level approach, which places a special emphasis on national borders and the heterogeneity of the country within these borders (Beugelsdijk &

Mudambi, 2013) for example by comparing GDP levels or the Hofstede dimensions per country (Hofstede, 2001). Variation, for example within culture, institutions or geography, used to be overlooked within IB research. However, in recent years there has been a growing focus on sub-national location choices for MNEs (Beugelsdijk & Mudambi, 2013). Sub-national locations are locations below a national level, for example cities or regions (Collins English Dictionary, 2014). A study by Hutzschenreuter, Matt & Kleindienst (2020) suggests that there is heterogeneity in administrative, economic and cultural perspectives among regions within the same countries. In addition, regions in different countries within Europe are sometimes seen as more similar by MNEs when comparing them to regions within their own country (Basile, Catellani & Zanfei, 2009). Both arguments suggest that variation exists within borders as well.

Global cities are an example of a location that shares similarities across borders and are often chosen as a specific location for MNEs to locate in. Examples of global cities include New York, London, Hong Kong and Singapore (GaWC, 2020). Despite global cities being located in different countries, Goerzen, Asmussen & Nielsen (2013) identified three characteristics that are similar across this type of cities. First, a high level of interconnectedness which includes the easy transfer of capital and information due to infrastructure and transportation possibilities. Second, the presence of APS businesses which include companies specialized in helping MNEs. Lastly, a cosmopolitan environment which includes a diversity of cultures (Goerzen et al., 2013).

(9)

9 As stated before, location is a key element to retrieve eventual performance (Chakravarty, Goerzen, Musteen & Ahsan, 2021). The characteristics of global cities tend to be beneficial for MNEs as they can help to overcome the Liability of Foreignness (LOF) or lower factors of distance (e.g. CAGE) which is faced when going abroad (Belderbos, Du & Goerzen, 2017; Goerzen et al., 2013). To illustrate, the interconnectedness of the city can reduce spatial distant costs (Belderbos et al., 2017) and the availability of APS businesses can support with reducing coordination costs and increase familiarity of the MNE within the given location (Castellani, Lavoratori, Perri & Scalera, 2021; Asmussen, Nielsen, Dohlmann Weatherall &

Lyngemark, 2019). Additionally, the cosmopolitan character can reduce the discrimination foreign firms might face (Goerzen et al., 2013). The lower levels of LOF indicate more chance to succeed and perform well in foreign countries (Zaheer, 1995).

An industry that might benefit from lower levels of LOF is the retail industry. MNEs within this industry often locate within global cities due to their cosmopolitan character (Asmussen & Goerzen, 2013) or because they seek new markets (Chakravarty et al., 2021).

However, this industry also has been hesitant to enter more distant markets (Asmussen &

Goerzen, 2013) and often face discrimination from local inhabitants (Maruyama & Wu, 2015).

Global cities might help with these hesitations and disadvantages of internationalization.

This thesis aims to fill the gap in the literature and elaborate on subsidiary performance within the retail (both food and non-food) industry. This industry might especially benefit from the lower levels of LOF within global cities. Therefore, it will contribute to the understanding of the profitability of subsidiaries of MNEs within global cities and why this profitability differs compared to other locations. Moreover, it will contribute to the existing literature by expanding on various reasons why MNEs might choose global city locations even if this does not translate into short-term profits in the first place. For managers, more insights about location and its effect on performance will help to make more informed decisions regarding location management. The following research question will be answered: ‘’Does being located in global cities affect MNE subsidiary performance within the retail industry?’’

The thesis will be structured as follows: First, previous literature in the field of IB, performance, location-management and global cities is presented. This literature review will place a special emphasis on the retail industry. After careful consideration of the literature, a research gap is identified which is the basis of the theoretical framework. Chapter three includes the theoretical framework and the hypotheses which have been researched. After, chapter four highlights how these hypotheses are researched, primarily focussing on the data and the independent, dependent, moderation and control variables. Chapter five indicates the results

(10)

10 based on the study. Lastly, a discussion is presented including the results of the study, the limitations, suggestions for future research and the managerial contributions. In the discussion, the results are further elaborated on with the help of previous explained literature.

(11)

11

Chapter 2: Literature review

In this chapter, various areas of literature are presented to provide a basis for the hypothesis creation. This literature review extends on various key and new theories with an emphasis on the retail industry. First, the literature focussing on the origin of multinationals and subsidiaries is presented. Second, more information is provided about strategy, context and institutions in relation to performance. Third, the basis of location choice and performance is elaborated on.

The chapter ends with the literature about global cities as a location choice for MNEs.

2.1 MNE and subsidiary performance 2.1.1 The concept of MNE and subsidiaries

A multinational enterprise is an enterprise that operates in more than one country, for example by delivering services or producing goods (Eurostat, 2019). MNEs coordinate and control the different activities via their subsidiaries (Birkinshaw, Hood & Jonsson, 1988). According to Buckley & Casson (1976), MNEs exist due to market imperfections which lead companies to internalize activities across borders to operate more efficiently. In general, this decision rests on lower transaction costs which come from information asymmetry, bounded rationality, opportunistic behaviour, uncertainty and risk (Buckley & Casson, 1976; Narula, Asmussen, Chi

& Kumar Kundu, 2019).

Internalisation theory has been developed further by Rugman (1981) and Rugman &

Verbeke (1992) with a focus on country-specific advantages (CSA) and firm-specific advantages (FSA). CSA focuses on the advantages which are linked to a specific location, whereas FSA places an emphasis on resources of the firm and the ability to leverage them worldwide (non-location-bound) or in a specific location (location-bound). FSA include stand- alone concepts, routines and the recombination of capabilities (Rugman, Verbeke & Nguyen, 2011). Whereas location-bound advantages are exploited in certain regions, non-location-bound advantages are easy to implement in the subsidiary network of the MNE.

This subsidiary network is the basis to control, coordinate or even create FSA in the host country, therefore, playing a large role in the success and the progression of the company by for example generating sustainable competitive advantage (Birkinshaw et al., 1998). Based on the origin of the company, the pressure to act globally and/or the pressure to act locally responsive, all subsidiaries take on different roles (Bartlett & Ghosal, 1989). For subsidiaries with high levels of global integration, such as global and international companies, subsidiaries

(12)

12 often act as pipelines of the HQ. Examples include Pfizer or Apple. On the other hand, subsidiaries can also have high levels of local responsiveness such as multi-domestic and transnational companies. Examples include retailers, or FMCG companies such as Unilever which need to diversify their products. The subsidiaries of multi-domestic companies often act autonomously to fit with the local demand. Transnational companies try to combine global standards with local standards, therefore combining CSA with FSA (Bartlett & Ghosal, 1989;

Rugman & Verbeke, 2011).

Subsidiaries often work as a bridge between the global parent company and the local environment (Meyer, Li & Schotter, 2020). National responsiveness has become the new standard which makes it even more important to develop location-bound advantages (Rugman

& Verbeke, 1992). Subsidiaries also play a role in overcoming LOF, can affect the overall image of the parent company and can help with new knowledge generation (Moeller, Harvey, Griffith & Richey, 2013; Rugman et al., 2011). Therefore, subsidiaries also have a role in the performance of MNEs, which will be further elaborated on in the next section.

2.1.2 Strategy and performance

Performance of MNE and subsidiaries has been widely studied in IB, taking on factors as the extent of internationalisation, resources of the company itself and its environment. In this section, performance of both the MNE and the subsidiary is discussed as combining the multi- level interactions with the local subsidiary is necessary to understand performance of the network of MNEs (Meyer et al., 2020). MNEs and subsidiaries also face LOF which they have to overcome to operate as successful as possible in a foreign country.

First, the extent of internationalisation of a company. Internationalisation creates certain benefits for MNEs such as the availability of new consumers (Dimitrova, Rosenbloom, Andras

& Kim, 2018), the distribution of overhead costs and the creation of a larger global scope (Contractor, Kundu & Hsu, 2003). However, the influence of expansion to other countries and performance has been disputable. Whereas most authors find a positive linear relationship between global expansion and performance (Han, Lee & Suk, 1998; Grant, 1987), some authors find a U-shaped relationship, predicting a negative relationship in earlier stages before the positive relationship, an inverted U-shaped relationship which suggests a downhill slope after some period of time (Contractor et al., 2003) or even no relationship between the variables.

Hennart (2007) argues that these contradicting findings might suggest that there is no relationship between internationalization and performance at all. MNEs often internationalize to increase performance and lower risk due to portfolio diversification. However,

(13)

13 internationalization is not always necessary to achieve higher performance as home markets might create the economies of scale and scope the MNEs are looking for in the first place.

Furthermore, costs and risks arising from CAGE distances might lower performance (Hennart, 2007). Other scholars, such as Contractor, Kunda & Hsu (2018) researched the service industry and suggested a three-tier staged model in which companies first show a negative slope when expanding abroad, then a positive slope and lastly, when overexpanding, a negative slope again.

In the first stage, companies deal with high initial investment costs and potential LOF. In the second stage, MNEs have the possibility to leverage the benefits of international expansion by the control of differences among companies and engage in price discrimination. Other studies found that LOF decreases in this stage, as companies gain more international experience (Lu &

Beamish, 2004). In the last stage, performance goes down when companies overexpand which causes higher incremental costs than income (Contractor et al., 2018; Lu & Beamish, 2004).

Moreover, the pace, rhythm and speed of the internationalization process of an MNE seem to play a role in performance. A wide product range to internationalize, a fast expansion to different countries and an asymmetrical rhythm moderates the assumed positive relationship between internationalization and performance (Vermeulen & Barkema, 2002). However, competition and the context a company is expanding to needs to be taken into account as well.

Companies that operate in a market with high globalizing levels tend to benefit from rapid expansion as competitors might internationalize rapidly (Chang & Rhee, 2011). To extend, the stronger the capabilities and resources of the firm, the better rapid expansion might go (ibid).

Another factor within performance success when expanding abroad comes from Yang &

Driffield (2012) who suggest that non-US companies benefit from internationalisation more by showing higher returns compared to US-based firms. It can be stated that in most cases and stages of internationalisation, the business itself and the business context seem to play a role (Johanson & Valne, 1990).

The company itself has physical, human and organisational resources. The creation of unique capabilities and resources, therefore the related FSAs, is necessary to distinguish from the competition. This view is complemented by the resource-based view which focuses on valuable, rare, in-imitable and organized resources. According to Barney (1991), these resources help to increase efficiency and effectiveness, which in turn lay the basis for the formation of sustainable competitive advantage and better performance. For subsidiaries specifically, the intangible resources of a MNE, such as intellectual resources, production processes and the brand name contribute to a better subsidiary performance. However,

(14)

14 subsidiaries do show differences in performance which in some cases is caused by contextual differences (Contractor, Yang & Gaur, 2016).

The context of the company includes, among others, the institution and the culture.

Every location shows certain variations within these aspects. How an MNE and its subsidiaries reacts to such variations partly explains performance differences (Galli Geleitate, Andrews &

Fainshmidt, 2019). Contextual differences are widely acknowledged in IB literature. Whereas Hofstede (2001) describes cultural differences by the hand of dimensions, others take a more personal approach or describe cultural differences as differences in shared knowledge (Chiu &

Hong, 2007). Companies that enter international markets must take the cultural environment into consideration (Drogendijk & Slangen, 2006). Context also includes the institutions.

Institutional differences are acknowledged by authors as North (1991), which focuses on the social, political and economic constraints which shape human life. Institutions become increasingly important due to globalization and the willingness of MNEs to manage uncertainty while maintaining open business networks (Cantwell, Dunning & Landan, 2010). Furthermore, the more internationalized an MNE becomes, the more difficult it is to manage the different institutions (ibid).

Context evidentially also has an influence on performance. For example, culture and cultural distance can affect the eventual performance of firms such as potential market growth and sales levels (Gomez-Meijia & Palich, 1997). But also, differences in language, culture, industrial development and education increases the distance. From a transaction cost perspective, the increase in distance causes more difficulties in coordination, communication and monitoring which causes higher costs and lower performance of subsidiaries in those locations (Hutzscenreuter et al., 2020). MNEs often try to find ways to cope, adapt or co-evolve with the institutions in a given country. The latter includes both the adaption as the introduction of new standards within the institution (Cantwell et al., 2010). This, among other arguments presented above, creates subsidiaries within the same MNE to show differences in performance as well as differences in strategies. To cope with these variances and often complex situations, numerous subsidiaries show autonomous behaviour (Galli Geleitate et al., 2019). These reactions are a consequence of the dynamic interaction between the organisation itself and the contextual differences (Peng, 2002).

This interaction is clearly shown when MNEs and subsidiaries act on the international playing field and try to overcome LOF. MNEs face LOF compared to local firms. LOF has

(15)

15 three main elements. First, complexity which includes the spatial distance. Second, the uncertainty includes the lack of familiarity with the local environment. Third, discrimination refers to the discrimination towards foreign firms. These features make it harder for MNEs to do business in foreign countries, which eventually will cause differences in performance (Zaheer, 1995). To overcome LOF, MNEs and subsidiaries often adjust their strategy based on their resources and the demands of the environment. For the retail industry this is the most equivalent as retail is a downstream industry that delivers products to the end-consumer.

Retailers face their end-consumer, therefore creating a high degree of social complexity and high demand for location-specific capabilities (Anand & Delios, 1997). Consequently, retailers tend to engage in diversification strategies to meet local demand (Swoboda, Elsner &

Morschett, 2014; Oh, Sohl & Rugman, 2015).

The context and the level of adjustment of strategies can affect differences in performance. Therefore, location choice is an important construct to take in consideration. This will be further elaborated on in the next section.

2.2 Location and performance

The location has been one of the main elements of the eclectic (OLI) model of Dunning (1980), which states that Foreign Direct Investment (FDI) should take place when a country has specific locational advantages. In general, MNEs seek natural resources, market, efficiency or strategic assets when expanding to a new location (Dunning, 1998). Natural resources include the availability of raw materials, but also the opportunity to update the quality of these materials with the help of local partners. Companies with a market-seeking motivation want to serve the local or regional market by serving clusters like the European Union or by creating certain proximity to a country. Efficiency motivated companies are generally seeking lower production costs or skilled labour which are centralized in certain areas. At last, companies that seek for strategic assets are mainly looking for superior skills to update their company (ibid). Other scholars group location choices on the basis of industry. For example, distinctions are made on the basis of the function of the subsidiary such as R&D, manufacturing or sales. R&D locations tend to choose a location to catch up with the competition, diversify their labour pool, to gain new knowledge or generate new researches which will create concerns at the home country (Demirbag & Glaiser, 2010). Manufacturing firms might be more drawn to a location based on physical assets in combination with available knowledge, people and technology (Lavoratori, Mariotti & Piscitello, 2020). Besides, scholars look at location choices on the basis of effect and motivation, such as done in the research by Mariotti, Piscitello & Elia (2010). They argue

(16)

16 that companies are either naturally drawn to a given location, drawn to effects of agglomeration such as specialized knowledge within that location or drawn to a location due to policy-induced effects such as beneficial laws or taxes (Mariotti, Piscitello & Elia, 2010). Agglomeration effects are often the motivation for retailers to locate in a specific location. Retailers tend to cluster due to for example potential knowledge spill-overs of competition (Picone et al., 2008).

These clusters are often found in global cities, such as New York, Singapore or Tokyo. Next to potential knowledge spill-overs, clusters provide stimuli for innovation (Bell,2005), collaboration (Saxenian, 1994) and higher labour productivity (Ciccone & Hall, 1996) which can all be beneficial for retailers. However, these agglomeration effects are not always beneficial for performance. Agglomeration and clusters come with costs such as negative knowledge spill-overs or increased competition. In later stages of the industry life cycle, isolated firms outperform clustered firms within the same industry (Kukalis, 2010). However, on the demand side clusters provide retailers with proximity towards their customers (Picone, Ridley & Zandbergen, 2008; Huang & Levinson, 2011). These new customers are often found in global cities, where a high density of potential, cosmopolitan customers is found (Goerzen et al., 2013). As retailers are generally driven by the saturated markets in their home country which forces them to seek new markets elsewhere, retailers are often drawn to global city locations (Dimitrova et al., 2018). Global cities as a location for these subsidiaries will be further discussed in the next section.

2.3 Global cities as a location

Whereas megacities differentiate themselves on size, global cities have different characteristics (Goerzen et al., 2013). In 1982, John Friedmann & Goetz Wolff (1982) introduced global cities as the key players within the world economy and argued that the economy was organised from them. Global cities have been conceptualized by the use of infrastructure indicators, such as transportation to show the transformation towards global cities or the connectivity with other cities (Smith & Timberlake, 2001; Otiso, Derudder, Bassens, Devriendt & Witlox, 2011). Other perspectives take a more firm-based approach and focus on the network of firms, states, cities and sectors in global cities (Beaverstock, Doel, Hubbrand & Taylor, 2002) or the availability of business services (Dunning & Norman, 1983). Global cities differ in terms of the institutional, economic and cultural domains (Chakravarty et al., 2021). Institutions can evolve differently due to path-dependency (North, 1991). Therefore, to coordinate global efforts, global cities evolved different institutional structures which led MNEs to flourish in them (Toly, 2017). In general, the institutional environment of global cities is found to be stronger and more

(17)

17 stable compared to other location within the host country. Moreover, the institutional environment tends to be more business-friendly (Chakravarty et al., 2021). The informal institutions, and therefore the cultural domain, differs in global cities as citizens are more diverse in for example religion, ethnics and sexuality and therefore show a higher degree of cosmopolitanism, tolerance and respect (Warf, 2015).

Goerzen et al. (2013) characterized global cities by three features, namely the interconnectedness of the city, the presence of APS businesses and the cosmopolitan character.

The interconnectedness of global cities is established due to the high presence of transportation possibilities and a good infrastructure. In addition, these cities have a centralized location in the world economy. This makes it easier for MNEs to move both people and products and capital and information(Goerzen et al., 2013). Second, APS businesses include, among others, legal, administrative and financial services. In general, these services are clustered in global cities (Goerzen et al., 2013). These services provide MNEs with the opportunity to establish local business relationships (Storper & Venables, 2004) and gain advice on strategic decision-making more easily (Pedersen & Tallman, 2016). The third characteristic of a global city is the cosmopolitan character. Cosmopolitanism is found in global cities due to certain developments in social factors such as education and culture (Goerzen et al., 2013). Cosmopolitanism has been linked to the diverse character of the city (Devadason, 2010; Goerzen et al., 2013) and is operationalized as open-mindedness, high international consumption preference and an appreciation for diversity (Riefler, Diamantopoulos & Syguaw, 2011). Moreover, cosmopolitan cities tend to be creative in terms of profession and education and show a high degree of egalitarianism. This is in line with the findings of Warf (2015) which shows that citizens in global, cosmopolitan cities tend to show higher levels of respect and tolerance towards others.

All these characteristics make global cities optimal location for MNEs investing abroad, reducing the costs related to LOF and operating abroad by being located in a highly connected place from an infrastructural perspective and through the global network of APS firms, as well as the connectivity generated by the international networks of inventors (Castellani et al, 2021).

However, different firms can prioritise different features of global cities, due to the sector in with they operate or the value chain activity they are performing abroad, such as HQ, R&D or manufacturing facilities (Belderbos et al., 2017; Castellani et al. 2021; Asmussen, et al, 2019).

Despite different literature pointing towards the beneficial side of global cities, these locations also show downsides for MNEs. First, MNEs located in global cities show lower survival rates due to a high density of competition and extra financial costs (Sassen, 2006).

Competition in global cities often shows excessive communication networks and value chains,

(18)

18 making it more difficult for new entries to survive in these locations (Chakravarty, 2018). Extra financial costs of global cities come from higher property costs per square meter, which causes a repulsive reaction of big production spaces (Goerzen et al., 2013). Another liability of global cities rests in the possibility of knowledge spill-overs, especially for firms with well-established resources such as human capital or value chains. However, firms with less established resources may benefit from these spill-overs (Shaver & Flyer, 2000). Furthermore, retailers often have low margins to cover the costs of doing business (Quix, 2019). As global city locations have the highest rental prices in the world (Statista, 2019), the costs of doing business increase as well which may affect potential survival rates. Likewise, retailers often face a high density of competitors in global city locations, which causes price competition (Picone et al., 2008).

To conclude, it can be stated that global cities both have benefits and disadvantages for retailers. The interconnectedness, availability of APS businesses and the cosmopolitan character are beneficial for retailers and might enhance their performance. On the other hand, locating in global cities causes higher rental and operating costs and increased competition due to clustering.

2.4 Research gap and research question

MNE internationalization and location choice have been important topics within the International Business literature. The structure of the MNE and its subsidiaries, the strategy, why to locate, where to locate and what benefits and disadvantages are combined with these choices have been widely discussed in the literature. For research on global cities specifically, the focus has been on its characteristics. However, the consequential performance of locating in a global city has been overlooked by research to the best of our knowledge. Whereas agglomeration effects and clusters and its relationship with performance have been discussed, global cities and its distinct characteristics are not extensively investigated by scholars. Gaining knowledge about the performance consequence of locating in global cities may help MNEs, and specifically retailers as for the focus of this thesis, to make more informed decisions.

Furthermore, it will help to understand whether the advantages of global cities are applicable for a sector that shows low margins and a dynamic environment. Retailers are often classified as multi-domestic companies, which means they show high levels of local responsiveness. This can result in extensive differences in performance among subsidiaries of the same retailer.

Therefore, it is an interesting industry to focus on when looking at global cities and financial performance.

(19)

19 This research aims to answer the following question: ‘’Does being located in global cities affect MNE subsidiary performance within the retail industry?’’.

(20)

20

Chapter 3: Theoretical Framework

In this chapter, four hypotheses are presented which will be answered in the research. These four hypotheses are based on the research gap presented in the previous chapter and the literature specific to these subjects.

3.1 Hypothesis 1

Firm financial performance is influenced by multiple factors, including culture, context and strategy (Gomez-Meijia & Palich,1997; Contractor, Yang & Gaur, 2016). In an international environment, MNEs need to cope with various circumstances. As retailers often have a multi- domestic nature, meaning they face high pressure of local responsiveness and lower pressure of global integration, subsidiaries often act autonomously to deal with the aspects they face at different locations (Rugman & Verbeke, 2011). Subsidiaries of retailers often choose to develop their own FSA which are only applicable to a certain location (ibid; Birkinshaw et al., 1998).

Therefore, the location of the retail’ subsidiary is extremely important.

Global cities seem to be a valuable location for retailers. Retailers tend to locate in global cities due to their cosmopolitan character (Asmussen et al., 2019) or because of market-seeking purposes (Chakravarty et al., 2021). Retailers tend to internalize late due to the complexity and uncertainty which causes hesitation to enter new markets (Asmussen & Goerzen, 2013;

Asmussen et al., 2019). Expected potential growth of a market and low levels of political and business risk might elevate these late internationalization considerations (Gaba, Pan & Ungson, 2002). Another reason retailers tend to internationalize late is because, MNEs within this industry often face discrimination from local inhabitants (Maruyama & Wu, 2015), which increases their LOF and therefore decreases their initial performance (Zaheer, 1994).

Locating in global cities might elevate these disadvantages of internalizing because of the interconnectedness, presence of APS businesses and cosmopolitan character which they can leverage for better performance. First, cosmopolitanism is linked with more open-mindedness towards foreign MNEs and international consumption (Riefler et al., 2011), which causes lower levels of cultural distance and LOF. Moreover, cosmopolitan cities show high levels of economic development and economic opportunities (Sevincer, Varnum & Kitayama, 2017) for example due to larger amounts of tourists visiting these locations (Goerzen et al., 2013;

Asmussen et al., 2020). A high density of touristic visits in a location has a positive relationship with the performance of retailers (Gholipour, Tajaddini & Andargoli, 2020). This is in line with the potential expected market growth presented by Gaba et al. (2002). The second aspect of

(21)

21 global cities, namely interconnectedness, helps to reduce LOF as well. Within global cities, MNEs can easier obtain local knowledge and can transfer capital, people and information cheaper and at a faster pace (Goerzen et al., 2013, Castellani et al., 2021). Moreover, the characteristics of global cities discharges the effect of large geographical distances (Belderbos et al., 2017). According to transaction costs theory, geographical distance normally increases costs of monitoring, managing and information which will eventually result in higher costs (Hutzschenreuter et al., 2015). As global cities dissimilate these distances, performance is likely to be higher. At last, the presence of APS businesses helps retailers to learn more about the local environment, lowers coordination costs and creates higher credibility for the MNE (Goerzen et al., 2013). Furthermore, successful internationalization needs new and existing recombination of resources which can either be at high costs or through contacts in the markets (Verbeke & Asmussen, 2016). APS businesses can help retailers to create local value chains and relationships with local stakeholders (Rugman, Li & Oh, 2009).

Retailers have small margins to operate with and global MNEs face the constant pressure of increasing efficiency and justifying the increased transactions costs of operating globally. Global cities show characteristics to elevate the disadvantages and higher transaction costs of doing business. Thus, the following hypothesis is created:

H1: Being located in global cities increases firm financial performance.

In a study by Rugman & Verbeke (2004), it was found that 320 out of 380 MNEs have concentrated sales in their home triad (European Union, United States, Japan). They argue that diversification of products is expensive and needs high investments, which leads to MNEs choosing to either standardize the product or sell their products in a regional market (Rugman

& Verbeke, 2004). As retailers need diversification to match with the local demand and have high capital investment costs, it is often chosen to focus on home region markets (Campbell &

Verbeke, 1994). Moreover, MNEs often internationalize via the Uppsala model, to gain market knowledge and international experience which means they are more likely to expand to countries with lower levels of distance (Johanson & Vahlne, 1997). Also for a global city location, distance is important. MNEs tend to choose global cities as a location when contextual distance increases and knowledge about the country decreases (Belderbos et al., 2020).

Contextual distance within regions is likely to be lower, as they have shared history or common rules and regulations similar to the CAGE model presented by Ghemewat in 2001. From a knowledge perspective, global cities tend to have clusters of different MNEs and therefore the

(22)

22 chance of negative knowledge spill-overs (Mariotti, Piscitello & Elia, 2010). As MNEs located within the home region might have more experience with this market, it can result in a preference to be located outside the cluster to reduce possible spill-overs (Rugman, 2007;

Mariotti et al., 2010). It is therefore expected that global city locations are chosen less often when the MNE is internationalizing in the home region.

Due to the above reasoning, it is expected that the benefits of global cities within the home region are less compared to the benefits it brings the MNE when located outside the home region. First of all, the clusters which are present in global cities are beneficial in terms of innovation, collaboration or labour productivity, but might also cause higher levels of competition or negative knowledge spill-overs. MNEs internationalizing within their home region face lower levels of LOF due to lower spatial distance and higher familiarity with the country (Zaheer, 1995; Kudina, 2012). Due to these lower levels, retailers might disregard the benefits due to the disadvantages. Companies outside the cluster tend to have higher financial performance compared to their clustered colleagues (Kudina, 2012). Second, locating within the home region often results in a higher firm financial performance (Sukpanick, 2007). This might result in retailers locating somewhere cheaper. Global cities specifically are expensive in terms of resources, for example rent and personnel (Goerzen et al., 2013). Due to this reasoning, it is argued that the cost of doing business in a global city might outweigh the benefits for retailers which expand within their home region, therefore the effect of being located in a global city is likely to be lower. The following hypothesis is created:

H2: The relationship between the global city location and firm financial performance will be negatively moderated when the subsidiary is located within the home region of the parent company.

When researching retailers, a general distinction is made between food and non-food due to differences in demand, logistics and preferences (Etgar & Moore, 2010; Moore & Batsakis, 2018). The main difference between food and non-food retailers is the level of diversification.

In a 2016 study, findings show that food retailers diversify their product offerings by 7-400%

compared to a rate of 5-65% for mixed retailers (Hart, 2016). Reasons for this difference include the tied connection between food and culture including differences in trends, diets and preferences (Azar, 2011). Food is an example of shared national identity among a group of individuals, which causes obstacles for food retail companies (Ghemewat, 2001; Buisson,

(23)

23 1995). For food retailers, this results in a less likely expansion to other countries which show high variation in food culture (Filippaios & Rama, 2011). Consequently, diversification is necessary which comes with costs including internal costs for monitoring and transferring knowledge internally and external costs which include costs of internationalizing the diversified offering (Shi, Lim, Weitz & France, 2017). Moreover, food retailers face legitimacy issues when expanding abroad (Shi et al., 2017). This is in line with the discrimination argument of LOF which companies face when expanding abroad (Zaheer, 1994).

However, food retailers have been found to generate higher returns when expanding abroad despite above-mentioned statements. It was found that expanding to more distant countries enhances the opportunities for food retailers as the new market is not saturated (Azar, 2014). Deloitte (2007) complements this finding by arguing that most revenue growth of food- retailers occurred outside developed countries. Moreover, the distance may also result in more comprehensive research before entering the market which causes less uncertainty and risk (Azar, 2014). Furthermore, the cosmopolitan character of global cities reduces ethnocentrism towards foreign companies (Carpenter, Moore, Alexander & Doherty, 2013). This theory is compatible with Goerzen et al (2013) which states that when expanding to global cities retailers face lower levels of the discrimination aspect of LOF as citizens are more exposed to international influences and therefore are more open towards foreign companies. So, despite food and culture being intertwined and the costs associated with diversification, it will be expected that global cities reduce these disadvantages. Following this line of thought, the following hypothesis is created:

H3: Being a food retailer positively moderates the relationship between global city location and firm financial performance.

(24)

24

Chapter 4: Empirical design

This quantitative research follows a deductive approach to investigate the effect of global city location on firm financial performance. In this chapter, the research design will be presented by visualization of the conceptual model. Further, the database, the sample, the dependent variable, the independent variable and the control variables are discussed.

4.1 Conceptual model

The conceptual model of this research is as followed:

4.2 Database

This research follows a quantitative design and gathers its data from the database Orbis developed by Bureau van Dijk. This database provides an extensive, global collection of economic data and financial performance of firms across country. Liu (2020) discussed the disadvantages of Orbis, which include downloading speed of the database and survivorship bias, which causes some companies to be deleted. Moreover, a lag of approximately two years was identified in the reporting of companies in Orbis and there have been troubles in the presentation of variables and the merging of data. Lastly, 15-25% of the data appears to be missing from Orbis. Despite these disadvantages, many researchers have chosen Orbis as a

Global city location Firm financial

performance

Being a food (vs.

non-food) company Being located within (vs.

outside) home region

H1 + H2

-

H3 +

Figure 1 Conceptual Model

(25)

25 reliable data source (Liu, 2020). Orbis was found to be most appropriate for studies taking a global perspective and for analysing top-performing multinationals (Bajgar, Berlingieri, Calligaris, Criscuolo & Timmis, 2020). This research compares global cities and firm financial performance on a global scale, and analyses the performance of subsidiaries of the biggest retail multinationals in the world, therefore making it a good research to investigate using the Orbis database. Moreover, to cover possible disadvantages, data was analysed using a timeframe of three years, namely 2019, 2018 and 2017 to cover the lack of data presentation.

4.3 Sample

The sample consists of the top fifth-teen retailers, which will be based on the Global Powers of retail list of 2021 of Deloitte UK. The retailers which are analysed in this thesis include Walmart Inc. (US), Amazon.com (US), Costco Wholesale Corporation (US), The Kroger Co. (US), Aldi (DE), Schwarz Group (DE), Walgreens Boots Alliances (US), The Home Depot (US), CVS Health Corporation (US), Tesco (UK), Target Corporation (US), Ahold Delhaize (NL), JD.com (CH), Aeon (JP) and Lowe’s (US). The majority of these companies originate from the United States (60%), followed by Europe (26,7%) and Asia (13,3%). Orbis is used as the main source of data for collecting the list of subsidiaries owned by these multinationals across countries.

A batch search for these companies was conducted in Orbis and the data was filtered for missing values and possible errors. This left 421 subsidiaries¹ to analyse. This sample size exceeds the overall recommended sample size of 380 based on a confidence interval of 95%

and a population of 20.000. This way, the results can be generalizable for the entire population, namely companies operating internationally and in the retail industry. Most of the subsidiaries in the dataset have a parent company with a European background, 38,5% have an Asian background and 15,9% originate from North America as presented in figure 1, which means that the proportions of the original sample have changed.

Frequency Percent

Valid Asia 162 38,5

Europe 192 45,6

North America

67 15,9

Total 421 100

1 This returns a successful rate of 10.5%, since in total these companies have 4009 subsidiaries.

(26)

26

Table 1 Origin of company – continent

4.4 Variables

4.4.1 Dependent variable

The dependent variable of this study is firm performance. Firm performance can be distinguished into two groups. First, financial performance, which includes profitability performance, market value performance and growth performance. Second, strategic performance which includes employee satisfaction, customer satisfaction, environmental performance, environmental audit performance, social performance and corporate governance performance (Selvan, Gayathri, Vinayagamoorthi & Kasilingam, 2016). The focus of this analysis will be on financial performance of MNE subsidiaries. Financial performance can be measured by multiple factors. Return on assets and return on sales are commonly used to determine firm financial performance in combination with locations (de Jong, Phan & van Ees, 2011; Pavelkova, Zizka, Homolka, Knapkova & Pelloneova, 2021). Also, these determinants have been used to research firm performance in retail companies, which is the main interest of this thesis (Mohr & Batsakis, 2016; Mohr, Wang & Shirodkar, 2014). The measure of return on assets explains the effectiveness of using firm assets and indicates future performance as it is interrelated with return on investment (Heikal, Khaddafi & Ummah, 2014). Return on sales explains the profit margin and represents operational efficiency (Bigcommerce, 2021). As these measures are widely used and accepted, this research will use return on assets and return on sales as determinants of firm performance. It was possible to generate the return on assets via the database for the years 2019, 2018 and 2017. An average was computed by adding these values and dividing them by three. The measure of return on sales was computed by dividing net income by operating revenue for 2019, 2018 and 2017. The average was generated by adding these values and dividing them by three. When values were missing for certain years, the added values were divided by the number of values available.

(27)

27 4.4.2 Independent variable

The independent variable of this research is the global city location of MNE subsidiaries. Differences exist focussing on the conceptualization of global cities (GCs). Some focus on transportation indicators to show the interconnectedness of the city (Smith &

Timberlake, 2001; Otiso et al., 2011), whereas others focus on the presence of APS businesses (Goerzen et al., 2013) to identify global cities. The Globalization and World Cities Network project (GaWC) combines

these approaches and produced a global city list based on APS business presence and interlocking networks. This approach is based on the view of Friedmann & Wolff (1982) who identified global cities as the centers of control for the world economy.

Critics of the GaWC point towards the bias of the list towards Western economies (Chubarov & Brooker, 2013). Therefore, new lists have emerged including the Powerlist of Mastercard, the Global Cities List of A.T. Kearney and the Global Power City Index of Mori Memorial Foundation. These lists place their focus on dimensions as infrastructure, liveability and ease of doing business. However, despite the criticism towards the GaWC, the list has been used in several academic studies and is consistent with the literature (Goerzen et al., 2013;

Belderbos et al., 2020; Chakravarty et al., 2021; Castellani et al. 2021). Therefore, it is chosen to use this list to identify global cities. The list of global cities presented is based on the year 2016 as this data was available to use in the research. The variable (Dummy_gc) is a binary measure that assumes value one if the subsidiary is located in a global city and value zero is the subsidiary is located in a non-global city.

4.4.3 Moderator

The moderator is expected to affect the strength of the relationship between global city locations and firm financial performance. Within this research, two moderators will be used which are tested to affect the strength of the relationship. The first moderator includes the subsidiary being located within or outside the home region of the parent company. In this thesis, the focus will

Figure 2 Global cities

(28)

28 be on North America, Europe and Asia as regions. These are chosen as the top-15 retailers of the world do not include companies from other regions such as Oceania or South America.

First, the regions of the parent companies and the subsidiary locations are identified. The regions of the parent company are based on the information presented by the Deloitte Global Power List, whereas the regions of the subsidiary are based on the country data derived by Orbis. After, these findings are compared and coded as ‘SAME’ or ‘DIFFERENT’. Subsidiaries that are flagged by ‘SAME’ were coded a 1 and subsidiaries which were flagged by

‘DIFFERENT’ were coded as 0. In so doing, the dummy ‘D_contin’ assumes value one if the subsidiary is located in the same region of the parent, zero otherwise.

The second moderator includes the differentiation between food and non-food retailers within the retail industry. This distinction is commonly used by other researchers (e.g. Etgar &

Moore, 2010). The differentiation of food vs non-food retailers will be made by the classification of the Deloitte Global Power List and therefore based on the origin of the company. For example, Walmart is flagged as a hypermarket/supermarket which indicates a substantial relationship with food, whereas Amazon is flagged as a non-store which indicates no relationship with food. Although Amazon does operate several supermarkets, the main intention of the company is not food-related. Subsidiaries that are flagged as ‘FOOD’ are coded 1, whereas subsidiaries flagged as ‘NON-FOOD’ are coded as 0. Thus, the variable

‘Dummy_food’ assumes value one if the parent group operates in the food industry, zero otherwise.

4.4.4 Control variables

Control variables are variables that are not in the direct interest of the research but may have a significant impact on the dependent variable. Adding control variables to the research will increase the power of the model, reduce error and enhance the accuracy (Allen, 2017). This way, it can be ruled out whether the effect of the independent variable on the dependent variable is based on various other influences. In this research, the control variables will be held constant by statistic control. The performance of MNEs is influenced by many external factors, which need to be taken in to account when conducting the research. Variables to control for are the size, age, debt and digital nature of the company. These variables can all change possible performance indicators. First of all, the size of the company is measured by the number of employees a company has. This control variable is chosen as bigger firms can exploit the advantages of economies more easily, which might lead to higher performance indicators. This is a widely acknowledge theory referred to as ‘economies of scale’theory (Hanson, 1964). The

(29)

29 data of the employees was retrieved for the years 2019, 2018 and 2017, after which an average was computed. Second, the control variable age is chosen as younger firms might have better performance indicators compared to older firms due to an assumably small amount of assets and high net income. This is supported by theories as the product or industry life cycle. This value was computed using the founding year of the company minus 2021 to compute the years a company is active. Third, the debt of a company can affect the net income which is needed to calculate the performance indicator. The debt of the company is measured by the liquidity ratio.

It was possible to retrieve the liquidity ratio from Orbis. Also for this variable, an average was computed based on 2019, 2018 and 2017. The choice of these control variables is in line with other research done on the topic of location, performance and retail (e.g. de Jong et al., 2011;

Mohr & Batsakis, 2016). Lastly, it was chosen to include a digital nature dummy which in the case of the sample used includes solely Amazon. This choice was made as e-commerce businesses have had tremendous growth the past years, which might affect the performance indicators of a firm.

4.4.5 Overview of variables

To give a better overview of the data, the measurement method and which source is used to retrieve the data is presented in table 2.

Type of variable

Variable Measurement method

Source Code

Dependent variable

Firm financial performance

Average Return on Assets (2019, 2018, 2017)

Retrieved from Orbis

AV_ROA

Dependent variable

Firm financial performance

Average Return on Sales (2019, 2018, 2017)

Retrieved from Orbis +

calculated by:

Net

income/operating revenue

AV_ROS

Independent variable

Location Global city vs non-global city

GaWC – 2016 Dummy_gc

Moderator Location Within or outside of home region

Retrieved from Orbis +

compared

location of parent company and subsidiary

D_contin

Moderator Location Food vs non- food company

Retrieved from Orbis + Deloitte Global Power list

Dummy_food

(30)

30 Control

variable

Size of company

Number of employees

Retrieved from Orbis

NEW_EMPLOYEES Control

variable

Age Years active Retrieved from Orbis

NEW_YEARSACTIVE Control

variable

Debt Liquidity ratio Retrieved from Orbis

NEW_LIQ Control

variable

Digital nature

Digital origin Deloitte Global Power List

DIGITAL_DUMMY

Table 2 Overview of variables

4.5 Data check

Before analysing the data, a data check was performed to ensure a good quality of the data and check for missing values, errors and outliers. To check for outliers, the Z-score of the dependent and control variables was created using the formula; Z = (x-μ)/σ. After, all data with a Z-score above three or below minus three, which is seen as a standard cut-off within statistics, was flagged as an outlier and removed from the data. This standard cut-off is based on the empirical rule that 99,7% of the data must fall within the range of three standard deviations from the mean. For return on assets, eleven values above 0,05 and below minus 0,07 were removed from the data. For return on sales, ten values under minus 2,84 and above 2,57 were removed from the data. For the control variable average employees, no values were removed from the data. For the control variable years active, two values were deleted which were both above 2000. For the control variable average liquidity, nine variables above 41,21 were removed from the data.

For the independent variable, a visual check was done on the list of the GaWC and the list of cities in the data file to understand which language was used. All city names were changed to their correct English spelling. A visual check was performed to verify whether all the cities were correctly spelled. To check, Excel was used to flag cities that did not

correspond with the cities listed on the GaWC list. This way, it was possible to manually verify the complete list. The spelling or specificity of 23 cities was changed for example due to traditional spelling or because the location had the name of a suburb within a city (e.g.

Southbank in Melbourne or Manhattan in New York).

After the check for outliers, missing values and errors, it was verified whether the variables follow a normal distribution. Normal distribution helps to generalize the research, which is especially useful for working with variables that are distributed differently (Rojas, 2010). Two approaches were used to check for a normal distribution. First, the dependent and control variables were visually examined using a histogram after that the skewness of these

(31)

31 variables was explored. The histogram of both AV_ROA as AV_ROS followed a normal distribution which can be seen in figure three and four in appendix one. Moreover, AV_ROA showed a small negative skewness of minus 0,540, which is however between the range of 0 and minus 0,5 and therefore is not seen as a problem. AV_ROS showed a moderate negative skewness of minus 0,729 which falls between minus 0,5 and minus 1. However, as the absolute variable is between the range of 1 and minus 1, this is not seen as a problem. Secondly, the control variables, with exception of digital due to the binary nature of the variable, were checked for normality. The histogram of all the control variables indicated that the variables were not normally distributed. This was confirmed by the skewness of the data which were all positive, namely 1,389 for employees, 2,210 for years active and 3,976 for liquidity. The values of the variables match with the characteristics of a lognormal distribution based on three reasons: All values were above 0, the data showed a high mean and standard deviation and the data shows a high peak and fast decrease in the histogram (Weibull, 2021). Due to these characteristics, the variables were converted using the formula: X*=Log10(X). After computing the new variables, the data was found to be more normally distributed as seen in the newly computed histograms in figure five in appendix 1 and the skewness was all between the accepted range of minus 1 and 1 as seen in table three and four in appendix two.

4.6 Methodology

Multiple analyses were used to test the hypotheses stated earlier. For all hypotheses, a threshold of 0,1 was chosen. A threshold of 0,05 is common for testing hypotheses as it indicates a chance of five percent or lower that certain results are found when hypothesis zero is true. A small p-value indicates stronger evidence to reject the null hypothesis. However, the study shows that significant results might be irrelevant due to small effect sizes or sample sizes and that non-significant results sometimes are highly relevant (Leo & Sardanelli, 2020).

A higher p-value threshold does increase the chance of a type 1 error, but decreases the chance of a type 2 error. As findings about location and its effect on performance have been contradicting, the result of this study might be highly relevant and steer towards new future research topics. Therefore, it was chosen to use a threshold of 0,1.

To test the first hypothesis, a preliminary analysis was performed using an

independent t-test. An independent t-test to test if there are differences observed between the means of a group. Secondly, a hierarchical linear regression model was performed. This

(32)

32 technique allows separating the independent variables in blocks to statistically control for certain variables (Sagehub, 2021). This control is generated by calculating the adjusted R2 when adding extra variables. Therefore, it indicates the variance explained by the model after taking the control variables in consideration (University of Virginia Library, 2016).

The second hypothesis had two elements. Hypothesis H2A tested whether global cities were chosen less often when located in the home region. This was tested using a chi-square test, which allows testing whether the data is distributed as expected. The following formula was used: χ2 = ∑(Oi – Ei)2/Ei, in which Oi is the observed value and Ei the expected value.

The second element, hypothesis H2B was tested using PROCESS by Andrew Hayes. This model is an extension of SPSS and allows to compute regression analysis containing moderators and control variables. The function allows choosing between 92 models to test, ranging from simple moderation and mediation to more complex models (Process, 2021). To test hypothesis H2B, the first model was chosen, namely simple moderation.

The third and last hypothesis tested the moderation effect of being a food vs non-food company on the relationship between global city location and firm financial performance.

Similar to hypothesis H2B, the PROCESS function developed by Andrew Hayes was used.

Based on above presented methodology and analysis, the results of the study are shown in the next chapter. First, the descriptive statistics of the results are given after which the correlation of the variables is presented. The chapter ends with a presentation of the results per hypothesis.

(33)

33

Chapter 5: Results

5.1 Descriptive statistics

The descriptive statistics for the dependent variables AV_ROA and AV_ROS are presented in table 21 and 22 in appendix two. This table shows the Number of observations (N), minimum, maximum, and mean values, standard deviation and skewness. The AV_ROA is shown in 345 of 421 cases and has a mean of 0,0003 or 0,03% with a minimum of -0,04 and a maximum of 0,05. This mean is considered rather low as a return on assets of five percent is normally considered a good return (Forbes, 2021). The AV_ROS is seen for 332 of 421 cases, which represents 78,8% of the total data. The data shows a range of -2,56 and 2,41 with a mean of 0,0377 or 3,7%. This indicates a positive profit to revenue ratio. The control variable employees is available in 115 cases, years active in 418 cases and liquidity in 297 cases. As the control variables were computed using a lognormal formula, it is important to look at the original means presented in table three in appendix two. For employees, the minimum statistic was zero, the maximum statistic 982 (rounded) and the mean 213 (rounded). For years active, the minimum presented was two, the maximum 133 and the mean 22,56 which translates to 22 years and 7 months. Lastly, the AV_liq shows a minimum of zero and a maximum of 41,21 with a mean of 3,3045.

As the independent variable, the moderators and the digital nature of the company are binary variables it was not possible to compute the minimum, maximum or mean. Therefore it was chosen to present the variables using a frequency tables which are shown in table 22 to 25 in appendix two. First, the independent variable of global city location which is presented in figure five. Within the dataset, 169 of the variables were flagged as a global city representing 40,1% and 252 values were flagged as a non-global city, representing 59,9% of the data. The total data set was flagged with either a global city or a non-global city label. The first

moderator is presented in table 23, which represents whether the subsidiary is located within or outside the home region of the parent company. It was found that 89 subsidiaries were located outside the home region, representing 21,1%, whereas 332 subsidiaries were located within the home region representing 78,9%. The second moderator is presented in table seven.

For the second moderator food vs non-food (table 24) companies, it was found that 334 subsidiaries originate from a food basis representing 79,3%, whereas 87 subsidiaries are flagged as non-food representing 20,7%. Lastly, the control variable digital nature (table 25) was verified. Within this data set, 394 subsidiaries are not digital in nature, whereas 27

Global cities as a location: The effect on firm financial performance