Foreign Banks and the Importance of Bilateral Distance. The Impact of Technological Progress.

(1)

Foreign Banks and the Importance of

Bilateral Distance. The Impact of

Technological Progress.

Koen van de Berg

(k.g.m.van.de.berg@student.rug.nl)

s2736039

June 20, 2017

University of Groningen

MSc. International Economics and Business

Faculty of Economics and Business

Master Thesis

(2)

Abstract

This thesis investigates the impact of technological progress on foreign bank presence and in particular the impact of technological progress on the importance of bilateral distance in explaining foreign bank presence. Using data from 2013, a sample of 8.368 observations and a logistic regression technique, technological progress in the host and home country is both found to be a significant factor in explaining foreign bank presence. Technological progress of the home and host country both diminishes the importance of geographical and regulatory distance in foreign banking.

(3)

List of Figures

1 Share of foreign banks as part of the total amount of banks . . . 5

2 Visualization of theoretical model . . . 12

List of Tables

1 Descriptive Statistics . . . 19

2 Baseline Regression Results . . . 23

3 Regression Results . . . 26

4 Robustness Tests on the Baseline Results . . . 28

5 Robustness Test on the Regression Results . . . 30

(4)

I. Introduction

For the past decades, financial markets have been characterised by a steady increase of financial integration and internationalisation. Particularly with the financial liberal-isation of emerging markets and developing countries in the 1980s and 1990s financial markets around the world have become more internationalised and integrated than be-fore (Haggard and Maxfield, 1996; Levine, 2001). The integration of financial markets is often referred to as financial globalisation. Financial globalisation provided new eco-nomic and strategical opportunities and incentives for financial institutions to expand its activities abroad (Kose et al., 2009). One striking feature of this is the unprecedented increase of foreign banks. Foreign banks across the world are now taking a significantly larger share of (domestic) financial markets, both in terms of assets (relative) and in the number of foreign banks (absolute) (Claessens and van Horen, 2014b). Future per-spectives indicate that the share of foreign banks in domestic banking markets will even become more important (Claessens et al., 2017).

Foreign banks have been linked to (i) higher efficiency levels of the banking system due to increased competition, (ii) higher levels of economic development and (iii) financial stability (de Haas and van Lelyveld, 2006; Demirguc-Kunt et al., 1998; Li et al., 2016). The e↵ects of foreign banks are even more pronounced in emerging markets and devel-oping countries1_.

While it is true that globalisation has lowered the barriers for banks to go abroad, banks still face barriers when expanding its activities to a foreign country. Distance or as most studies (see for e.g. Berger, 2003; Claessens and van Horen, 2008) refer to as bilateral distance remains important in explaining foreign bank presence. Bilateral distance refers to the distance between two countries and can take on di↵erent forms of distances like physical distance, cultural distances such as language and colonial links and institutional distance like having the same regulatory system or not. The existing literature on the importance of bilateral links between the home and host country on foreign banks have linked several significant types of distances. Foreign banks that decide to expand abroad are more likely to choose countries that (i) are institutionally close to the home country, (ii) provide the best profit opportunities and (iii) have good access to both soft and hard information (Buch, 2000; Claessens and van Horen, 2014b; Rajan and Gopalan, 2015).

New developments in the literature on foreign banking have argued the importance of country heterogeneity (Claessens and van Horen, 2014b) and the diminishing

impor-1_{The global financial crisis of 2007/2008, however, has caused concerns about these e↵ects and}

(5)

tance of the bilateral links in explaining foreign bank presence (Buch, 2005). Countries are heterogenous and have di↵erent characteristics. Therefore, the importance of bi-lateral links around the world can di↵er in magnitude and importance. In trade, the development of new technologies has had a large impact on the transaction and in-formation costs (Friedman, 2005). These diminishing costs may also apply to foreign banks, and this would lead to lowering the importance of bilateral links. So while bi-lateral distance is still of great importance, still few is known why the importance of bilateral distance di↵ers from banks across di↵erent countries. Claessens and van Horen (2014a) linked the importance of bilateral distance to OECD countries and non-OECD countries and argued that the importance of bilateral distance is less for banks from OECD countries. However, do not elaborate why banks from OECD countries have an advantage.

In this thesis, the di↵erences in the importance of bilateral distance are investigated into more detail. Since technological progress is associated with better technological capa-bilities and resources for banks (e.g. internet banking, screening capacapa-bilities) (Berger, 2003) this thesis investigates how technological progress has a↵ected the importance of bilateral distance in foreign banking.

The following research question will be examined in this thesis: What is the e↵ect of technological progress on the importance of bilateral distance in foreign bank presence?

To answer the research question, an original dataset is constructed based on the Claessens and van Horen bank ownership database and absolute bilateral distance variables. The dataset covers 103 di↵erent host countries and 84 di↵erent home countries from di↵er-ent income class levels (developing, emerging and advanced). For each country pair (in total 8.386), foreign bank presence between the home and host country is determined in 2013 (most recent data).

(6)

This study adds to the existing literature on the importance of bilateral links in foreign banking in several ways. First, it re-examines the importance of bilateral distance in foreign banking. Is it still a significant factor a↵ecting locational choices of foreign banks? Second, the impacts of technological progress on the presence of foreign banks in both the home and host country are studied. Third, this thesis expands the litera-ture on foreign banking by explaining that the importance of bilateral distance is not a homogenous relationship. Focussing on technological progress, this thesis investigates whether technological progress in the home and host country a↵ects the importance of bilateral distance in foreign banking. Fourth, the study uses an original dataset, which compared to other studies (for e.g. Buch, 2005) more comprehensive in the number of variables used and observations included.

The remainder of this paper is as follows. Section II reviews the existing literature to formulate appropriate hypotheses to answer the research question. It also includes a theoretical model. Section III provides the data discussion and the methodology that is used to test the hypotheses. Section IV shows the empirical results. This section shows and elaborates on the results of the regression analyses. It also includes several robustness tests. Section V gives the conclusion and discussion. It also provides areas for future research.

II. Literature Review

To be able to answer the research question (see section I.), the literature review consists out of several subsections that gradually narrow down the literature of interest. First, subsection A. discusses globalisation as the driver of the existence of foreign banks. Subsection B. reviews the importance of foreign banks in the domestic banking and international banking markets. Subsection C. examines the importance of bilateral distance in the locational decisions of foreign banks. Finally, subsection D. reviews the impact of technological progress on the importance of bilateral distance.

A. Financial Globalisation

(7)

Transporta-tion costs decreased, which made the geographical distance in the trade less relevant. Communication costs decreased as well due to the development of new communication systems. This allowed the sharing of information over long(er) distances (Friedman, 2005). Second, governments and in particular governments of developing countries and emerging markets have and still are consequently lowering regulations to allow oper-ations of foreign institutions like international banks and foreign banks (Detragiache et al., 2008).

Globalisation did not only shape the domestic financial market (Cetorelli and Goldberg, 2012), it also changed the international banking market. In this paper, we focus on the increase of foreign banks active in the host country (more on this in section II.B). The literature on foreign banking often uses the terms foreign banks and international banks interchangeably. However, they are di↵erent in definition (Goulding and Nolle, 2012). To be precise and concise in this study, we di↵erentiate between the two. International banks are banks that have lending and borrowing activities outside its home country but do not have a physical presence in the foreign host country. It therefore only con-ducts in cross-border operations like lending and borrowing. Foreign banks, however, do have a physical presence in the host country. Foreign bank entry, therefore, relates to a foreign bank that enters the financial market of a country and start operating in the domestic market (Clarke et al., 2003).

Globalisation, therefore, has created new opportunities for domestic institutions (both financial and non-financial) to expand its activities abroad and benefit from the lower costs of communication and transportation. In the next section, the importance of foreign banks is examined into more detail.

B. Foreign Banks

(8)

10% 20% 30% 40% 50% 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

Global High Income Emerging Developing

Figure 1: Share of foreign banks as part of the total amount of banks

Figure 1 displays the share of commercial foreign banks active as part of the total amount of commercial banks (domestic and foreign) for high-income countries, emerging markets and

developing countries between 1995 and 2013

crisis has hit the financial world, the share of foreign banks in emerging markets and developing countries increased by (approx.) 10 percent. Although it is true that some foreign banks exited these markets in the years after the financial crisis, most of the in-crease in foreign bank share is due to the significant dein-crease in domestic banks (either stopping its operations or due to merger & acquisition)(see Appendix C1). Appendix C1 provides a detailed overview of the number of domestic banks and foreign banks active in each year2_.

Besides the change in the importance of foreign banks in the domestic banking market (host country), there were also significant changes in the home countries of the foreign banks. Most foreign banks are owned by developed countries and in particular OECD countries. However, as Claessens and van Horen (2015) argue, foreign banks that come from OECD countries become less important. There is thus a shift from foreign banks that come from OECD countries and other developed countries to banks from emerging markets and developing countries.

Banks that decide to expand its activities abroad basically have two methods to do so (Clarke et al., 2003). First, a bank can choose to serve its foreign clients without having a physical presence in the foreign country (host country) via cross-border lending and borrowing (transactions). The other method is by establishing a physical presence in

2_{Note that the number of domestic banks and foreign banks is based upon the CvH database.}

(9)

the host country of interest by either starting a new branch or subsidiary (de novo) or via merger & acquisition (M&A)(Buch, 2000). In this paper, we are interested in banks that have a physical presence in the host country via both de novo and M&A. Throughout the paper, we do however refer to cross-border transactions, but this will not be the main field of interest.

When examining the di↵erences between foreign banks and domestic banks in the host country, large di↵erences are noticeable. One strand of literature argues that foreign banks focus their lending on large multinational enterprises, while domestic firms often serve small businesses and consumers (Berger et al., 2001). Foreign banks also tend to increase the economic growth of developing countries and emerging markets (Bruno and Hauswald, 2014). Foreign bank presence leads to more competition in the domestic banking market and therefore increases the productivity and efficiency of the domestic banks. Foreign banks also have greater abilities to absorb losses which can enhance the financial system in the host country where the foreign bank is present (Crystal et al., 2002).

The literature on foreign banks has established numerous reasons why banks decide to expand its activities abroad. Focarelli and Pozzolo (2005) argue that banks expand their activities abroad for strategical and financial reasons. Strategic wise, banks can benefit from the competitive advantage they have over their competitor banks in the host country and therefore increase their profits (Claessens and van Horen, 2014a). Banks that expand abroad also can benefit from economies of scale since. There is also evidence that banks expand to countries that are more distant from them. Since business cycles in developing countries and emerging countries are di↵erent from those in developed countries, (Rand and Tarp, 2002) banks might have an incentive to invest in more distant countries because of portfolio diversification reasons. Banks, therefore, might benefit from their more distant investment since the possibility of having a shock in both countries is less than investing in less distant countries (which most often im-plicates that the country of interest has the same development stage).

(10)

C. Bilateral Distance and Foreign Bank Presence

While globalisation implicates that distance has become less relevant, it remains impor-tant in explaining the di↵erences between countries (McCann, 2008). In trade theory, bilateral distance (defined as the distance3 _{between the two countries of interest)}

be-tween countries have been studied for a long time and found to be an important deter-minant of trade between countries (Deardor↵, 1998; Groot et al., 2004). More recently, the importance of bilateral distance on locational decisions of financial institutions to expand abroad is studied.

When banks decide to expand abroad, they prefer to choose a country that is in the proximation of the home country. Galindo et al. (2003) argue that the absolute dis-tance between the home and host country acts as a barrier for foreign banks to enter a particular host country. The more remote the host country is, the less likely a bank from the home country will expand to the host country. Claessens and van Horen (2014b) argue that while absolute distance is important in explaining the foreign bank presence, relative distance is important as well. Banks that are relatively close to their host country compared to other countries are more likely to establish a foreign bank in that host country.

Other than non-financial institutions, banks face another constraint when expanding abroad. Banks provide loans to businesses and consumers and act as financial intermedi-aries. Therefore, banks rely on both soft and hard information to monitor the borrower to prevent adverse selection and moral hazard (Agarwal and Hauswald, 2010). Screen-ing borrowers and their creditworthiness is more difficult for more distant banks, and therefore foreign banks have more difficulties providing loans in an efficient way. Soft information relates to financial statements, income records and personal characteristics of the borrower. When soft information has to travel more within the banks network (e.g. from the foreign country to the home country), the information quality transmit-ted deteriorates and becomes less useful (Aghion and Tirole, 1997; Liberti, 2005; Liberti and Mian, 2008). Foreign banks, therefore, choose countries where (soft) information on the borrower is more transparent and easier to obtain. Claessens and van Horen (2014a) find evidence for this and argue that the distance in information quality between the home and host country is significantly related to the presence of foreign banks in the host country. The importance of information quality is also echoed in preceding studies that explain the ’following the customer theory’. Goldberg and Saunders (1981) argue that foreign banks tend to follow its domestic customers and businesses abroad to act as the local lender. Since the foreign bank does have good information on these borrowers.

3_{The distance between two countries does not necessarily refer to the physical distance between}

(11)

Section II.A discussed that globalisation went along with the improvement of regula-tions (i.e. lowering the restricregula-tions on foreign activity in the host country). Banks that expand abroad prefer to choose countries that have better regulations since banks are freer to implement its strategy or to transform its strategy to the local environment. Lensink et al. (2008) argue that regulatory quality of the host country is indeed impor-tant in determining the presence of foreign banks. Host countries with better regulatory quality have more foreign banks than countries with lower regulatory quality. Claessens and van Horen (2014a) extend the literature on the importance of regulatory quality on the locational decisions of foreign banks by examining the distance in regulatory quality between the host and home country. Using their original database, Claessens and van Horen find that bilateral distance in regulatory quality is important in explaining the presence of foreign banks as well. The less remote the home and host country are in terms of regulatory quality, the more likely foreign banks will be present in the host country.

Another strand of literature focusses on the legal environment of the host country and foreign bank presence. For financial institutions (and non-financial institutions), legal rights are important since it explains how well these institutions are protected by law (Porta et al., 2000). For banks, contract enforcement is an important issue and there-fore it is expected that banks that expand abroad choose countries with good legal institutions (Bae and Goyal, 2009). Detragiache et al. (2008) argue that foreign banks are more likely to enter the country when the country has good legal rights. Claessens and van Horen (2014a) argue that banks are more likely to choose countries that have the same level of legal rights since they have the experience of operating in the same level of legal rights in the home country. Distance in legal quality between the home and host country is, therefore, an important determinant in explaining foreign bank presence.

(12)

Section II.C examined the importance of bilateral distance in the presence of foreign banks. While the literature reveals that bilateral distance is still important in the decision-making of foreign banks, it cannot be ignored that technological progress has lowered the costs associated with distance. Remarkably, little is known about the drivers that a↵ect the importance of bilateral distance in explaining foreign bank presence. Section II.D examines the e↵ects of technological progress on bilateral distance.

D. Technological Progress and Bilateral Distance

It cannot be ignored that banks rely heavily on technological progress and the improve-ment of information technology (IT) (Berger, 2003). These technologies are used to build and monitor their portfolio, create new securities and employ economic and sta-tistical models. IT is also used to gather the soft information required to assess their potential customers. Berger and DeYoung (2006) argue that technological progress has provided bank holding companies to have better control over its (foreign) affiliates and that agency costs accompanied with the management of these affiliates have decreased. This suggests that distance has become less of a constraint in the locational decisions of foreign banks. Technological progress or development refers to the efficiency improve-ments of the production or use of a product, device, or process (Koh and Magee, 2006). It is expected that technological progress in the home country provides banks with better technological capabilities and resources. These technological capabilities and resources should provide the bank with better and more possibilities to expand abroad. To test whether the technological progress of the home has an e↵ect on foreign bank presence in the host country, the following hypothesis 1A is constructed.

Hypothesis 1A: A higher level of technological progress in the home country is as-sociated with a higher level of foreign bank presence in the host country.

Since one might expect that host countries with a higher level of technological progress provide foreign banks with better technologies (e.g. new technologies for screening borrowers), attract more foreign banks, hypothesis 1B is constructed.

Hypothesis 1B: A higher level of technological progress in the host country is asso-ciated with a higher level of foreign bank presence in the host country.

(13)

countries remains unknown. Claessens and van Horen debate that OECD countries are better able to deal with information asymmetries than foreign banks located or from non-OECD countries. However, do not provide any empirical proof of this. The study of Claessens and van Horen (2014a) however does imply that the importance between the home and host country of foreign bank is heterogeneous among countries.

Buch (2005) has studied if the importance of distance in locational choices of banks to expand abroad has decreased over time. This study is based on the assumption that technological progress has lowered banks’ information costs and therefore geographi-cal distance should become less important over time. Using asset data of commercial banks from five OECD countries (France, Germany, Italy, UK and US) between 1983 and 1999, no evidence is found that distance is becoming less important for foreign banks location choices. However, as Buch argues, one reason for the lack of evidence might be that improvements in information technology might take more time.

New technologies lower the transportation, transaction and information costs (Cairn-cross, 2001).Berger (2003) argues that technological progress has an e↵ect on the ge-ographic expansion of banking organisations. New services (e.g. internet banking) created due to the technological progress and the improvement of traditional services. Therefore, it is no longer needed for customers to be in the proximity of the banks’ location. Berger also debates upon the improvement in capabilities of credit scoring of banks. Therefore, it is no a requirement for credit scoring by banks to be in the near geographical distance of the customer or potential loan applicant.

Literature thus far has acknowledged the heterogeneity among countries in the im-portance of bilateral distance in foreign banking but have not yet provided a well-established answer. Therefore, this thesis adds to this strand of literature by examining the impact of technological progress on the importance of bilateral links in foreign banking. Similar to hypotheses 1 it is expected that technological progress in the home country provides banks with better technological capabilities and resources. These technological capabilities and resources should provide the bank with a surplus to deal with bilateral distances. To examine if there is evidence for this, hypothesis 2A is con-structed. Hypotheses 2 are based on geographical distance and hypotheses 3 will test the other bilateral distance variables.

(14)

distance when the host country has a high level of technological progress. To examine if this is the case, hypothesis 2B is constructed.

Hypothesis 2B: A higher level of technological progress in the host country has a negative e↵ect on the importance of geographical distance in the foreign banking Financial development of countries financial institutions and the increase of bank tech-nology provide banks with better equipment to analyse the customer in the foreign host country. Therefore, these banks are less dependent on the quality and need of soft-information (Sedunov, 2017).

Hypothesis 2A and 2B studies the impact of technological progress on the importance of geographical distance in explaining foreign bank presence. To examine which channels make distance so important and on what channel technological progress has an e↵ect on the importance of bilateral distance, hypothesis 3A and 3B are created. Hypothesis 3A and 3B focus on the specific measurements of bilateral distance (Legal distance, Regulatory distance and Informational distance, hereafter referred to as bilateral dis-tance) that have been found significant in explaining foreign bank presence.

The same line of reasoning for hypothesis 2A and 2B is applied to hypothesis 3A and 3B. It is expected that banks from countries with a higher level of technological progress have better technological capabilities or resources to overcome the difficulties associated with bilateral distance.

Hypothesis 3A A higher level of technological progress in the home country has a negative e↵ect on the importance of bilateral distance between the host and home country

It is assumed that host countries with a higher level of technological progress provide foreign banks with better technologies. Therefore it is expected that foreign banks are less dependent on the importance of bilateral distance when the host country has a high level of technological progress.

Hypothesis 3B A higher level of technological progress in the host country has a negative e↵ect on the importance of bilateral distance between the host and home coun-try

(15)

E. Theoretical Model

Based on the literature review and the hypotheses discussed in section II, a theoretical model is constructed. A visual design of the theoretical model is shown in figure 2. This model does show arrows indicating the direction of the expected e↵ect but does not include positive or negative signs. However, the expected signs are discussed below. The model in figure 2 only includes hypotheses 2 and 3 since these are the ones that will answer the research question.

In line with the methodology and literature, it is expected that bilateral distance has a negative e↵ect on foreign bank presence4_{. Figure 2 also includes the modifier e↵ect}

(technological progress in both the home and host country) that a↵ects the relationship between the importance of bilateral distance and foreign bank presence.

Figure 2: Visualization of theoretical model

Bilateral Distance Foreign Bank Presence

Technological Progress Home Country Technological Progress

Host Country

Links between the independent variable, dependent variable and mediation variables. The variable bilateral distance refers to the geographical distance, legal distance, regulatory distance and

informational distance.

It is expected that both technological progress in the home country and technological progress in the host country will have a positive sign. This is in line with hypotheses 2 and 3 since this implicates that the negative e↵ect between the importance of bilateral distance on foreign bank presence will be of less importance if the home and host country have a higher level of technological progress.

4_{As argued in the data and methodology section, the bilateral distance can both have a positive and}

(16)

III. Data and Methodology

The data and methodology section discusses the data collection and the proxies used for the dependent and explanatory (independent) variables of interest5_{. First, section III.A}

gives a description of the dependent variable and how the dependent is constructed. Section III.B provides an overview and explanation of the explanatory variables of interest. Section III.C discusses the control variables that according to the literature review, might a↵ect the significance of the explanatory variables. Finally, section III.D discusses the methodology that is most appropriate to obtain clean empirical results in section IV.

A. Dependent variable

The dependent variable of interest in this thesis is foreign bank presence. Data on for-eign bank presence is retrieved from the Claessens and van Horen (2015) bank ownership database (CvH database)6_{. The CvH database is used since it is the most suitable and}

more detailed than other databases on bank ownership (e.g. Micco et al., 2007). The CvH database is unique since it tracks the home country of the foreign bank active in a country and when it has entered or exited the country. By using individual banks websites, annual reports, parent companys websites, banking regulatory agency/Central Bank websites, reports on corporate governance, local stock exchanges, SECs Form F-20, and country experts the bank ownership database is well constructed and precise (Claessens and van Horen, 2015).

To determine the home country of the foreign bank, the CvH database sums up all the foreign shareholders. The foreign shareholder with the largest share is then con-sidered as the host country. Determining the home country of the foreign bank via this way has to be taken with caution. As Claessens and van Horen (2014a) argue, it remains ambiguous if the largest shareholder also has the largest e↵ect on the banks characteristics and strategy. When the bank is owned by a large group of shareholders, the largest foreign shareholder may be rather small. However, since data availability is limited and determination of the e↵ect of di↵erent shareholders and how the ownership is structured among these banks, this measurement is seen as most favourable at the moment (Claessens and van Horen, 2008; Clarke et al., 2003).

For this study, a new and original database is constructed building upon existing data

5_{See Appendix A1 for a detailed overview of all variables including a short description, unit of}

measurement and data sources.

6_{The Claessens and van Horen bank ownership database is the most comprehensive database that}

(17)

and own calculations. First of all, all host and home countries that have foreign banks active in 2013 are included in the database. To examine the distances between the home and host country of the foreign bank, country pairs are generated and included in the database. There is, however, one problem when including all available countries from the CvH database to the data set used in this thesis. Foreign banks that are active in countries that operate as tax havens or o↵shore locations possibly have di↵erent drivers (e.g. regulatory and tax advantages) to be active in these countries than the reasons of foreign banks active in non-tax haven countries (Demirguc-Kunt and Huizinga, 2001). Therefore we exclude these countries7 _{from the database used in this thesis.}

There-fore, the dataset covers 103 di↵erent host countries and 84 di↵erent home countries. Subtracting the country pairs that are paired with itself, 8.386 possible country pairs could have established a foreign bank from the home country in the host country of that pair. All possible country pairs are included since this provides the opportunity to di↵erentiate between country pairs that have established a foreign bank presence and country pairs that have not.

B. Independent variables

According to the literature discussion and theoretical model in section II. several bilat-eral distances between countries that have established a foreign bank presence have been found significant. In this section, we discuss the independent variables or explanatory variables (these terms will be used interchangeably) of interest. The following sections will discuss geographical distance, information distance, legal distance and regulatory distance into more detail.

Geographical Distance

The geographical distance between the host and home country of the foreign bank is measured using the physical radial distance between the capitals of both countries in each country pair (CIA, 2016).

(|coordinatesofcapitali coordinatesof capitalj|)

To capture the relative distance instead of the absolute distance, the log of the number of kilometres between home and host country capitals is used. This is in line with (Claessens and van Horen, 2014a).

The average distance between all our country pairs is 6.954 kilometres. The average distance between the home and host country that have established a foreign bank between in 2013 is 4.277 kilometres. While these results are just mean statistics, they

7_{Andorra, Antigua and Barbuda, Bahamas, Bahrain, Barbados, Bermuda, British Virgin Islands,}

(18)

show that the distance is indeed an important element in foreign bank relationships. The descriptive statistics of the geographic distance variable are further elaborated in detail in section III.D. The descriptive statistics do not include the geographical distance in kilometres but in logs, since the log variable will be used in the regression (for more information on this see sectionIII.D). In line with Buch (2005); Claessens and van Horen (2008), it is expected that geographical distance is negatively related to foreign bank presence. Thus, the more physical distance there is in a country pair, the less likely the chance on a foreign bank presence of the home country in the host country.

Informational Distance

The Doing Business Report depth of creditor information is a good proxy to control for the use of hard and soft information (Claessens and van Horen, 2014b). The creditor depth information variable is obtained from The World Bank (2017a) using the getting credit dataset, hereafter referred to as informational quality.

Informational quality is measured based upon the accessibility, and quality of credit information available via public or private credit institutions. The variable ranges from 0 to 8. A higher value is associated with the better availability of credit informa-tion. A higher value, therefore, implies a greater possibility to use hard information in a country, whereas a lower value indicates that the use of soft information is more likely. Using the country pairs that are formed in the database (see section III.A), absolute distance between the home country’s informational quality and the host country’s in-formational quality is calculated.

(_{|Informationalquality}i Inf ormationalqualityj|)

By measuring the absolute distance between the informational quality a new variable is constructed called informational distance. The absolute distance is created since it is expected that countries which are closer to each other regarding credit quality, find it easier to operate in that countrys environment. Since these countries are already familiar with the same kind of environment in the home country. An increase in infor-mational distance between the country pair is, therefore, expected to decrease foreign bank presence.

Legal Distance

(19)

protect borrowers and lenders to collateral and bankruptcy. The index ranks countries from 0 to 12. The higher the score, the higher the quality of the laws active in the country.

Using the country pairs that are formed in the database (see section III.A), absolute distance between the home country’s legal quality and the host country’s legal quality is calculated.

(|legalqualityi legalqualityj|)

The newly created variable is called legal distance. It is expected that countries which are closer to each other regarding legal quality, find it easier to operate in that countrys environment. Since these countries are already familiar with the same kind of environ-ment in the home country. An increase in legal distance between the country pair is, therefore, expected to decrease foreign bank presence.

Regulatory Distance

Data on the regulatory quality of the home an host country is obtained from the World Governance Indicators (Kaufmann et al., 2005). Regulatory quality is measured by the ability of the government to implement sound policies and regulations. The Regulatory quality is ranked between -2.5 (worst regulatory quality) and 2.5 (best regulatory qual-ity).

Using the country pairs that are formed in the database (see section III.A), absolute dis-tance between the home country’s regulatory quality and the host country’s regulatory quality is calculated.

(_{|regulatoryquality}i regulatoryqualityj|)

the newly created variable is called regulatory distance. It is expected that countries which are closer to each other regarding regulatory quality, find it easier to operate in that countrys environment. Since these countries are already familiar with the same kind of environment in the home country. An increase in regulatory distance between the country pair is, therefore, expected to decrease foreign bank presence.

Technological Progress

(20)

and 1990s by The Conference Board and Groningen Growth and Development Centre (University of Groningen, The Netherlands).

While the independent variables of interest (geographical distance, regulatory distance, legal distance and credit information distance) are calculated with use of the absolute di↵erence between the home and host country, technological progress will not be mea-sured this way. One might expect that when two countries that both have a high level of technological progress (so the absolute di↵erence is close to zero), these countries will likely have the same level of regulatory, credit information and legal quality, which leads to low absolute di↵erences as well. This causes serious correlation problems. Therefore, technological progress will be measured as a stand-alone value for the host and home country. Using both the level of technological progress of the host and home country allows this study to estimate whether the importance of geographical distance, regula-tory distance, legal distance and credit information distance is a↵ected when the host or home country has a high level of technological progress.

C. Control variables

Since one might expect that other variables that are of less interest in this thesis can a↵ect the presence of foreign banks, several control variables are added to the study. In the ideal case, control variables should only have a relationship with the depen-dent variable and not with the independepen-dent or explanatory variables. Therefore, these control variables should be selected with care. Correlation and collinearity among the explanatory variables and control variables is tested in section IV.A.

Equal Language

Based upon studies from Deardor↵ (1998) and Hutchinson (2005) a dummy variable for language8 _{is added as a control variable. Having a common language is linked to}

higher levels of bilateral trade and economic integration. Language data is obtained from The World Factbook (CIA, 2014). The variable is a binary variable and equals 1 if the home and host country in the country pair share the same language and 0 if they do not. It is expected that having a common language will have a positive e↵ect on the presence of foreign banks from the home country in the host country.

Equal Currency

Ghemawat (2001) argues the importance of having a common currency in explaining trade patterns between countries. Having the same currency is often used to measure the administrative distance between the two countries. One can expect that when two countries share the same currency, foreign banks find it easier to adopt in that country.

8_{The most commonly spoken language in the country is used since it is expected that most financial}

(21)

Therefore, one might expect that when the home and host country of the foreign bank share the same currency or are located within the same currency union, bank presence will be higher. Therefore, a currency dummy is added to the regression which equals 1 if the host and home country of the foreign bank have the same currency and 0 of they have not.

Equal Region

Since countries within a region are more likely to trade and therefore have foreign bank presence of one another country from the same region, a region dummy is added to the regression. All countries are divided into seven di↵erent regions according to the World Bank. A country is then assigned to one of the following regions, South Asia (SA), Europa & Central Asia (ECA), Middle East & North Africa (MENA), East Asia & Pacific (EAP), Sub-Saharan Africa (SSA), Latin America & Caribbean (LAC) and North America (NA). When the home and host country of the foreign bank are located in the same region, the dummy variable will have a value of 1 and 0 when the region of the home and host country are di↵erent.

Host and Home Country Characteristics

Next, to the binary control variables, a variable to control for home and host character-istics is added to the analysis. Claessens and van Horen (2014a) and Buch (2005) argue that the economic size of a country is positively related with foreign bank presence. Countries that have a large economy do provide more opportunities to allow (multiple) foreign banks since the borrower and lender base is larger as well. Therefore, a control variable for both the home and host country’s economic size is added. Economic size of a country is commonly measured as Gross Domestic Product (GDP) and is obtained from The World Bank (2017b).

Final Sample

The final sample includes 8.386 observation (see section III.A on country pairs). The total database that is created for this thesis includes all the variables discussed in section III.A,III.B and III.C9_{. Since the explanatory variables do not change (much) over time}

and the control variables (with exception of the home and host country characteristics) remain equal a cross-sectional analysis is selected as the most appropriate analysis method.

Descriptive Statistics

(22)

Table 1: Descriptive Statistics

Variable Obs. Mean Std. Dev. Min. Max.

Dependent Variable

Foreign bank presence [binary] 8,386 .0714 .256 0 1

Foreign bank presence [contineous] 8,386 .125 .588 0 16

Independent Variables Geographical Distance 8,386 8.569 .855 4.770 9.898 Regulatory Distance 8,386 1.085 .774 0 3.8 Legal Distance 8,386 3.000 2.216 0 12 Informational distance 8,386 2.684 2.63 0 8 Interaction variables

Technological progress 2013 Host 8,386 10.739 .879 7.975 12.166 Technological progress 2013 Home 8,386 10.33 1.095 7.428 12.100

is added to Appendix 2A. The descriptive statistics reveal that there are 8.386 obser-vations for all variables. The number of obserobser-vations is equal for all variables because the number of possible pairs is dependent on the availability of variable information. As discussed in section III.A, two di↵erent measurements for the dependent variable are used. Foreign bank presence [binary] takes on either 0 or 1 and has a mean of .0714. This implicates that .0714 percent of the pairs do have a foreign bank presence in the host country. Since the standard deviation is large compared with the mean, there is much variability in the foreign bank presence as measured by 0 or 1. The second dependent variable, foreign bank presence [continuous], captures the number of foreign banks from the home country active in the host country. The maximum number of foreign banks between the host and home country is 16 (e.g. there are 16 banks from the United States active in Germany in 2013). The mean is slightly larger than in the binary variable and has a larger standard deviation. This is however normal since the range of the dependent variable runs between 0 and 16.

The independent variables are all continuous variables of which the geographical dis-tance is measured in logs. The regulatory disdis-tance between the home and host country ranges from 0 (minimum) to 3.8 (maximum). The mean value of the regulatory distance is 1.085 with a standard deviation of .774. According to the descriptive statistics, the maximum legal distance between two countries in a country pair is 12, and the least is 0. The mean legal distance between country pairs is 3.00 with a standard deviation of 2.216. The maximum informational distance between country pairs 8 and the least is 0. The mean informational distance between country pairs is 2.684 with a standard deviation of 2.63.

9_{More information on the dataset is attached to Appendix D. The complete data set can be received}

(23)

The descriptive statistics of the technological progress variables are more or less equal for the home and host country. There is a slight di↵erence since some host countries only act as host countries and others acts as home and host country.

D. Methodology

The dependent variable is measured using two di↵erent variables. The standard regres-sion consists out of a binary dependent variable which takes on 1 if there is a foreign bank present between the home and host country and 0 if there is no foreign bank presence between the host and home country. As robustness test, the number of foreign banks between the host and home country is used to include the relevant importance of foreign banks (e.g. when there are two banks from the home country present in the host country).

Since the baseline regression uses a binary dependent variable, caution should be taken with the regression technique. Standard OLS regression does not account for binary dependent variables and is not appropriate for this kind of regression. According to Hosmer et al. (2013) a logistic regression (logit model) is the most appropriate regression technique when the dependent variables takes on either 1 or 0. Logistic regressions are notated as:

logitP ( = 1) = ↵ + 1⇤ X1+ ..n⇤ X..n+ ✏ij (1)

Where logitP ( = 1) denotes the chance that the dependent variable equals 1 and

..n⇤X..nrefers to the coefficient and variable of interest with ’n’ total variables included

to the regression. To test the hypotheses (see section II.D and II.E), equation (1) is used to assess wether the hypotheses can be rejected or not. F BPij captures the chance that

a foreing bank from the home country ’i’ is located in the host country ’j’ in country pair ’ij’ and takes on 1 if their is a foreign bank and 0 if there is none.

F BPij = ↵ + 1⇤Bij+ 2⇤Mi+ 3⇤Mj+ 1(Bij⇤Mi) + 2(Bij⇤Mj) + 1⇤Xij+ ✏ij (2)

Furthermore, ↵ indicates a constant term, Bij indicates the bilateral distance

measure-ments between the host and home country. That is geographical distance, informational distance, legal distance and regulatory distance. Mi and Mj refer to the technological

progress in the host and home country respectively. (Bij ⇤ Mi) and (Bij ⇤ Mj) capture

(24)

1, 2, 3, 1 and 2 are the coefficients of interest, and 1 is a vector of coefficients, Xij

is a matrix of control variables, including home and host country characteristics like the size of the country and bilateral aspects such as having a common language, and ✏ij is the error term.

IV. Empirical Results

A. Preliminary Analysis

UNIVARIABLE ANALYSES

Before a logistic regression can be applied, the data set and the variables have to be controlled if they are appropriate to use and comply with the rules of a logistic regres-sion. Therefore, a univariable analysis is conducted to test if the explanatory variables of interest have a relationship with foreign bank presence (dependent variable) in iso-lation. In the case an explanatory variable is not significant on its own, there is no need to include this variable to the multivariate regression (The University of Sydney, 2017). The results of the univariable analysis are attached to the appendix (Appendix B1). Next, to the explanatory variables, the variables for technological progress in the home country and the technological progress in the host country are included in the univariable analysis.

The univariable analysis indicates that all the explanatory variables (geographical dis-tance, legal disdis-tance, regulatory distance and informational distance) are significant. The two variables that measure technological progress at the home and host country also show up significant. Therefore, all the variables can safely be included in the mul-tivariable analysis.

(25)

COLLINEARITY

Correlation and collinearity among the explanatory variables can lead to serious prob-lems in the estimation of the multivariable regression. Since their explanation in the outcomes are co-variated and have the same variability (Hosmer et al., 2013). To test if collinearity is present among the variables used in this data set, a correlation matrix is created, and a variance inflation factor (VIF) test is conducted. The correlation matrix (see Appendix A3) reveals that there are no serious correlation problems. All correlation values are below .50. Also when adding the control variables, the correlation among the control variables is not leading to serious correlation problems.

Collinearity is tested by conducting a VIF test10_{. All variables have a VIF value that}

is between 0 and 2. It is commonly accepted that a VIF higher than 10 is indicating evidence for collinearity.

B. Multivariable Results

Baseline Regression

To examine if the explanatory variables have a significant relationship with foreign bank presence, a baseline regression is constructed. The results of the baseline regression are shown in Table 2. The baseline results consist out of four di↵erent models. Model 1, reveals the results of equation (2)(see section III.D) without the interaction variables of interest and the technological progress variables. Model 2 adds the control variables to the regression in model 1. Model 3 adds the technological progress in the home and host country with the explanatory variables, but without the control variables and model, 4 combines the variables of model 1, 2 and 3.

The baseline regression results in model 1 show that both geographical distance and informational distance are significant and negatively correlated with foreign bank pres-ence of the home country in the host country. This implicates that the more physical distance there is between two countries, the less likely there will be a foreign bank present in the host country from the home country. This is in line with the findings in existing literate were the distance (both physical and informational) is negatively as-sociated with foreign bank presence (see for e.g. Buch, 2005; Claessens and van Horen, 2008). The other two variables, legal distance and regulatory distance, have a positive and negative relationship with the presence of a foreign bank but remain insignificant. This is contrary to the existing literature, were legal distance and regulatory distance

10_{To test if there is collinearity among the explanatory variables an OLS regression is conducted.}

(26)

Table 2: Baseline Regression Results

(1) (2) (3) (4)

Explanatory Explanatory and

VARIABLES Explanatory and controls Technological Progress Combined

Geographic distance -0.808*** -0.880*** -0.775*** -0.907*** (0.0447) (0.0695) (0.0512) (0.0748) Informational distance -0.0692*** -0.0155 -0.0652*** -0.0482** (0.0189) (0.0205) (0.0215) (0.0218) Legal distance 0.0248 0.0456** 0.0162 0.0315 (0.0204) (0.0213) (0.0205) (0.0216) Regulatory distance -0.0473 -0.0475 -0.0921 -0.133** (0.0623) (0.0655) (0.0643) (0.0676) Technological progress 0.748*** 0.678*** home country (0.0753) (0.0880) Technological progress -0.108** -0.394*** host country (0.0469) (0.0581) Equal language 1.179*** 1.269*** (0.413) (0.409) Equal region 0.625*** 0.708*** (0.133) (0.134) Equal currency 0.217 0.198 (0.214) (0.225) GDP host country 0.209*** 0.293*** (0.0307) (0.0318) GDP home country 0.801*** 0.820*** (0.0380) (0.0420) Constant 4.272*** -22.93*** -3.042** -28.66*** (0.348) (1.436) (1.272) (2.028) Observations 8,386 8,386 8,386 8,386

(27)

are presented as significant variables in the explanation of foreign bank presence (see for e.g. Claessens and van Horen, 2014a). There can be multiple reasons, why legal dis-tance and regulatory disdis-tance are not significant in this study. First, the time sample is di↵erent to that of most other studies. Claessens and van Horen (2014a) found signifi-cant results for 2009. Second, legal distance and regulatory distance can be measured using di↵erent proxies to evaluate the importance of these variables in explaining for-eign bank presence. It, thus, shows that distance variables and in particular measuring legal and regulatory distance should be taken with caution. The robustness test will examine if there are di↵erences in the significance of the explanatory variables when using a di↵erent year for the sample and other regression techniques.

To examine the strength of the explanatory variables, control variables are added in model 2. When the home country and the host country share an equal language, there is a significantly higher chance that a foreign bank is located in the host country. Also, when the home country and the host country are located within the same geographical region, foreign bank presence is more likely. The baseline regression results reveal that having an equal currency is not significant in explaining the presence of a foreign bank between the host and home country. Therefore, the currency control variables will be dropped in further analysis. The results in table2 also show that economic size of both the home and host country are important in explaining the presence of a foreign bank. Since both GDP home and GDP host have a positive sign and are significant, having a larger economy (either the home and host country) leads to a higher chance in foreign bank presence.

Model 3 adds the technological progress variables to model 1 and model 4 includes all the variables. Both model 3 and 4 show that technological progress in the home and host country is significant at the 1 percent level. Technological progress in the home country is positively related to foreign bank presence. It, therefore, suggest that the higher the technological progress of the home country the higher the change in foreign bank presence in the host country. This is in line with hypothesis 1A. There is, however, no evidence that hypothesis 1B is true. Technological progress in the host country leads to lower levels of foreign bank presence. This could be because more distant banks have more difficulties adapting to the high level of technology in the host country. Model 4 shows that geographical distance remains negative and significant at the 1 percent level, while of a slightly larger negative magnitude. Informational distance and regulatory distance also are negative but only significant at the 5 percent level. Legal distance is not significant.

(28)

combined model (4) implicates that informational distance and regulatory distance also explain the presence of foreign banks.

Regression Results With Interaction Variables

Table 3 shows the results of the multivariable analysis using a logistic regression with the interaction terms between the explanatory variables of interest and technological progress in the home and host country. Using these interaction terms allows this model to explain the e↵ects of technological progress in the importance of bilateral distance in foreign banking (see Claessens and van Horen, 2014a; Menard, 2003). Model 1 to 4 show the results for the explanatory variables and the interaction variable for each distance variable on its own (also see the univariable analysis at Appendix B1). Model 5 includes all variables, tested in model 1-4 combined. All models 1 to 5 include control variables, but these are not shown in table 3 (for the complete results overview see Appendix D1). Model 1 shows that adding the interaction term to the regression, geographical dis-tance, while, of a larger magnitude (due to the logistic regression and the interaction terms) remains negative and significant at the 1 percent level. The interaction terms that combine geographical distance and the technological progress in the home and host country are positive and significant at the 1 percent level. While the magnitudes of the variables have to be interpreted with caution11_{, the significant signs they have, can be}

interpreted (Menard, 2003).

The positive signs of the interaction term between geographical distance and technologi-cal progress at the home and host country suggest that technologitechnologi-cal progress does have an e↵ect on the relationship between geographical distance and the presence of foreign banks. The higher the technological progress in the home and host country, the more likely a distant foreign bank from a home country will be present in the host country. These results are in line with hypothesis 2A and 2B (see section II. D). Model 1 thus provides evidence of the hypothesis that banks from home countries with a high level of technological progress probably have better technology to overcome the difficulties that arise with geographical distance. New technologies like internet banking have a large impact on how banks operate (Martins et al., 2014). Model 1 also provides evidence of the hypothesis that banks in host countries with a high level of technological progress find it easier to overcome the difficulties that arise with geographical distance.

Informational distance and regulatory distance are both insignificant in Model 2 and 4. While the interaction term between the technological development of the host country and informational distance is significant at the 5 percent level, it cannot be interpreted

11_{Since a logistic regression is used, and the geographical distance and technological progress are}

(29)

Table 3: Regression Results

(1) (2) (3) (4) (5)

VARIABLES Geographical Informational Legal Regulatory Combined

Geographical distance -8.174*** -6.640***

(1.078) (0.685)

Geographical distance 0.318*** 0.309***

* Technological progress home (0.0938) (0.0632)

* Technological progress host (0.0615) (0.0412)

Informational distance -0.203 -0.121

(0.399) (0.286)

Informational distance 0.0369** 0.0364***

Legal distance -1.287*** -0.601*

(0.470) (0.322)

Legal distance 0.0792** 0.0135

Legal distance 0.0385* 0.0422***

Regulatory distance -0.166 2.122*

(1.568) (1.145)

Regulatory distance -0.0128 -0.114

Regulatory distance 0.00656 -0.0881*

Technological progress home -1.839** 0.814*** 0.551*** 0.808*** -1.707***

(0.769) (0.110) (0.130) (0.143) (0.497)

Technological progress host -3.170*** -0.327*** -0.310*** -0.260*** -2.157***

(0.515) (0.0739) (0.0789) (0.0908) (0.317)

Constant 28.19*** -34.87*** -32.05*** -35.67*** 19.81***

(8.574) (1.927) (2.079) (2.008) (5.132)

Observations 8,386 8,386 8,386 8,386 8,386

Control variables Yes Yes Yes Yes Yes

(30)

since informational distance is not found to be a significant variable in explaining foreign bank presence.

Legal distance, however, in model 3, is negatively correlated with foreign bank pres-ence and significant at the 1 percent level. The interaction term between legal distance and technological progress home is positive and significant at the 5 percent level. The interaction term between legal distance and technological progress host is positive as well but only significant at the 10 percent level. Model 3 thus provides evidence in line with hypothesis 3A and 3B. It suggests that foreign banks from home countries with a higher level of technological progress find it easier to deal with legal distances between the host and home country.

Model 5 in Table 3 combines all the regressions from model 1 to 4. When adding all the variables to the regression, only geographical distance, legal distance and regulatory distance remain significant at respectively the 1, 10 and 10 percent significance level. The interaction e↵ect between geographical distance and technological progress at the home country remains significant and positive. So the evidence from model 1 still hold and are in line with hypothesis 2A and 2A. The interaction e↵ect between geographical distance and technological progress at the home country remains significant and nega-tive. Thus the evidence from model 1 also holds and is line with hypothesis 2B. Legal distance also remains negative. However, the interaction term between legal distance and technological progress in the home country is not significant. The interaction term between legal distance and technological progress in the host country is positive and significant at the 1 percent level. This indicates that in countries with a high level of technological progress, legal distance is less important in explaining foreign bank presence in that country. Model 5 also shows that regulatory distance is important in explaining foreign bank presence in the host country.

C. Robustness Check

To test the strength of the regression results in section B., several robustness checks are conducted. First, the baseline regression results (see table 2) are tested with several robustness tests. Table C. shows the regression results of four di↵erent tests. All models are based on the combined model shown in table 2 under model 5 which includes all variables and the control variables12_.

Model 1 shows the results based on data from 2007. Compared to model 5 in table 2, only geographical distance is significant in explaining foreign bank presence. These results thus are in line with the previous assumption that studies explaining the im-portance of distance in the presence of foreign banks might be biased by the year or

12_{Table 4 does not show the control variables. For a complete overview of the baseline robustness}

(31)

years they are using. It also suggests that, while, the importance of legal distance, in-formational distance and regulatory distance remain ambiguous, geographical distance is important in explaining foreign bank presence.

Model 2 therefore, uses the average values of the variables between 2008 and 2013. These results are more in line with the baseline regression results. Geographical distance and informational distance are significant and both negative. Suggesting that the more distance there is between two countries the less likely a foreign bank presence.

Table 4: Robustness Tests on the Baseline Results

(1) (2) (3) (4)

VARIABLES 2007 2008-2013 Poisson Tobit

Geographical distance -0.840*** -0.870*** -0.834*** -1.370*** (0.0720) (0.0679) (0.0431) (0.109) Legal distance 0.00979 0.0133 0.0188 0.0326 (0.0227) (0.0215) (0.0148) (0.0329) Informational distance -0.0297 -0.0551*** -0.0303** -0.0481 (0.0215) (0.0206) (0.0140) (0.0309) Regulatory distance -0.0704 -0.0986 -0.0733 -0.225** (0.0687) (0.0652) (0.0467) (0.0992)

Technological progress Home 0.797*** 0.737*** 0.625*** 0.873***

(0.0789) (0.0795) (0.0578) (0.116)

Technological progress Host -0.333*** -0.393*** -0.288*** -0.497***

(0.0600) (0.0547) (0.0385) (0.0832)

Constant -32.09*** -28.26*** -26.44*** -40.84***

(1.701) (1.619) (1.102) (2.587)

Observations 8,386 8,386 8,386 8,386

Control variables Yes Yes Yes Yes

Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

(32)

is a continuous variable including a lot of zero’s (McDonald and Moffitt, 1980).

Table 5 shows the same models and variables as in table 3, however, the dependent variable in table 5 represents the number of foreign banks instead of a binary value. Model 1-4 show that both geographical distance and legal distance are robust since these variables are still significant when testing for the number of foreign banks. Model 5 specifies the combined variables. Geographical distance is still negative and significant at the 1 percent level. Also, the interaction terms between geographical distance and technological progress in the home and host country remain positive and significant at the 1 percent level. The robustness check, therefore, confirms the findings of the regression analysis in Table 3. When the home country of the foreign bank has a high level of technological progress, the bank is more likely to expand over large distance than foreign banks from home countries with less technological progress. As well, when the host country has a high level of technological progress, it is more likely that banks from distant geographical countries would enter the domestic banking market.

Table 6 includes several other robustness checks. For this, model 5 from table 5 is used as the full sample regression. Model 1-5 include control variables, but these are not shown. For the complete results, including the control variables, see Appendix D4. Model 1 uses the binary dependent variable but for 2007 to test whether variables are significantly di↵erent between 2013 and 2007. Using data from 2007, technolog-ical progress at the home country is no longer significant in explaining foreign bank presence in the host country. As well as, technological progress is no longer significant in explaining the importance of geographical distance in foreign banking. This thus suggests that using data from di↵erent years may lead to di↵erent conclusions.

Model 2 partly accounts for this problem by using aggregate data of the years 2008-2013. While the regression is still based on cross-sectional data, using aggregates nets-out yearly a↵ects13_.

Model 3 to 5 use the weighted dependent variable that measures the number of foreign banks from the home country active in the host country. Model 3-4 are based on data from 2007 and model 5 on data from 2013, like the multivariable results in section IV.B. The robustness tests suggest that there are di↵erences across the results when using di↵erent samples and regression techniques. While geographical distance remains an important variable in the explanation of foreign bank presence, the interaction e↵ect between geographical distance and technological progress at the home country is only significant in model 2,3 and 5. However, the interaction term between geographical distance and technological progress in the host country is positive and significant at the 1 percent level among all the robustness tests. Foreign banks that expand to host

(33)

Table 5: Robustness Test on the Regression Results

(1) (2) (3) (4) (5)

VARIABLES Geographical Informational Legal Regulatory Combined

Geographical distance -6.875*** -6.640***

(0.642) (0.685)

(0.289) (0.286)

Informational distance 0.0486*** 0.0364***

Legal distance -1.198*** -0.601*

(0.333) (0.322)

Legal distance 0.0564** 0.0135

Legal distance 0.0525*** 0.0422***

Regulatory distance -0.367 2.122*

(1.118) (1.145)

Regulatory distance 0.00368 -0.114

Regulatory distance 0.0118 -0.0881*

Technological progress home -1.763*** 0.790*** 0.608*** 0.763*** -1.707***

(0.490) (0.0766) (0.0910) (0.102) (0.497)

Technological progress host -2.201*** -0.252*** -0.255*** -0.167*** -2.157***

(0.307) (0.0490) (0.0542) (0.0644) (0.317)

Constant 20.42*** -32.83*** -30.75*** -33.61*** 19.81***

(5.042) (1.278) (1.433) (1.332) (5.132)

Observations 8,386 8,386 8,386 8,386 8,386

Control variables Yes Yes Yes Yes Yes

(34)

Table 6: Multiple Robustness Tests on the Regression Results

(1) (2) (3) (4) (5)

Logit Logit Poisson Tobit Tobit

VARIABLES 2007 2008-2013 2007 2007 2013

Geographical distance -5.702*** -7.916*** -4.683*** -7.809*** -10.61*** (1.125) (1.084) (0.740) (1.716) (1.540)

Geographical distance 0.0898 0.284*** 0.123* 0.123 0.416***

* Technological progress home (0.0940) (0.0926) (0.0633) (0.141) (0.132) Geographical distance 0.358*** 0.361*** 0.222*** 0.466*** 0.430*** * Technological progress host (0.0669) (0.0638) (0.0420) (0.104) (0.0953)

Informational distance 0.420 -0.0282 -0.0747 0.490 0.191

(0.445) (0.403) (0.318) (0.643) (0.548) Informational distance -0.0480 -0.0217 -0.0286 -0.0858* -0.0668 * Technological progress home (0.0361) (0.0331) (0.0263) (0.0517) (0.0446) Informational distance 0.00996 0.0231 0.0374*** 0.0452 0.0518** * Technological progress host (0.0181) (0.0173) (0.0121) (0.0278) (0.0255)

Legal distance -0.351 -0.831* -0.345 -0.622 -0.936

(0.479) (0.462) (0.338) (0.723) (0.663)

Legal distance 0.0125 0.0517 0.00416 0.0158 0.0420

* Technological progress home (0.0376) (0.0374) (0.0270) (0.0562) (0.0536)

Legal distance 0.0201 0.0233 0.0289* 0.0426 0.0452

* Technological progress host (0.0231) (0.0218) (0.0156) (0.0359) (0.0328)

Regulatory distance 3.520** 2.706* 3.207*** 5.187** 4.261*

(1.574) (1.579) (1.106) (2.297) (2.226)

Regulatory distance -0.153 -0.107 -0.139 -0.224 -0.230

* Technological progress home (0.119) (0.122) (0.0847) (0.174) (0.171) Regulatory distance -0.181*** -0.156** -0.159*** -0.273*** -0.186* * Technological progress host (0.0682) (0.0649) (0.0479) (0.106) (0.0985) Technological progress home 0.262 -1.547** -0.107 0.490 -2.223** (0.764) (0.740) (0.505) (1.156) (1.058) Technological progress host -3.204*** -3.329*** -2.022*** -4.273*** -4.145***

(0.532) (0.509) (0.323) (0.839) (0.766)

Constant 4.068 28.07*** 0.950 -0.349 32.15***

(8.815) (8.267) (5.701) (13.37) (11.64)

Observations 8,386 8,386 8,386 8,386 8,386

(35)

countries with well-developed technological resources and capabilities, therefore, have fewer difficulties with geographical distance.

Model 1 to 5 also reveals that regulatory distance is important in explaining foreign bank presence. Regulatory distance is significant in all the robustness models. The interaction term between regulatory distance and technological progress in the host country is significant among all the models as well. This is in line with the regression results of table 3. The next section will elaborate on the findings of the empirical research.

Results Discussion

Section IV.B first elaborated on the baseline regression results. In the baseline regression results, geographical, informational and regulatory distance are significant determinants in explaining foreign bank presence. The negative signs of the coefficients suggest that an increase in geographical, informational and regulatory distance results in a lower chance of foreign bank presence from the more remote home country. Technological progress in both the home and host country also are significant determinants in ex-plaining foreign bank presence in the host country.

After the robustness tests, the baseline regression suggests that geographical distance and informational distance have a significant e↵ect on foreign bank presence. Also technological progress in the home and host country remain significant. This provides evidence that foreign banks that come from home countries with a high level of techno-logical progress have better technotechno-logical capabilities and resources to expand abroad. A higher level of technological development in the host country, however, leads to a lower presence of foreign banks. This is contrary to hypothesis 1B. A possible reason for this could be that when domestic banks in the foreign country have good technolog-ical capabilities and resources, these banks likely be more efficient. Therefore, foreign banks might find it hard to enter the foreign country. This is in line with the findings of Claessens and van Horen (2014a) who argue that competitive advantage is important in explaining foreign bank presence.

Foreign Banks and the Importance of Bilateral Distance. The Impact of Technological Progress.