THREE ESSAYS IN INTERNATIONAL ECONOMICS

By

Hannah Claire Gabriel

A DISSERTATION

to Michigan State University

in partial fulfillment of the requirements

Submitted

for the degree of

Economics – Doctor of Philosophy

2020

ABSTRACT

THREE ESSAYS IN INTERNATIONAL ECONOMICS

By

Hannah Claire Gabriel

Chapter 1: This paper contributes to a growing literature on the effects of credit constraints on
international trade, and an existing body of literature on the exporting advantage of multinational
firms. Using 2013 data from Estonian and Hungarian exporting firms, I find that traditional
measures of credit constraints (cash flows, debt to sales ratio, and tangible asset share) have a
significant negative effect on the intensive margin of international trade, and being a multinational
affiliate has a positive effect on trade. Multinational affiliates export nearly twice as much as
domestic firms. I find no strong evidence that multinational affiliates are less credit constrained
than domestic firms conditional on firms already exporting. Therefore, any differences between
the two types of firms appear on the extensive margin of trade or in domestic activities. Estonia
and Hungary are relatively recent EU members that experienced an influx of foreign investment
during their transition periods in the 1990s. Therefore, these results provide useful information for
studying the long-term benefits of EU accession and foreign investment in transition economies.

Chapter 2: This paper analyzes the impact of multinational banking and multinational ownership
on the performance of exporting Central European firms. Using a panel of Hungarian, Croatian,
and Estonian exporters, I find that controlling for a firm’s borrowing behavior leads to a 12.5%
reduction in the coefficient on multinational status. Such a non-trivial amount indicates that the
“multinational advantage" in export revenue falls from 122% to 102%. This outcome is strongest
for Hungarian firms and weakest for Croatian firms, indicating a long-run benefit to increases in
foreign banking among recent EU members. Additionally, I find that the Hungarian government’s
effort to increase domestic banking presence by purchasing two major multinational banks, MKB
and Budapest Bank, led to a $270,000 decrease in loans among domestically owned exporters.
However, this purchase had no effect on the loans of multinational affiliates. These results provide

evidence that multinational affiliates are better able to smooth their borrowing behavior in the
presence of a tumultuous banking sector, and that the firms most affected by anti-global banking
policies are smaller, locally owned firms.

Chapter 3:

In 2014, the Austrian bank Hypo Group Alpe Adria went bankrupt and was
purchased and rebranded as Addiko bank. The bank was purchased by the American banking
group Advent International, and the European Bank for Reconstruction and Development (EBRD),
an international organization that serves as an investment bank in former transition economies.
Their primary goal is to assist European transition countries in establishing or enriching a market-
based economy.
In this paper, I explore how this bailout of Addiko Bank by an international
organization affected exporting firms in Croatia. Specifically, I investigate whether there was a
positive effect on firm performance, therefore justifying the need for intervention by an international
financial institution. I find that the turnover of Addiko bank led to a $260,000 decrease in loans
taken out by firms. However, this effect seems to occur immediately after the turnover, and vanishes
over time. Additionally, I find no effect on the export revenue, and total revenue of firms, and a
small increase in the domestic revenue amongst the firms. These results indicate that after an initial
period of turmoil, the intervention by EBRD and Advent International had no lasting negative (and
perhaps slightly positive) effects on firm outcomes.

Copyright by
HANNAH CLAIRE GABRIEL
2020

ACKNOWLEDGEMENTS

I would like to thank my advisor Raoul Minetti, and committee members: Susan Chun Zhu,
Qingqing Cao, and Hao Jiang as well as Red Cedar committee member Luis de Araujo for their
invaluable help and guidance during my time at MSU. Additionally I would like to acknowledge
the Michigan State Macro Group, Michigan State Applied Economics Workshop, the members of
my cohort, my office-mates in Berkey 5L, Susan J. Linz, Paul E. Gabriel and Susanne Schmitz for
their insightful suggestions and feedback on all chapters of this dissertation. Special thanks go to
D.A.G for her moral support while writing this dissertation. All errors are my own.

v

TABLE OF CONTENTS

LIST OF TABLES .
LIST OF FIGURES .

.
.

.
.

.
.

.
.

.
.

.
.

.
.

.
.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

x

CHAPTER 1 THE ADVANTAGE OF MULTINATIONAL FIRMS UNDER CREDIT

.
.

.
.

. .

Introduction .

.
1.1
1.2 Review of Literature .

Investment in Central Europe and Historical Context

CONSTRAINTS: EVIDENCE FROM ESTONIA AND HUNGARY . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2.1
. . . . . . . . . . . .
1.2.2 Exporting Under Financial Constraints . . . . . . . . . . . . . . . . . . . .
1.2.3 Multinational Activity Under Financial Constraints . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1
1
2
2
3
4
5
7
9
1.5.1 Results by Country (Robustness) . . . . . . . . . . . . . . . . . . . . . . . 11
.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

1.3 Data .
.
1.4 Empirical Specification .
1.5 Results .
.

1.6 Conclusion .
APPENDIX .

.
. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

.
.

.

.
.

.
.

.

.

.

.

.

CHAPTER 2 MULTINATIONAL ACTIVITY AND BANKING: THE EFFECTS OF

.

.
.

.
.

. .

. .

. .

. .

Introduction .

.
2.1
2.2 Review of Literature .

2.3.1 Estonia
2.3.2 Hungary .
2.3.3 Croatia
. .

OWNERSHIP STATUS ON EXPORTERS . . . . . . . . . . . . . . . . . . 24
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.2.1
Investment in Central Europe . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.2.2 Multinational Activity and Export Under Financial Constraints . . . . . . . 27
2.2.3 Multinational Banking . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.3 Background and Institutional Details . . . . . . . . . . . . . . . . . . . . . . . . . 29
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.4 Data .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.5 Descriptive Motivation and Conceptual Framework . . . . . . . . . . . . . . . . . 35
2.5.1 Conceptual Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.5.2 Descriptive Evidence and Omitted Variable Bias
. . . . . . . . . . . . . . 36
. .
2.5.3 Results
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.5.3.1
Exports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.5.3.2 Domestic Revenue . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.5.4 Robustness Checks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
. . . . . . . . . . . . . . . 42
2.6.1 Difference-in-Differences Framework . . . . . . . . . . . . . . . . . . . . 42
. . . . . . . . . . . . . . . . . . . . . . 44
2.6.2 Difference-in-Differences Results

2.6 Causal Effects of Bank Turnover on Firm Performance

.
. .
.
.

. .
.
.

. .
. .

. .

. .

.

.

vi

2.7 Conclusion . .
APPENDIX . .

. .

. .

. .
.

.
.

.
.

.
.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
.

CHAPTER 3 BANKRUPTCY AND INTERNATIONAL INTERVENTION: THE CASE

.

.

.

.

.

.

. .

Introduction .

OF ADDIKO BANK . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
3.1
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
3.2 Review of Literature and Background . . . . . . . . . . . . . . . . . . . . . . . . 82
3.2.1 Multinational Banking . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
.
3.2.2 The EBRD .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. 86
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100

.
3.3 Data .
.
3.4 Empirical Specification .
3.5 Results .
.
Firm Switching .
.
.

3.6 Conclusion .
APPENDIX .

3.5.1

.
. .

.

.
.

.

.
.

.

.
.

.

. .

. .

. .

. .

. .

. .

. .

. .

.

.

. .

BIBLIOGRAPHY .

.

.

.

.

vii

LIST OF TABLES

Table 1.1 Overall Firm Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

Table 1.2 Summary Statistics .

.

.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

Table 1.3 Firm Industry (NACE Main) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

Table 1.4 Baseline Results

.

.

.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

Table 1.5 Baseline Results with Standardized Financial Variables . . . . . . . . . . . . . . 18

Table 1.6 Baseline Results with Interaction Terms . . . . . . . . . . . . . . . . . . . . . . 19

Table 1.7 Results by Country .

.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

Table 1.8 Results by Country with Interaction Terms . . . . . . . . . . . . . . . . . . . . . 21

Table 1.9

ISO Country Code Key .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

Table 2.1 Summary Statistics- Firm Characteristics

. . . . . . . . . . . . . . . . . . . . . 48

Table 2.2 Summary Statistics- Banking Characteristics

. . . . . . . . . . . . . . . . . . . 49

Table 2.3 List of Banks–Hungary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

Table 2.4 List of Banks–Estonia

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

Table 2.5 List of Banks–Croatia

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

Table 2.6 Most frequent foreign owners – Exporters . . . . . . . . . . . . . . . . . . . . . 52

Table 2.7 Most Frequent Banks for Multinational Affiliates . . . . . . . . . . . . . . . . . 53

Table 2.8 Firm Industry (NACE Main) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

Table 2.9 Export Results for All Firms . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

Table 2.10 Export Results for Hungarian Firms . . . . . . . . . . . . . . . . . . . . . . . . 56

Table 2.11 Export Results for Estonian Firms . . . . . . . . . . . . . . . . . . . . . . . . . 57

Table 2.12 Export Results for Croatian Firms . . . . . . . . . . . . . . . . . . . . . . . . . 58
Table 2.13 Export Results for Small/Medium Enterprises (≤ 250 Employees)

. . . . . . . . 59

viii

Table 2.14 Export Results for Large Firms (> 250 Employees) . . . . . . . . . . . . . . . . 60

Table 2.15 Domestic Results for All Firms . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

Table 2.16 Domestic Results by Country . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

Table 2.17 Domestic Results by Firm Size . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

Table 2.18 Difference in Differences Results for All Firms

. . . . . . . . . . . . . . . . . . 64

Table 2.19 Difference in Differences Results for Domestic Firms . . . . . . . . . . . . . . . 65

Table 2.20 Difference in Differences Results for Multinational Affiliates . . . . . . . . . . . 66

Table 2.21 Size and History of Banks- Hungary . . . . . . . . . . . . . . . . . . . . . . . . 72

Table 2.22 Size and History of Banks- Estonia

. . . . . . . . . . . . . . . . . . . . . . . . 73

Table 2.23 Size and History of Banks- Croatia . . . . . . . . . . . . . . . . . . . . . . . . . 74

Table 2.24 Mergers and Acquisitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

Table 2.25 Robustness Check:Total Revenue Results for All Firms . . . . . . . . . . . . . . 76

Table 2.26 Robustness Check:Total Revenue Results by Country . . . . . . . . . . . . . . . 77

Table 2.27 Robustness Check:Total Revenue Results by Firm Size . . . . . . . . . . . . . . 78

Table 2.28 Robustness Check: Export Results for All Firms (Controlling for Bank Country)

79

Table 2.29 Robustness Check: Export Results for All Firms (Different MNA Definition)

. . 80

Table 3.1 Summary Statistics- Firm Characteristics

. . . . . . . . . . . . . . . . . . . . . 91

Table 3.2 List of Banks .

.

.

.

.

.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

Table 3.3 Firm Industry (NACE Main) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

Table 3.4 Difference in Differences Results for Loans . . . . . . . . . . . . . . . . . . . . 94

Table 3.5 Difference in Differences Results (Revenue) . . . . . . . . . . . . . . . . . . . . 95

Table 3.6 Difference in Differences Results for Export Share

. . . . . . . . . . . . . . . . 96

Table 3.7 Size and History of Banks- Croatia . . . . . . . . . . . . . . . . . . . . . . . . . 99

ix

LIST OF FIGURES

Figure 1.1 Global Ultimate Owner Countries . . . . . . . . . . . . . . . . . . . . . . . .

. 22

Figure 2.1 Example From the Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. 67

Figure 2.2 Event Study results for Exports (All Firms) . . . . . . . . . . . . . . . . . . . . 68

Figure 2.3 Event Study results for Loans (All Firms) . . . . . . . . . . . . . . . . . . . .

. 69

Figure 2.4 Event Study results for Exports (Domestic) . . . . . . . . . . . . . . . . . . .

. 69

Figure 2.5 Event Study results for Loans (Domestic) . . . . . . . . . . . . . . . . . . . .

. 70

Figure 2.6 Event Study results for Exports (MNA) . . . . . . . . . . . . . . . . . . . . .

. 70

Figure 2.7 Event Study results for Loans (MNA) . . . . . . . . . . . . . . . . . . . . . .

. 71

Figure 3.1 Event Study results for Loans . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

Figure 3.2 Event Study results for Exports . . . . . . . . . . . . . . . . . . . . . . . . . . 97

Figure 3.3 Event Study results for Total Revenue

. . . . . . . . . . . . . . . . . . . . . . 98

Figure 3.4 Event Study results for Domestic Revenue . . . . . . . . . . . . . . . . . . . . 98

x

CHAPTER 1

THE ADVANTAGE OF MULTINATIONAL FIRMS UNDER CREDIT CONSTRAINTS:

EVIDENCE FROM ESTONIA AND HUNGARY

1.1 Introduction

Multinational firms tend to be larger and more efficient than their domestic counterparts, and the
size and scope of these firms have increased over time1. Consequently, they are also more resilient
than domestic firms during periods of economic hardship. There are several explanations for
this phenomenon, and the main question explored in this paper is whether or not the affiliates of
large multinational firms perform better in the presence of credit constraints when compared with
domestic firms. One possible explanation for this advantage is that multinational affiliates have
access to different, or rather more, means of financing their operations. They are privy to financing
through themselves, and through the resources of their parent company.

There are significant startup costs for a firm to engage in international trade. If a firm cannot
gain access to adequate financing, it can significantly hamper their export decisions on the extensive
and intensive margins. This problem is more prevalent for firms in countries that are less financially
developed 2. Thus, it is logical that an affiliate of a multinational enterprise that can use financial
channels afforded to it by the parent company is at an advantage to engage in trade over a domestic
firm without these additional means of financing.

In this paper I analyze two Central European economies (CEE): Estonia and Hungary. Both are
European Union members, having joined in 2004, that experienced increases in foreign investment
during the time period around their EU accession. However, this occurrence is not unique to Estonia
and Hungary as other nations, including Slovakia and the Czech Republic, have experienced a
similar influx of foreign investment surrounding EU accession. There are several reasons for

1Markusen and Venables (1998)
2Manova (2013)

1

this inflow of capital including greater access to other EU markets through the lower-cost Central
European members and access to an emerging, and later rapidly developing, market in the CEE. This
makes Central Europe an interesting region to study because it has experienced rapid development
while remaining a relatively “low-cost" area in which to invest.

The remainder of this paper proceeds as follows: section two provides a review of literature,
section three describes the data, section four specifies an empirical design, section five presents the
results, and section six concludes and discusses avenues for further research. Tables and figures
are presented in section seven.

1.2 Review of Literature

The previous literature on this topic is largely divided into two categories: studies of the effects
that financial constraints have on exporting, and studies that explore the behavior of multinational
firms under imperfect capital markets. An additional body of literature that is relevant to this paper
covers foreign investment in Central Europe.

1.2.1

Investment in Central Europe and Historical Context

After the Iron Curtain fell in the late 1980s, several Central European countries applied for EU
membership. In 2004, Hungary and Estonia joined the EU along with eight other Central European
countries, resulting in a significant enlargement of the European Union (EUROPA). During their
EU candidacy phase, there was a large increase in foreign investment in Central European countries.
The candidacy phase for most Central European economies also corresponded with the country’s
transitions from planned economies to a market economies.
It is important to understand why
Central Europe saw an increase in foreign investment in the transition and candidacy period of the
1990s and early 2000s.

Bevin and Estrin (2004) find that public announcements of EU membership negotiations had a
positive impact on foreign direct investment (FDI) in the candidate country. Additionally, Bevin and

2

Estrin conclude that the increase in foreign investment in central and eastern European countries
during the 1990s was due to lower skilled-labor costs, market size, and proximity. Unsurprisingly,
the authors find that Western European EU members such as Germany and Italy invested in Central
European economies disproportionally more than other large economies such as the United States.
Carstensen and Toubal (2004) find that the level of privatization in Central and Eastern Europe
affects the amount of foreign investment in a host country. Privatization is a key variable in the
attractiveness of a host country because a higher level of privatization indicates a more successful
transition to a market economy from a planned economy. Thus Carstensen and Toubal show that
Central European economies (Hungary, Poland, etc.), which were more privatized during this time
period, experienced more foreign investment than Eastern European economies (Belarus, Bulgaria,
etc.). Additionally, Lansbury, Pain, and Smidkova (1996) corroborate these results and show that
increased privatization in Central Europe, relative to Eastern Europe, led to the influx of foreign
investment by EU members in Western Europe.

More recently, a paper by Bilir, Chor, and Manova (2016) shows that when a multinational
enterprise is choosing an affiliate location, host country financial development plays a significant
role. Since Central European economies were more financially developed than Eastern European
economies during the transition period, they were considered a better location choice for the
multinational affiliates of large Western European MNAs.

1.2.2 Exporting Under Financial Constraints

There is a recent, and growing, literature on the effects of credit constraints on international
trade, particularly on a firm’s exporting behavior. In her 2013 paper, Kalina Manova constructs a
theoretical model that introduces credit constraints into a heterogeneous firm model (à la Melitz
(2003)). She finds that credit constraints affect a firm’s activities through three mechanisms:
domestic production, a firm’s selection into exporting (extensive margin), and the level of firm
exports (intensive margin). However, her paper uses aggregate level data to confirm the theoretical
predictions, as opposed to firm level micro-data, thus making it difficult to draw conclusions about

3

an individual firm’s export behavior under credit constraints.

Importantly, further studies have been conducted using firm level micro-data in different coun-
tries including Peru 3, Italy 4, and China 5, all of which conclude that credit constraints affect
exports both on the extensive and intensive margin. The wide range of countries used in these
studies gives credence to the significant role that credit constraints have on international trade, and
that this phenomenon is not regionally concentrated.

1.2.3 Multinational Activity Under Financial Constraints

Other studies have found that multinational affiliates are at an advantage over domestic firms
when financial frictions are present. Alfaro and Chen (2012) find that foreign owned firms fared
much better during the 2008 global financial crisis than domestic firms, indicating an advantage
of multinational affiliates during periods of economic hardship. Additionally, Desai, Fritz, and
Foley (2008) find that U.S. multinational affiliates respond much better to currency depreciations
in a host country than domestic firms. In particular, multinational affiliates increase their sales
and investment much more than the corresponding domestic firms. In their 2009 paper, Antras,
Desai, and Foley study the effects of financial contracting and investor protections (imperfect
capital markets) on a firm’s decision to engage in FDI. The authors show theoretically that FDI
flows and multinational activity arise endogenously for a parent company under financial frictions
and imperfect monitoring in the host country. Multinational activity occurring as a workaround
to imperfect capital markets indicates that there is an advantage to multinational firms over their
domestic counterparts.

Furthermore, Buch et al. (2014) use German micro-data, including balance sheet data, and
find that firms are less likely to engage in FDI when credit constraints are present. The effect is
strongest for the most productive firms as they are the most likely to engage in FDI. Their analysis

3Paravisini et al. (2014)
4Minetti and Zhu (2011)
5Feenstra, Li, and Yu (2014)

4

is conducted on the extensive margin, i.e. financial constraints lower the probability that a firm
chooses FDI. Interestingly, the authors find that unproductive firms who will never invest abroad
do not face this negative impact of financial frictions.

Lastly, a recent paper by Manova, Wei and Zhang (2015) addresses the question of multinational
activity under financial constraints using Chinese microdata. The authors do not have balance
sheet data available in their dataset, and thus cannot compute a measure of firm-level financial
vulnerability.
Instead, they study financial vulnerability at the sector level. Their results show
that while financial frictions (financially vulnerable industries) negatively affect exports, this effect
is not as large for multinational affiliates. This suggests that multinational affiliates have access
to better means of financing than domestic firms, and thus can circumvent the affect of domestic
market imperfections. However, multinational affiliates in China and Central Europe are likely very
different, thus it is important to analyze other areas of the world in order to see if thus phenomena
occurs elsewhere.

My paper contributes to the existing literature by examining a region that is historically known
to have a high level of foreign investment; it is important to know if foreign ownership gives firms
in Central Europe an advantage over domestic firms, particularly when firms are credit constrained.
In analyzing an area of the world that has largely been neglected by the existing literature on the role
of credit constraints in international trade, I am attempting to bridge the three branches of literature
mentioned above. Studying Central European firms and their ownership provides valuable insight
into the long-term benefits of foreign investment in transition economies and EU membership.

1.3 Data

The data used in this paper is from ORBIS, provided by Bureau van Dijk. I am using the data
collected from Estonia and Hungary, which are part of the AMADEUS (European) subset of ORBIS
for the year 2013. The firm-level data is collected annually from businesses through websites and
various business registries and reports. ORBIS is a well-known firm-level datasource and it is used

5

in several other papers on foreign direct investment and multinational enterprise behavior6.

One advantage of the ORBIS dataset is that it provides detailed balance sheet data for firms of
all sizes (including unlisted firms and publicly owned firms). Examples of these useful balance
sheet variables are: cash flows, external debt, and tangible fixed assets. These measures have all
been used in previous papers such as Buch et al. (2014) to proxy for financial constraints. A novel
aspect of the ORBIS data is that it provides detailed ownership information on each firm including
the country of the global ultimate owner in the case of a multinational affiliate. There is also an
independence indicator which shows how much autonomy a multinational affiliate has. These are
unique features that I exploit in my analysis. To define a multinational affiliate in my paper, I follow
the work of Cravino and Levchenko (2017). To be considered a multinational affiliate, a firm must
have a global ultimate owner (GUO) country that differs from the home country, and additionally,
a multinational affiliate will an independence indicator of “D". An independence indicator of “D"
means that the GUO shareholders have direct ownership over 50% of a Hungarian or Estonian firm.
I believe these firms have more direct access to the parent companies, and their resources, and are
thus at an advantage over other firms whose international shareholders have less direct control of
the Estonian or Hungarian firm.

Other useful variables from the dataset included in my analysis are: NACE industry classi-
fication, total employees and firm profitability (measured as Gross Profit/Sales). These provide
other controls for the firm that might also affect exporting behavior. Summary statistics for the key
variables for the exporting firms are presented below in Table 1.1 through Table 1.3.

Table 1.1 breaks the sample down by firm country; 60% of the firms are Estonian and 40%
are Hungarian. Table 1.1 also summarizes multinational affiliate status; 12% of the firms in the
sample are multinational affiliates and 88% are domestic firms. From Table 1.2, it is evident that
the Hungarian firms are larger in assets (fixed and total), employees, export revenue, and total
sales. However, Estonian firms are more profitable on average and the two countries have a similar
percentage of multinational affiliates: 14% for Hungarian firms and 10% for Estonian firms. When

6For example, Cravino and Levchenko (2017) and Buch et al. (2014)

6

separating the sample by firm ownership status in Table 1.2, it is clear that multinational affiliates are
larger than domestic firms in almost every regard, with the exception of profitability. Additionally,
multinational affiliates have a higher percentage of Hungarian firms than domestic firms.
It is
also apparent that export activity varies by firm sector. As shown in Table 1.3, the most common
industries for firms are wholesale and retail trade, manufacturing, and transportation and storage.
Figure 1.1 illustrates the countries with the largest number of multinational affiliates in this
dataset. Excluding Estonia and Hungary, the countries with the most affiliates are: Finland,
Sweden, Germany, and the United States. Unsurprisingly, other EU member countries (e.g. Italy,
France, and Great Britain– which was an EU member in 2013) also have several affiliates in Estonia
and Hungary. This phenomenon is consistent with the result of Bevin and Estrin (2004) that Western
European EU members invested more in Central Europe in the 1990s than North American and
Asian countries.

Currently, my sample only includes exporting firms. Hence, my analysis of the effect of credit
constraints and firm ownership on exports is limited to the intensive margin of international trade.
In the future I will include non-exporting firms to study these effects on the extensive margin of
trade.

1.4 Empirical Specification

It is reasonable to expect that credit constraints and firm ownership status would affect both
the extensive and intensive margins of trade. Additionally, multinational firms should have the
advantage on both margins as they are usually larger, more productive firms who should perform
better than domestic firms. Therefore, I propose two specifications: one to study the extensive
margin of trade (which will be included in a later draft of this paper) and one to study the intensive
margin.

My main specification for analyzing the effects of firm ownership status and credit constraints

7

on the intensive margin of international trade for firm f in industry i is:

LogExports f i = α + θM N A f + βX f + γFin f + ϕi + 
 f i

(1.1)

MNA is an indicator for firm f being a multinational affiliate. X f refers to the firm characteristics:
size, profitability, and an indicator for Hungarian firms. Fin f is a vector of financial variables that
act as proxies for financial constraints. The variables are: debt to sales ratio, tangible asset ratio,
and cash flows. The ϕi are industry dummies.

In the above specification, size is measured as the log of the number of employees at a firm.
Naturally, the expected sign on this variable is positive, indicating larger firms export more. Prof-
itability serves as a measure of firm quality, where more profitable firms are “better" and should
thus also export more. The Hungarian indicator accounts for differences between Hungarian and
Estonian firms, and the MNA firm ownership variable is the main characteristic of interest. A posi-
tive coefficient is expected for that variable, demonstrating an exporting advantage of multinational
affiliates over domestic firms.

These financial variables were chosen because they are the most common measures of financial
constraints in the literature. Following Buch et al. (2014), the tangible asset ratio acts as a proxy
for fixed cost of exporting and thus the higher the tangible asset ratio, the higher the fixed costs
associated with exporting. Consequently, one would expect a negative coefficient associated with
this variable. In Manova (2013), the tangible asset ratio acts as a proxy for pledgeable collateral
and would therefore have a positive sign. The fixed cost interpretation seems more logical in this
specification since I am using balance sheet data for individual firms rather than industry average
values for tangible asset ratio. A firm with a high debt to sales ratio would be considered more
financially constrained than a firm with a low ratio because it would be more difficult to obtain
additional external financing when your debt to sales ratio is already very high. A lender would
be more reluctant to lend to these firms. Thus the expected sign on the debt to sales ratio is
negative. Lastly, cash flows are used in Buch et al. (2014) and Hericourt and Poncet (2009) to
proxy for internal funds. A firm with higher cash flows is able to fund more projects internally and

8

is consequently less financially constrained, and also less dependent on borrowing, so the expected
sign on this term is positive.

In order to address the question of whether or not multinational affiliates are less credit con-
strained than their domestic counterparts, I include interaction terms7 between the ownership status
indicator and each of the financial variables, yielding the following specification:

LogExports f i = α + θM N A f + βX f + γFin f + δ(Fin f × M N A f ) + ϕi + 
 f i

(1.2)

A multinational affiliate having an advantage with respect to credit constraints would have positive
interaction terms for cash flows, tangible asset ratio, and debt to sales ratio. The interpretation
is that effect of these credit constraints is less strong for multinational affiliates. This would give
credence to the belief that multinational affiliates benefit financially from their parent companies
in ways that put them at an advantage over domestic firms. They can alleviate financial market
imperfections in their country through parent firms.

1.5 Results

The first column in Table 1.4 presents baseline results for the effect of firm characteristics on the
log of exports without the financial data or industry controls. Unsurprisingly, size, and profitability
have a positive and significant effect on exports, as does being a Hungarian firm instead of an
Estonian firm (Hungarian firms export at least 43% more than Estonian firms). And importantly,
the foreign ownership variable has a positive and significant coefficient indicating an advantage of
multinational affiliates over domestic firms with regards to exporting. The coefficient on the MNA
variable is 1.25 which means that multinational affiliates export over twice as much as domestic
firms8. The coefficients remain significant after controlling for firm industry (see column 2 of
Table 1.4). Controlling for sector is necessary to account for any differences between industries

7All financial constraint variables are centered in the interaction term analysis
8The 95% confidence interval on the MNA coefficient in Column 1 is between 1.14 and 1.35,

and exp(1.14) − 1 = 2.12

9

that affect exports such as skilled-labor intensity and producing goods (which are easier to export)
versus services.

Column 3 of Table 1.4 includes firm characteristics, industry controls, and the financial vari-
ables. Multinational affiliates still export nearly twice as much as domestic firms after adding these
additional controls9. The results show that a high debt to sales ratio has a significant negative effect
on exporting, indicating a more financially constrained firm exports less, as expected. Also, there
is a negative, and significant, coefficient on a firm’s tangible asset ratio, which gives credence to
interpreting the tangible asset ratio as a proxy for fixed cost rather than pledgeable collateral. The
higher the fixed costs for a firm, the less they export. Cash flows have the expected positive sign,
implying that the more internal financing available to a firm, the more it can export. These findings
are consistent with the existing literature on the effects of credit constraints on international trade.
The more financially constrained a firm is, the less it exports on the intensive margin. Overall,
these results support previous findings that credit constraints negatively affect exporting on the
intensive margin and crucially, that multinational affiliates have an advantage over domestic firms
with regards to exports.

In Table 1.5, the financial variables are standardized in order to provide a more meaningful
interpretation to the coefficients on these terms. The .15 coefficient on Cash flows indicates that
a one standard deviation increase in cash flows for a firm (measured in millions of 2013 USD)
leads to a 16% increase in exports. The coefficients on the Tangible Asset share and the Debt/Sales
ratio are almost identical (although the Debt/Sales ratio has a larger confidence interval). A one
standard deviation increase in the Tangible Asset ratio decreases a firm’s exports by more than
39%10, and a one standard deviation increase in the Debt/Sales ratio decreases exports by more
9Similarly, the 95% confidence interval on the MNA coefficient in Column 3 is between 1.07
and 1.30, and exp(1.07) − 1 = 1.92
10The 95% confidence interval for Tangible Asset is between -.68 and -.50, and 1− exp(−.50) =
.39. Thus, a one standard deviation increase in the tabgible asset share lowers exports by more than
39%

10

than 10%11. These are relatively large effects, consistent with the literature12, which confirm that
credit constraints have a significant dampening effect on the intensive margin of international trade.
Table 1.6 includes interaction terms between the multinational enterprise indicator and the
various measures of financial constraints for the full sample. These terms address the question
of whether or not multinational enterprises are less financially constrained than domestic firms.
One would assume that a multinational firm might have better resources for obtaining financing
through their parent company.
Interestingly, the only highly significant interaction term in the
results shown in Table 1.6 is MNA x Debt/Sales Ratio, which is negative. This means that a
credit constrained multinational affiliate exports much less than a credit constrained domestic firm
when credit constraints are measured as the debt to sales ratio. The interaction term with cash
flows is only marginally significant (see Column 3). Since the decision to engage in international
trade involves high fixed costs and requires firms to obtain additional financing, we might see
more of an advantage of the multinational enterprises in this regard on the extensive margin. In
other words, multinational firms might be less financially constrained than domestic firms on the
extensive margin of international trade where there is a larger dependence on external financing.
This question remains to be explored in a later paper.

1.5.1 Results by Country (Robustness)

Since there are large differences in the characteristics of Hungarian and Estonian firms, Table 1.7
shows the results when the sample is split by firm country as a robustness check. Doing so allows
coefficients to differ in each regression, and is more straightforward than including interaction terms
for all the variables with the Hungarian indicator. As shown in the table, the size, profitability,
tangible asset share, and cash flow coefficients are similar to the results from the full sample
regressions in Column 3 of Table 1.4 (i.e.
their 95% confidence intervals include the estimates
11Similarly the 95% CI for Debt/Sales ratio is between -1.33 and -.11, and 1 − exp(−.11) = .10
12For example, Minetti and Zhu (2011) find that credit rationing reduces intensive margin exports

by more than 38%

11

from Column 3 of Table 1.4). All of those coefficients are also significant with the exception of
cash flows in Estonia. Crucially, the multinational coefficient is very different for Estonian and
Hungarian firms: 1.05 compared to 1.32. This result indicates that multinational affiliates have a
larger advantage over domestic firms in Estonia than in Hungary. However, the 95% confidence
intervals for the MNA coefficient in each country do nest the estimate on the MNA indicator from
Column 3 of Table 1.4. Additionally, the debt to sales ratio is no longer significant in Hungary and
only marginally significant, but negative, in Estonia.

Table 1.8 includes interaction terms between the multinational affiliate indicator and the mea-
sures of financial constraints separating the sample by firm country. By doing so, I test whether
multinational firms are less credit constrained than domestic firms while allowing for firms in dif-
ferent countries to be constrained in different ways. The interaction term between the multinational
indicator and the tangible asset share is insignificant for both countries, and the other two interac-
tion terms are significant (albeit only marginally significant for the debt to sales ratio interaction
term) and negative, for Hungarian firms. For the Estonian firms, MNA x Cash Flows is positive
and significant which supports the claim that multinational affiliates are financially at an advantage
with cash flows, they export more than domestic firms when they have more cash flows on hand.
However, the interaction term with the debt to sales ratio is not significant for Estonian firms. These
results do support the interesting claim that the multinational affiliates are constrained in different
ways between the two countries given that the significant interaction terms has a different sign in
Hungary and Estonia.

1.6 Conclusion

The main findings in this paper are that multinational firms have an advantage over domestic firms,
and export nearly twice as much on the intensive margin. Additionally, credit constraints, measured
by tangible asset ratio, debt to sales ratio, and cash flows reduce exports among firms in Estonia and
Hungary. Moreover, Hungarian firms export more than Estonian firms. This is unsurprising since

12

Hungary is a larger country than Estonia and is closer to large economies such as Germany and
Italy. These results support the conclusion that the influx of multinational investment that occurred
in Central Europe after the fall of the Iron Curtain still had lingering effects in 2013. Multinational
affiliates have an advantage over their domestic counterparts, and these affiliates appear to be
mostly owned by countries that are nearby EU members. These results provide useful information
to studying the benefits of EU membership and foreign investment in transition economies.

However, further work remains to be done on this topic. Primarily, I would like to obtain
data from ORBIS on non-exporting firms in Estonia and Hungary. This is crucial in order to
study extensive margin effects of firm ownership and credit constraints on international trade.
Furthermore, obtaining several years of data from ORBIS to construct a panel would also be
useful to add a time dimension to this analysis. For example, if domestic firms are acquired by
multinational enterprises, we can study how this affects their international trade behavior over time.
Lastly, there has been an increased presence of multinational banks in this area of the world.
It would be useful to parse out the difference between benefits of multinational ownership versus
access to multinational banks. To do so would require finding a way to measure a firm’s exposure to
multinational banks. The large and significant MNA coefficient could partially be reflecting access
to multinational banks by an affiliate firm.

13

APPENDIX

14

Table 1.1 Overall Firm Characteristics

Number Percent

Estonian Firms
Hungary Firms

9,039
6,064

Domestic Firms
Multinational Affiliates

13,321
1,782

59.85
40.15

88.20
11.80

Total

15,103

100.00

Table 1.2 Summary Statistics

Full Sample

MNA

Domestic

Hungary Estonia

0.11
(0.32)
69.31
(533.56)

5.29
(16.14)
15,946
(133,122)

7,954

(108,512)

1,054
(13,113)
1,356
(9,194)
12,878
(162,843)

4,422
(61,515)

169.45
(641.36)

3.70
(14.59)
62,972
(328,168)
41,926
(305,483)

3,874
(32,734)
4,174
(17,427)
44,842
(371,523)
11,938
(99,059)

55.19
(515.99)

5.50
(16.33)
9,655
(73,192)
3,410
(26,412)

676

(7,106)

979

(7,351)
8,602

(107,027)

3,417
(54,496)

0.14
(0.35)
1150.87
(831.72 )

3.71
(11.71)
35,300
(206,366)
17,831
(169,024)

2,258
(20,393)
2,989
(14,185)
29,018
(254,924)

9,905
(95,048)

0.10
(0.30)
14.70
(64.64)
6.35
(18.45)
2,963
(24,955)
1,328
(20,020)

246

(2,598)

260

(1,813)
2,050
(20,564)

745

(15,132)

Variable

MNA

Number of Employees

Profitability

Sales

Export Revenue

Cash Flows

Total Debt

Total Assets

Tangible Fixed Assets

Number of Observations

15,103

1,782

13,321

6,064

9,039

15

Table 1.3 Firm Industry (NACE Main)

Industry

Frequency Percent

A - Agriculture, forestry and fishing
B - Mining and quarrying
C - Manufacturing
D - Electricity, gas, steam and air conditioning supply
E - Water supply; sewerage, waste management and remediation activities
F - Construction
G - Wholesale and retail trade; repair of motor vehicles and motorcycles
H - Transportation and storage
I - Accommodation and food service activities
J - Information and communication
K - Financial and insurance activities
L - Real estate activities
M - Professional, scientific and technical activities
N - Administrative and support service activities
O - Public administration and defence; compulsory social security
P - Education
Q - Human health and social work activities
R - Arts, entertainment and recreation
S - Other service activities
Total

424
54
3,822
47
92
1,258
4,326
1,479
101
749
101
258
1,418
623
2
54
63
88
144
15,103

2.81
0.36
25.31
0.31
0.61
8.33
28.64
9.79
0.67
4.96
0.67
1.71
9.39
4.13
0.01
0.36
0.42
.58
.95
100

16

Table 1.4 Baseline Results

Dependent Variable: log(Exports)

(1)

(2)

(3)

1.25***
(0.060)
0.83***
(0.015)
0.011***
(0.001)
0.41***
(0.052)

1.31***
(0.058)
0.76***
(0.016)
0.012***
(0.001)
0.49***
(0.052)

MNA

Size

Profitability

Hungarian

Tangible Asset Share

Debt/Sales Ratio

Cash Flows

Industry Controls

No

Yes

Observations
R-squared

Robust standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

15,103
0.329

15,103
0.356

1.20***
(0.056)
0.76***
(0.016)
0.011***
(0.001)
0.49***
(0.051)
-1.08***
(0.078)
-0.028**
(0.012)
0.012***
(0.003)

Yes

15,103
0.382

17

Table 1.5 Baseline Results with Standardized Financial Variables

Dependent Variable: log(Exports)

MNA

Size

Profitability

Hungarian

Tangible Asset Share

Debt/Sales Ratio

Cash Flows

Industry Controls

(1)

1.20***
(0.058)
0.76***
(0.016)
0.011***
(0.001)
0.49***
(0.052)
-0.59***
(0.043)
-0.72**
(0.314)
0.15***
(0.045)

Yes

Observations
R-squared
Robust standard errors in parentheses

15,103
0.382

*** p<0.01, ** p<0.05, * p<0.1

18

Table 1.6 Baseline Results with Interaction Terms

Dependent Variable: log(Exports)

(1)

(2)

(3)

MNA

Size

Profitability

Hungarian

Tangible Asset Share

Debt/Sales Ratio

Cash Flows

MNA x Tangible Asset Share

MNA x Debt/Sales Ratio

MNA x Cash Flows

1.19***
(0.058)
0.76***
(0.016)
0.011***
(0.001)
0.49***
(0.052)
-1.07***
(0.082)
-0.028**
(0.012)
0.012***
(0.003)
-0.072
(0.232)

1.16***
(0.058)
0.75***
(0.016)
0.011***
(0.001)
0.49***
(0.052)
-1.08***
(0.078)
-0.016***
(0.006)
0.012***
(0.003)

-0.21***
(0.027)

Industry Controls

Yes

Yes

Observations
R-squared

15,103
0.382
Robust standard errors in parentheses

15,103
0.382

*** p<0.01, ** p<0.05, * p<0.1

1.20***
(0.058)
0.75***
(0.016)
0.011***
(0.001)
0.49***
(0.052)
-1.08***
(0.078)
-0.028**
(0.012)
0.020***
(0.005)

-0.011*
(0.006)

Yes

15,103
0.382

19

Table 1.7 Results by Country

Dependent Variable: log(Exports)

MNA

Size

Profitability

Tangible Asset Share

Debt/Sales Ratio

Cash Flows

Industry Controls

(1)

(2)

1.05***
(0.085)
0.79***
(0.025)
0.01***
(0.003)
-1.39***
(0.147)
-0.121
(0.100)
0.011***
(0.004)

Yes

1.32***
(0.077)
0.73***
(0.022)
0.01***
(0.001)
-0.98***
(0.091)
-0.013*
(0.007)
0.017
(0.024)

Yes

Observations
R-squared

Robust standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

6,064
0.322

9,039
0.269

20

Table 1.8 Results by Country with Interaction Terms

Dependent Variable: log(Exports)

(1)

(2)

1.33***
1.04***
(0.085)
(0.085)
0.73***
0.78***
(0.025)
(0.021)
0.010*** 0.011***
(0.001)
(0.003)
-0.97***
-1.39***
(0.158)
(0.095)
-0.01
-0.05
(0.008)
(0.089)
0.004
0.021***
(0.006)
(0.024)
-0.056
-0.062
(0.297)
(0.361)
-0.19*
-0.12
(0.150)
(0.099)
0.094***
-0.013**
(0.006)
(0.030)

MNA

Size

Profitability

Tangible Asset Share

Debt/Sales Ratio

Cash Flows

MNA x Tangible Asset Share

MNA x Debt/Sales Ratio

MNA x Cash Flows

Industry Controls

Observations
R-squared

Yes

6,064
0.324

Yes

9,039
0.271

Robust standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

21

Figure 1.1 Global Ultimate Owner Countries

22

Table 1.9 ISO Country Code Key

ISO Code

Country

Austria
Belgium
Switzerland

Cyprus
Germany
Denmark
Estonia
Finland
France

Hungary
Ireland
Italy
Japan

Great Britian

Lithuania
Luxembourg

Latvia

Netherlands

Norway
Panama
Poland
Russia
Sweden

United States

AT
BE
CH
CY
DE
DK
EE
FI
FR
GB
HU
IE
IT
JP
LT
LU
LV
NL
NO
PA
PL
RU
SE
US

23

CHAPTER 2

MULTINATIONAL ACTIVITY AND BANKING: THE EFFECTS OF OWNERSHIP

STATUS ON EXPORTERS

2.1 Introduction

In recent years there has been growing interest in the broad question of globalization and specifically,
whether or not the expansion of multinational enterprises is beneficial. Multinational enterprises
are firms that are headquartered in one country, but have operations or branches in one or more
different countries abroad. It is well known that multinational firms tend to be larger and more
efficient than their domestic counterparts, and the size and scope of these firms have increased over
time1. However, it is difficult to determine if this expansion is harmful or beneficial to the countries
receiving foreign investment. Foreign investment can increase domestic wages2 and productivity
3, but conversely, can also lead to a detrimental dependence on foreign technology in the domestic
country and hinder economic development4.

An additional factor to consider in the expansion of multinational enterprises is the spread of
multinational banking. Similarly defined, a multinational bank is a bank that operates in multiple
countries. Usually these banks are part of a larger “banking group" with branches in several
countries. Naturally, these two concepts are related, as multinational firms encounter multinational
banks in their financing decisions.

In this paper I analyze whether or not a naive “Multinational Advantage" coefficient on exports
suffers from omitted variable bias when neglecting to include a firm’s banking behavior. Addi-
tionally, this paper contributes to a growing literature studying the impacts of credit constraints
and imperfect capital markets on international trade. This suggests that multinational firms and

1Markusen and Venables (1998)
2Figlio and Blonigen (2000)
3Lipsey (2004) and Fons-Rosen et al. (2017)
4Vissak and Roolaht (2005)

24

banks (in effect, “globalization") have a positive impact in a country with imperfect capital mar-
kets. Previous literature on the benefits of multinational ownership may overstate the importance
of parent companies if multinational bank access also benefits affiliates. I then use a staggered
difference-in-differences approach to analyze the channels through which multinational banking
affects exporting firms.
I exploit variation in the country of bank ownership through different
mergers and acquisitions. Specifically, I study the effects of the Hungarian government purchasing
two large multinational banks in Hungary on an affected firm’s borrowing and export performance.
My empirical results indicate that multinational banking plays an important role (in addition to
foreign ownership) in a firm’s export performance.

This paper uses a novel firm-level panel data from three Central European economies (CEE):
Croatia, Estonia, and Hungary. All three are European Union members, Estonia and Hungary
having joined in 2004 and Croatia in 2013. These countries all experienced increased foreign
investment during the time period around their EU accession5. Central and Eastern European
countries experienced high levels of foreign investment and banking in the 1990s and early 2000s,
but have largely been neglected in the growing body of literature on financial constraints and
international trade.
In addition, studying firm ownership and banking in Central Europe can
provide information on the long-term benefits of EU accession and foreign investment in transition
economies.

The remainder of this paper proceeds as follows: section two provides a review of literature,
section three explains institutional details, section four describes the data, section five presents
a conceptual framework and descriptive evidence for the importance of firm banking behavior,
section six presents the results for the effect of two major bank purchases on exporting firms in
Hungary, and section seven concludes and discusses avenues for further research. Tables and
figures are presented in section eight.

5Janicki and Wunnava (2006) study this phenomenon in the case of the 2004 enlargement, and
more recently several papers including Vachudova (2014) analyze the Balkan region from which
Croatia was the first EU member.

25

2.2 Review of Literature

2.2.1

Investment in Central Europe

It is important to understand precisely why Central Europe saw an increase in foreign investment
in the transition period of the 1990s and early 2000s. This was the time after the Iron Curtain fell
and several Central European countries applied for EU membership. In 2004, ten countries joined
the EU, eight of which were Central and Eastern European countries such as Hungary and Poland.
The result was a significant enlargement of the European Union (EUROPA).

In a 2004 paper, Bevin and Estrin find that the increase in foreign investment in central and
eastern European countries during the 1990s was mainly due to lower skilled-labor costs, market
size, and proximity. Interestingly, and perhaps unsurprisingly, the authors discover that Western
European EU members invested in Central European economies disproportionally more than other
large economies such as the United States and Japan. Additionally, Bevin and Estrin conclude
that public announcements of EU membership negotiations had a positive impact on foreign direct
investment (FDI) in the candidate country.

These results are corroborated by Carstensen and Toubal (2004) who also find that the level
of privatization in Central and Eastern Europe affects the amount of foreign investment in a host
country. Privatization is a key variable in the attractiveness of a host country because a higher
level of privatization indicates a more successful transition to a market economy following the fall
of the Iron Curtain. Thus, Carstensen and Toubal show that Central European economies, which
were more privatized during this time period, experienced more foreign investment than Eastern
European economies. Lansbury, Pain, and Smidkova (1996) also show that increased privatization
in Central Europe, relative to Eastern Europe, led to the influx of foreign investment by EU members
in Western Europe.

A recent paper by Bilir, Chor, and Manova (2016) shows that host country financial development
plays a significant role in the affiliate location choice of a multinational enterprise. Since Central
European economies were more financially developed than other potential affiliate locations during

26

the transition period, they were a good location choice for multinational affiliates.

2.2.2 Multinational Activity and Export Under Financial Constraints

In their 2009 paper, Antras, Desai, and Fritz Foley study the effects of financial contracting
and investor protections (imperfect capital markets) on a firm’s decision to engage in FDI. The
authors show theoretically that FDI flows and multinational activity arise endogenously for a parent
company under financial frictions and imperfect monitoring in the host country. Multinational
activity occurring as a workaround to imperfect capital markets indicates that there is an advantage
to multinational firms over their domestic counterparts.

Other studies have found that when financial frictions are present, multinational affiliates are at
an advantage compared to domestic firms. For example, Alfaro and Chen (2012) find that foreign
owned firms fared much better during the 2008 global financial crisis than domestic firms, indicating
an advantage of multinational affiliates during periods of economic hardship. Additionally, Desai,
Fritz Foley, and Forbes (2008) find that U.S. multinational affiliates respond better to currency
depreciations in a host country than domestic firms. In particular, multinational affiliates increase
their sales and investment much more than corresponding domestic firms.

Lastly, a recent paper by Manova, Wei and Zhang (2015) addresses the question of multinational
activity under financial constraints using Chinese micro-level customs data. Since they do not have
balance sheet data available in their dataset, the authors cannot compute financial vulnerability at
the firm level. Instead, they study financial vulnerability at the sector level. Their results show that
while financial frictions (financially vulnerable industries) negatively affect exports, this effect is
not as large for multinational affiliates. This suggests that MNCs have access to different means of
financing than domestic firms, and thus can ameliorate the affect of domestic market imperfections.
There is also a recent, and growing, literature on the effects of credit constraints on international
trade, particularly on a firm’s ability to export. Importantly, studies have been conducted using firm
level micro-data in different countries including Peru (Paravisini et al. (2014)), Italy (Minetti and
Zhu (2011)), and China (Feenstra, Li, and Yu (2014)), all of which conclude that credit constraints

27

affect exports both on the extensive and intensive margin. The wide range of countries used in these
studies gives credence to the significant role that credit constraints have on international trade, and
that this phenomenon is not regionally concentrated.

2.2.3 Multinational Banking

In the previous section, I discuss the advantage that multinational affiliates face. This result could
stem from better access to multinational banks (MNBs), and not just from resources afforded to
them by the parent company. There is an extensive literature on the effects that multinational
banking can have on an economy, and perhaps too much credit is given to multinational enterprises,
and not enough to banks.

However, the impact that international banks have can be ambiguous. Historically, international
banks opening, or acquiring, existing banks in less financially developed countries has led to inef-
ficiently run banks. In a 2010 paper, Weller argues that an increase in the number of multinational
banks in Poland after 1991 actually led to a negative effect on business investments due to lower
credit supplies by all Polish banks. This was an unintended consequence immediately following
the transition process. One goal of the transition process in CEEs was to achieve more financial
liberalization and in the long run develop a more sophisticated financial sector.

Chang, Hasan, and Hunter (2010) find that multinational banks are more inefficient than
domestic banks, however, their study is conducted using the US banking sector. One would
not expect the same dynamic to hold in less financially developed countries. Another paper by
Lensink, Meesters, and Naaborg (2008) studies the relationship between foreign ownership and bank
efficiency in Central Europe’s post-transition period (1991-2003). The authors measure inefficiency
in the standard way as expenses divided by revenue. During this time period, they studied over 2000
banks in over 100 countries scattered globally, including countries in Asia and Oceania. While they
find inefficiency in foreign-owned banks is present, this inefficiency diminishes with the quality of
host country governance and with the similarity between home and host country. Central Europe
is a good example of an area in which this inefficiency would be reduced due to its close proximity

28

to Western Europe and EU membership goals during this period.

Additionally, De Haas and Van Lelyveld (2010) find multinational banks weather local financial
crises better than domestic banks. The internal capital markets afforded to them by financially strong
parent banks enable the subsidiaries to face minimal changes during periods of local crisis6. Another
paper by Havrylchyk and Jurzyk (2011) shows that Central and Eastern European banks acquired
by multinational banking groups experience increases in market share and profitability following
the turnover. These banks experience a transmission of knowledge from parent to subsidiary and
are thus at an advantage over banks that remain domestically owned. A similar phenomenon is
found between multinational enterprise parent companies and their affiliates.

My paper contributes to the existing literature by examining a region that is historically known
to have a high level of foreign investment and banking; it is important to know if foreign ownership
gives firms in Central Europe an advantage over domestic firms, particularly when firms are credit
constrained. Furthermore, I investigate how much of that advantage is due to bank access that has
been historically misattributed to parent companies. In analyzing an area of the world that has largely
been neglected by the existing literature on the role of credit constraints on firm performance, I am
attempting to bridge the three branches of literature mentioned above. Studying Central European
firms and their ownership and borrowing behavior provides valuable insight into long-run benefits
of EU membership and the corresponding foreign investment from Western Europe.

2.3 Background and Institutional Details

The history of the banking sectors in Central European economies (and particularly the three
countries studied in this paper) mirrors the increase in multinational activity and FDI during
the post-transition period. Financial development is a crucial factor in economic development,
and during the transition to market-based economies, many Central European countries began to
develop or enrich their banking sector. As shown in Table 2.21, Table 2.22, and Table 2.23 many
6These results do not extend to global crises (2008). De Haas and Van Lelyveld (2014) show

that spillover effects existed between parent bank and subsidiary during the Great Recession.

29

of the modern day banks in Croatia, Estonia, and Hungary were established in the late 1980s and
1990s (when the Iron Curtain fell). The banking sector in each country is unique, with a different
makeup of banks and international presence.

2.3.1 Estonia

Estonia began a rapid transition to a market-based economy in 1992 shortly after gaining inde-
pendence from the Soviet Union in 1991. The country experienced rapid economic development
and in 2004 became a member of the European Union. During the 1990s several Estonian Banks
were established and there was an additional shakeup of the banking sector in Estonia following
the financial crisis with several banks re-branding, being turned over, or closing.

Presently, the largest banks in Estonia are owned by international banking groups. Swedbank
and SEB Bank are both owned by Swedish groups and account for over 60% of the market share in
2016 (see Table 2.22). Luminor Bank is the third largest bank in Estonia (with 14% of the market
share), and it is owned by Norwegian and Finnish groups. Not only is the banking sector in Estonia
dominated by multinational banks, these banks are geographically close to Estonia, representing
Scandinavian or other Baltic countries. This is consistent with the findings of Lensink, Meesters,
and Naaborg (2008) that multinational bank inefficiency diminishes with similarity between home
and host country.

2.3.2 Hungary

Hungary has historically remained an independent state since the collapse of the Austro-Hungarian
Empire. However, following World War II, the country operated as a Communist state with
some free-market aspects until 1989. Following the fall of the Iron Curtain, Hungary began rapid
privatization and experienced a large influx of foreign investment from Western European countries.
Many of the current banks in Hungary were established in the 1990s during this period of transition.
Hungary joined the EU in 2004, but it was already host to several multinational banks. Presently,

30

Hungary is ranked as the 36th largest export economy in the world (2017) and has benefited from
it’s proximity to economically large neighboring countries.

Similarly to Estonia, there were several changes to the banking sector following the financial
crisis in 2008. Notably, the election of a far-right government in 2010 put the domestication of
the banking sector into the media spotlight. This has led to the turnover and government purchase
of several international banks. While many domestic banks were bought-out by international
groups during the transition, the reverse phenomenon has been happening recently. The Hungarian
government has expressed a goal for over 50% of the banking sector to be domestic. This is a
topical field of study. However, despite this recent shakeup, there is still a large multinational
banking presence in Hungary (see Table 2.21). Many of the largest international banks in Hungary
are owned by groups in Germany, Austria, and Italy.

2.3.3 Croatia

Croatia is the most recent EU member, having joined in 2013 after a lengthy candidacy which
began in 2003 and became official in 2004. Following a similar timeline as Estonia and Hungary,
Croatia declared it’s independence from Communist Yugoslavia in 1992, and had already started the
privatization process in 1991. There was already an established banking system in place (Reininger
and Walko (2005)) with several domestic banks with establishment dates in the late 19th and early
20th century (see Table 2.23). This is a phenomenon unique to Croatia relative to other Central
and Eastern European countries. There are also several present-day banks that were established
during the transition to a market economy in the late 1990s. Similarly to Estonia and Hungary, a
few banks went through restructuring and mergers after the global financial crisis, and many of the
internationally-owned banks are from countries in close geographic proximity.

It is reasonable to compare these three countries because they went through similar transition
patterns. Each declared independence from a communist country, or transitioned from commu-
nism in the early 1990s. This was then followed by the rapid privatization of industry and the
development/enrichment of a banking sector. Lastly, each country was granted EU membership or

31

candidacy in 2004, and experienced some shakeups in their banking sector following the financial
crisis. Each country also has a strong multinational banking presence, with variation in the own-
ership countries. This variation comes from a strong geographic link between a country and the
international banking groups operating there. Therefore, it is valid to analyze these three countries
in one study.

2.4 Data

The data in this analysis is from ORBIS, provided by Bureau van Dijk, and is comprised of
exporting firms from the three EU members: Croatia, Hungary, and Estonia. 33% of the firms in
my sample are Estonian, 44% are Croatian, and 23% are Hungarian. These countries are part of the
AMADEUS (European) subset of ORBIS for the years 2010-2017. The firm-level data is collected
annually from businesses through websites and various business registries and reports. ORBIS
is a well-known datasource used in several papers on FDI and multinational enterprise behavior7.
Croatia joined the EU in 2013, towards the beginning of the panel. Therefore it is interesting to
compare a newer EU member to more established members from 2004.

There are several advantages to this dataset. First, ORBIS provides detailed balance sheet data
for firms of all sizes (including unlisted firms). Examples of these variables are: cash flows, debt,
and fixed assets. These measures have all been used in previous papers to construct proxies for
financial constraints8. Additionally, ORBIS provides detailed ownership information on each firm
including the country of the global ultimate owner, and an independence indicator showing how
much autonomy a multinational affiliate has. For my analysis, I follow the work of Cravino and
Levchenko (2017) and consider firms with a global ultimate owner (GUO) country that differs
from the home country, and with an independence indicator of “D" as a multinational subsidiary.
7For example, Cravino and Levchenko (2017) study business cycle transmission through multi-
national firms and Buch et al. (2014) use ORBIS linked with German customs data to analyze the
effects of credit constraints on firm’s FDI behavior.

8See Buch et al. (2014)

32

This means that the GUO (and thus international) shareholders have direct ownership over 50%.
I believe these firms have more direct access to the parent companies and their resources, and are
therefore at an advantage over other internationally owned firms with less direct control.

Other useful variables from the dataset included in my analysis are: NACE industry classifica-
tion, total employees, profitability, and city as these provide other controls for the firm that might
also affect firm performance. Summary statistics for the key variables for the exporting firms are
presented below in Table 2.1 and Table 2.2. For clarity, I present the means for key variables for
the full sample, then divide by firm ownership status, then finally present the means across country.
Overall, apporiximately 11% of firms are considered multinational affiliates by my definition, with
some differences across countries. Hungary has a higher ratio of MNAs and Croatia’s ratio is
slightly below the mean. Additionally, in line with previous literature and findings, MNAs are
much larger (both in terms of employees and revenue). They also have higher average amounts for
debt and assets, indication a higher level of financial activity. Interestingly, multinational affiliates
are more likely to be located in the capital city of a country. However, this is unsurprising given
the three countries in my sample. For Hungary, Croatia, and Estonia, the capital is the largest and
most “global" city.

When looking at summary statistics across country, Croatian and Estonian firms are similar
across many firm characteristics. These include percentage of MNAs, profitability, capital city
share, and export revenue. However, they differ along many of the balance sheet variables with
Croatian firms having larger means for all types of assets, debt, and cash flows. The Hungarian
firms dominate along all variables except profitability. They have more financial activity and are
larger in terms of revenue and employees. This is logical since Hungary is a much larger country
than either Croatia or Estonia. Table 2.6 shows the most frequent foreign ownership country across
the three countries. Most of the countries listed are Western European EU members (several of
which border the countries in the sample), corroborating the findings of Bevin and Estrin (2004)9.
9Central Europe was an attractive area for FDI among Western EU members in the period

following the transition to market economies and prior to EU membership

33

Table 2.2 presents the banking characteristic summary statistics for all firms, firms separated
by ownership status, and across countries. It is evident that for the full sample of firms that borrow
from banks, most of them use multinational banks as either the sole means of financing or combined
with domestic banks. Moreover, MNAs are more likely to only borrow from foreign banks than
domestic firms. Domestic firms are more likely than MNAs to combine foreign banking with
domestic banking. Another fact to note is that 16% of multinational affiliates borrow from a bank
that is owned by a group in the same country as their parent enterprise. This is a non-trivial amount,
considering that not every GUO country has a corresponding multinational banking presence in
Croatia, Estonia, or Hungary. One last takeaway from Table 2.2 is that multinational affiliates
borrow from a slightly larger number of banks on average that domestic firms, and that they have
more non-bank advisors, indicating more diversification in their financing behavior.

When comparing banking behavior across countries in Table 2.2, there are vast differences
It is important to look
between Croatian, Hungarian, and Estonian firms’ borrowing behavior.
across countries as there are idiosyncrasies in the banking sector in each country and this could
lead to different borrowing patterns. In addition, the options available to firms in different countries
are unique. For example, in Estonia only 437 firms borrow from an Estonian bank, because the
domestic banks are much smaller and tend to cater to households and not firms. Thus, when firms
borrow, they more likely choose multinational banks. Conversely, Hungary has several large and
competitive domestic banks (OTP), along with many big international banks (ING) which gives
exporters more options. This is seen in the Hungarian column of the table where firms have more
diversity among the banking options (Foreign Borrower (excl.), Foreign Borrower (comb.), and
Domestic Borrower (excl.)). Estonia and Croatia have similar shares among the three variables
since the domestic banks tend to be small, rendering multinational banking more attractive.

Measuring a firm’s multinational bank exposure is not immediately clear. Firms could use
banks in their parent country, domestic country, or any other country. One of the advantages
of ORBIS is that for most firms, banks are included along with other firm information in the
dataset. The banks from which a firm borrows are listed in the “Advisor" column. In order to

34

code which banks a firm uses, I compiled a list of the largest banks in a country along with the
country of ownership, and create an indicator for each bank the firm lists in their “Advisor" column.
This required scraping the column for the name of the bank (sometimes in the original language)
since the “Advisor" variable is simply a list of names and sometimes up to 15 banks were listed.
Table 2.3, Table 2.4, and Table 2.5 list all the banks and owner countries for the exporting panel
separated across country and Table 2.7 lists the most frequent banks for each country. I coded
46 unique banks across the three countries in the dataset, with a good mix of domestic banks and
multinational banks. These were the most prevalent across all the firms. I then aggregated the
individual banks to the bank-country level. This allows for a direct comparison of bank borrowing
behavior to multinational affiliate ownership country (e.g. one can see if an Italian owned firm in
Hungary borrows from an Italian bank etc.).

2.5 Descriptive Motivation and Conceptual Framework

2.5.1 Conceptual Framework

An exporting firm has several means of financing available: domestic banks, multinational banks,
and non-bank financing (e.g. private individuals). Each of these options comes with different
costs (interest rates, ease of access). It is reasonable to assume that these costs are heterogeneous
across ownership status. Multinational affiliates could use banks in their parent country, domestic
country, or any other country. For example, it would be more straightforward for a Estonian firm
owned by a Danish company to borrow from a Danish bank than it would be for an Estonian firm
with no connection to Denmark. This would be an example of a firm using the internal capital
markets between MNA and parent companies found by Desai, Fritz Foley, and Hines (2008).
Furthermore, it could also be easier for the Danish-owned firm to access banks in Germany or
Sweden, both neighboring countries to Denmark with strong banking presences there. I propose
that multinational affiliates have better access to international banks through the networks afforded
to them by their parent company. Therefore multinational affiliates face a lower cost of borrowing

35

internationally than domestic firms.

For clarification I present the following example: Figure 2.1 shows an example of two Hungarian
multinational affiliates: Louis Vuitton Hungaria and Lidl Hungary. Lidl is a German affiliate and
Louis Vuitton is a French affiliate. In the advisor column (ADV) Louis Vuitton borrows exclusively
from the BNP Paribas Hungarian branch, which is a French-owned bank. This borrowing behavior
is relatively straight forward: a French company borrows from a French bank. However, Lidl
borrows from five different banks: BNP Paribas, CIB, ING, and two OTP branches, which are
French-owned, Italian-owned, Dutch-owned, and Hungarian-owned respectively. Lidl’s pattern
behavior is not as simple. The firm borrows from domestic banks and larger multinational banks
that are not from the same country as their parent company, but are banks from nearby Western
European countries. Both firms borrow internationally, and through parsing out the different banks
in the advisor column, I can differentiate between firms that borrow exclusively from foreign banks
(Louis Vuitton) and firms who use both domestic and multinational banks (Lidl). In both cases it
would be naive to just examine a multinational ownership coefficient and neglect their borrowing
from multinational banks.

2.5.2 Descriptive Evidence and Omitted Variable Bias

It is reasonable to expect that firm ownership status and credit constraints would affect firm
performance, both domestically and through exporting. Additionally, multinational firms should
have the advantage in both cases as they are usually larger, more productive firms who should
perform better than domestic firms. Therefore, I propose two specifications: one to study exporting
behavior of a firm, and a second to study domestic behavior.

For analyzing the effects of firm ownership status, multinational banking, and credit constraints
on the intensive margin of international trade for firm f in industry i in country j I estimate the
following random effects model:

Log(ExportRevenue f i jt) = α + θMNA f + χMNB f + βX f + γFin f + ϕi + ϕ j + ϕt + 
 f i jt (2.1)

36

Where MNA f is an indicator for a firm being a multinational affiliate; MNB f is a measure
of multinational banking access; X f
is a vector of firm characteristics: capital city indicator,
profitability, and size measured as log(Employment); Fin f is a vector of financial variables that act
as proxies for financial constraints: debt to sales ratio, tangible asset ratio, and cash flows; Lastly,
ϕ are industry (NACE 2 digit), country, and year effects. Controlling for sector is necessary to
account for any differences between industries that affect exports such as skilled-labor intensity and
producing goods (which are easier to export) versus services.

The variables in MNB f include an indicator if the firm borrows from a multinational bank
from the same country as its headquarter country (Same Country) and two indicators for whether
or not a firm borrows from a foreign bank exclusively, or a combination of foreign and domestic
banks. In the Lidl and Louis Vuitton example from above, Lidl would have a value of 1 for the
combined foreign borrowing indicator and a 0 for the exclusive foreign borrower indicator, and vice
versa for Louis Vuitton. Louis Vuitton would also have a value of 1 for the Same Country variable.
One would expect the sign on the banking variables to be positive since firms that borrow from
international banks should be able to export more as they are better able to finance their activities
due to access to larger more financially sophisticated banks. Furthermore, the sign on the Same
Country variable should be positive. This follows because borrowing from a bank owned by a
group headquartered in the same country as your parent company should give you an advantage
such as better knowledge of the banking sector in that country. This would lead to more efficient
financing activities and therefore, a better ability to export.

Once the variables for multinational banking are included, along with controlling for each bank
(columns 3 and 4 in Table 2.9 - Table 2.14), the coefficient on MNA f should fall. This occurs since
omitting banking information causes the coefficient on MNA f to be upwardly biased. This result
is expected because both:

and

Corr[MNA, MNB] > 0

Corr[Log(ExportRevenue), MNB] > 0

37

(2.2)

(2.3)

These findings demonstrate that neglecting to account for firm borrowing and multinational banks
overstates the effect of multinational ownership on a firms levels of exports. Additionally, non
bank advisors (e.g. private borrowing) also have a positive effect on exports that could have been
mistakenly captured in the coefficient on multinational ownership.

These financial variables were chosen because they are the most common measures of financial
constraints in the literature. Following Buch et al. (2014), the tangible asset ratio acts as a proxy
for fixed cost of exporting and thus the higher the tangible asset ratio, the higher the fixed costs
associated with exporting. Consequently, one would expect a negative coefficient associated with
this variable. In Manova (2013), the tangible asset ratio acts as a proxy for pledgeable collateral
and would therefore have a positive sign. The fixed cost interpretation seems more logical in this
specification since I am using balance sheet data for individual firms rather than industry average
values for tangible asset ratio. A firm with a high debt to sales ratio would be considered more
financially constrained than a firm with a low ratio because it would be more difficult to obtain
additional external financing when your debt to sales ratio is already very high conditional on
already exporting. A lender would be more reluctant to lend to these firms. Thus the expected
sign on the debt to sales ratio is negative. Lastly, cash flows are used in Buch et al. (2014) and
Hericourt and Poncet (2009) to proxy for internal funds. A firm with higher cash flows is able
to fund more projects internally and is consequently less financially constrained, and also less
dependent on borrowing, so the expected sign on this term is positive. The expected signs on the
size, profitability, and capital city variables are also positive. More profitable firms should have
higher export revenue (mechanically), and firms with more employees should also export more.

Similarly, my specification to study the effects of MNA status and multinational banking on

domestic performance of exporters under financial constraints is:

Log(DomesticRevenue f i jt) = α +θMNA f + χMNB f + βX f +γFin f + ϕi + ϕ j + ϕt +
 f i jt (2.4)

Where

DomesticRevenue f i jt = TotalRevenue f i jt − ExportRevenue f i jt

(2.5)

38

As a robustness check, I also measure firm performance as Log(TotalRevenue f i jt) and use the
following random effects specification (for these results please see appendix Table 2.25, Table 2.26,
and Table 2.27):

Log(TotalRevenue f i jt) = α + θMNA f + χMNB f + βX f + γFin f + ϕi + ϕ j + ϕt + 
 f i jt (2.6)

2.5.3 Results

2.5.3.1 Exports

The results from estimating Equation 1 for all exporters show that including banking information
(comparing Column 1 to Column 4 in Table 2.9) causes the coefficient on MNA to fall by 12.5%.
This is a non-trivial amount and shows that the “multinational advantage" in export revenue falls
from 122% to 102%10. To avoid concern that this drop is attributed to generic “bank borrowing",
I test whether the MNA coefficient stays the same when including a dummy variable for a firm
borrowing form any bank. The results of this test are shown in column 2 of Table 2.9. While
this variable is large and significant, it has no effect on the ownership status variable. However,
upon controlling for the type of bank (see Column 3) and including dummies for each of the 46
unique banks (see Column 4), there is a clear reduction in the MNA effect. This indicates that this
phenomenon is not attributed to “banks," but rather, multinational banks.

Additionally, the signs on the firm characteristics are as expected. More profitable firms export
more, as do larger firms with more employees. The capital city indicator is insignificant. Since
the capital is the largest major city in all three countries, there is no clear distinction between
the type of firm that locates in the capital or elsewhere. Firms with a higher debt/sales ratio and
a higher tangible asset share also export less, while higher cash flows are associated with more
exports. This is in line with the literature about financial constraints negatively affecting exports,
even on the intensive margin11. These signs are consistent throughout specifications and for clarity
10The coefficient on the MNA indicator falls from .798 to .705, and exp(.798) − 1 = 1.22 &
exp(.705) − 1 = 1.02
11As shown in Minetti and Zhu (2011) in Italy, and other countries in related literature.

39

of presentation the variables have been removed from the subsequent tables.

An advantage of splitting the MNB coefficient into Foreign Borrower (exclusive) and Foreign
Borrower (combined) is that one can analyze whether it is better to borrow solely from international
banks or diversify between types of bank. I find that for the full sample, the coefficients on Foreign
Borrower (e) and Foreign Borrower (c) are both positive but the magnitude is larger for Foreign
Borrower (e), indicating that it is better to borrow only from foreign banks. This effect holds for
all groups except Estonian exporters. The coefficient on Same Country is large and significant
which indicates an advantage associated with borrowing from a bank owned by a group in the same
country as an affiliate’s parent company.

When running the specification across countries, it is evident that the reduction of the MNA
coefficient is most pronounced for Hungarian firms (see Table 2.10). The coefficient falls by 36%
(from .537 to .342), which is the largest drop across countries in the full sample, and corresponds
in a change in “multinational advantage" from 71% to 41%. This result is relatively unsurprising
as Hungary is a larger country with more investment from large Western European countries
(see Table 2.6). The coefficient also falls for Estonian firms (13.7%), but not for Croatian firms.
Since Croatia is the most recent EU member in the sample, it is plausible that it has not experienced
the same influx of foreign investment and banking from Western Europe as Estonia and Hungary.
These results also indicate that there is a long term benefit to EU membership and globalization.
The dataset spans from 2010 to 2017 and thus, the timeline is too short to see a large benefit from
multinational banking in Croatia.

Additionally, I run the specification across firm size and the results are shown in Table 2.13
and Table 2.14. I divide the sample into small/medium enterprises (SME) and large firms. I use
the same designation for a SME as the European Union. An SME is a firm with less than 250
employees, and large firms are those with more than 250 employees. The vast majority of firms in
the dataset, approximately 84%, are SMEs. This ratio is about the same across all three countries
and across ownership status. The multinational affiliate coefficient falls in both cases, but the
magnitude of the drop as a percentage of original coefficient is greater for large firms. However, it

40

is important to note that the coefficient on MNA is not significant in either column for large firms,
most likely due to a lack of power. There is still evidence that the coefficient on multinational
ownership was overstated for both types of firms but more so for large enterprises (i.e. accounting
for banking behavior has a stronger effect on large firms).

2.5.3.2 Domestic Revenue

The results from estimating Equation 4 across the full sample, separating by country, and separating
by size are presented in Table 2.15 through Table 2.17. The firm characteristics and financial
variables have similar magnitudes and signs as those when using export revenue as the dependent
variable. However, it is interesting to note that there appears to be no significant “multinational
advantage" across any specification for the exporting firms. This follows from the fact that the mean
amount of exports as a percentage of total revenue is 53% for multinational affiliates and 44% for
domestic firms. The percentage is relatively constant across countries with Estonian firms having
the highest amount with 56%. This suggests that once a firm selects into exporting, the majority
of their operations will be concentrated there. Although the coefficient on MNA falls once one
controls for banking status, the coefficients are small and insignificant. The results for total revenue
mirror those of export revenue at a smaller scale. This follows from the high Export/Total Revenue
ratio among all firms in the sample. The results are shown in the appendix.

2.5.4 Robustness Checks

I conduct several sensitivity analyses to verify whether the dynamics between multinational affiliate
status and multinational banking behavior holds. First, I estimate Equation 6. Since the majority of
firm revenue comes from export activity, one would expect the same relationship between the MNA
coefficient and the multinational banking coefficients. I present the results with Log(TotalRevenue)
as the dependent variable in Table 2.25 through Table 2.27. While the point estimates on the
MNA indicator are lower, the coefficient falls by about 10% when controlling for specific banking

41

behavior, a result which is close to the 12% found for exporters in Section 4.2. The results are also
heterogeneous across country. The small result for Croatian firms is similar to those for exports,
but there is a very large decrease in the coefficient for Estonian firms and a smaller decrease for
Hungarian firms. This phenomenon diverges from results found in Table 2.10 through Table 2.12.
Furthermore, I run the specifications in Table 2.9 controlling for bank country instead of
individual bank indicators. This might better capture idiosyncrasies among each parent country’s
banking sector. There are a few countries with several banks in the sample, for example: Italy and
Austria. I find that this does not change the results significantly from Column 4 of Table 2.9. The
coefficient on MNA falls by 11% (instead of 12%) but the coefficients are not statistically different
from each other. The results are presented in Table 2.28.

Finally, I allow for my definition of multinational affiliate to include any firm whose Global
Ultimate Owner country does not match the firm country. Doing so absorbs an extra 1,700 firms
into the MNA category. These firms are more independent from their parent companies, and
their international shareholders have direct company ownership of less than 50%. When I run the
specification from Equation 1 (see Table 2.29), I find that although the point estimates are higher,
the drop on the MNA coefficient is consistent with the results from the original MNA definition
(11% compared to 12%). The point estimates are higher because I have reclassified some of the
“best" domestic firms (in terms of export revenue) as multinational affiliates. It is interesting that
some of the better domestic firms have some minority of foreign-owned shareholders.

2.6 Causal Effects of Bank Turnover on Firm Performance

2.6.1 Difference-in-Differences Framework

In 2014 and 2015, the Hungarian government purchased two major banks from international owners.
Budapest Bank was purchased from the German group Bayern LB and MKB was purchased from
the American group GE Capital. The purchases of MKB and Budapest Bank by the Hungarian
government were in 2014 and 2015 respectively. These purchases were announced and completed

42

within a year of each other, and affected 26% of Hungarian exporters in the dataset (Table 2.24). I
assert that these turnovers are both instances of a multinational bank switching to a domestic bank.
These banks were purchased by the government with the intention to privatize them12 and were
thus run as domestic private banks rather than public banks.

To study the impact this turnover in banking ownership had on exporting firms in Hungary, I

use the following difference-in-differences model:

(cid:88)

Yf it = α f + δt +

γkD f k + ϕi + 
 f it

(2.7)

Which I estimate as:

k

y f i = β1Borrower f + β2PostPurchase f + β3PostPurchase f ∗ Borrower f + βX f + γFin f ϕi + 
 f i

(2.8)
Where y f it is the measure of firm outcome: Log(ExportRevenue f i jt) or Loans f i jt; Borrower f
is an indicator that takes a value 1 if firm f borrowers from either MKB or Budapest Bank (or both);
PostPurchase f is a dummy variable that equals 1 if the observation for firm f is in the period
following the purchase of the banks. Post treatment year is 2015 for Budapest Bank borrowers and
all control firms, and 2014 for MKB borrowers. β3 represents the causal effect of the turnover from
international to domestic banking on firm exports and loans and thus, is my coefficient of interest.
This follows from differencing out any unobserved changes shared across all firms. I present the
results for all firms and then subsequently separate the firms into two groups: domestic firms and
multinational affiliates, and estimate Equation 8 for each group.

In order for this estimation method to be valid, the assumption of parallel trends must be
satisfied. Specifically, I test whether the firms borrowing from MKB and Budapest bank had
common trends in loan amount and exports prior to the government purchase as the unaffected
firms (non-MKB and Budapest Bank borrowers). I can use the following related model to analyze
12According to contemporaneous articles in several Hungarian news sources such as the Budapest

Business Journal and Daily News Hungary

43

pre and post trends and test this assumption:

Yf it = α f + δt +

βtT f t +

−1(cid:88)

t=m

Which I estimate as:

y f it = α f + δt +

−1(cid:88)

τ=m

βτ1(τ = t) ∗ Borrower f +

g(cid:88)

t=1

g(cid:88)

τ=1

γtK f t + ϕi + 
 f it

(2.9)

βτ1(τ = t) ∗ Borrower f + ϕi + 
 f it

(2.10)
Figure 2.2 through Figure 2.7 plot the estimated βs from estimating Equation 10 across all firms, then
separating multinational affiliates from domestic firms. As seen in and Figure 2.6 and Figure 2.7,
the coefficients before the purchase are statistically indistinguishable from zero, lending credence
to the parallel trend assumption for multinational affiliates. In Figure 2.2 through Figure 2.5 the
coefficients are also statistically indistinguishable from zero in the pre-period with the exception of
the earliest coefficient. The time period closest to the purchase satisfy this assumption for all firms
and, separately, for domestic firms. One can therefore interpret the findings as causal, particularly
when separating the sample into multinational affiliates and domestic firms, since affiliates and
their domestic counterparts are quite different (see Table 2.1 and Table 2.2).

2.6.2 Difference-in-Differences Results

The acquisitions of MKB and Budapest Bank appear to have no significant impact on loans and
exports for all firms in the sample together (Table 2.18). However, upon splitting the analysis by
firm ownership status, a more interesting dynamic becomes evident. For domestic firms, there is
a decrease in total loans taken out by the firm of about $270,000 (with a p-value of .06 in both
Columns 1 and 2 in Table 2.19). There is also a very small, increase in Log(Exports) among
domestic firms.
I find no evidence of an effect on loans or exports for multinational affiliates.
Table 2.19 and Table 2.20 indicate that the transition of an international bank to a domestic bank
negatively affects loans for domestic firms, but not multinational affiliates. This result provides
evidence that multinational affiliates are better able to smooth borrowing behavior in a tumultuous

44

banking environment than their domestic counterparts. It is well-known that multinational firms
are more efficient and larger (for example, the mean loan amount for affiliates is $2.21 million
whereas for domestic firms it is $1.05 million) and Columns 1 and 2 of Table 2.19 and Table 2.20
show that this extends to the amount of loans taken out by a firm in the presence of international
bank turnover. These results show that the firms affected most by nationalistic banking policies are
smaller locally owned firms.

2.7 Conclusion

This paper aims to address the broad question: “Does multinational banking matter in the discussion
of globalization of firms?" and subsequent to finding positive evidence asks: “How does it play
a role?" I concentrate my analysis to three Central European Economies: Hungary, Estonia,
and Croatia. These countries are all EU members and experienced an increase in investment
following the fall of the Iron Curtain. To incorporate firm borrowing behavior, I include the
following variables: a dummy variable if the firm borrows from a multinational bank from the
same country as its headquarter country and dummy variables for whether or not the firm borrows
from a foreign bank, and a domestic bank.
I demonstrate that the coefficient corresponding
to the “multinational advantage" falls upon controlling for borrowing behavior. Furthermore, I
utilize a staggered difference-in-differences methodology to analyze the impact of the Hungarian
government purchasing two multinational banks on an exporting firm’s loans and exports.

I find that once banking behavior is included in the regression of Log(Exports) on MNA status,
the coefficient on MNA to fall by 12.5%. This corresponds to the “multinational advantage" in
export revenue falling from 122% to 102% which is a non-trivial amount. The result is strongest
for Hungarian exporters where the “multinational advantage" falls from 71% to 41%. Additionally,
I find evidence that multinational affiliates are able to smooth their borrowing behavior in the
presence of international turnover of banks. When the Hungarian government purchased MKB
and Budapest Bank (effectively transforming them from multinational to domestic banks), loans

45

among domestic exporters fell by over $300,000, but loans among multinational affiliates remained
unchanged. There appeared to be no adverse effect on exports from this purchase. However, more
post-turnover analysis may be necessary to see whether long-term effects of this event exist. This
represents one avenue for further research.

Another expansion of this project is a study how the purchases of MKB and Budapest Bank
affected domestic firms. This paper is limited to the analysis of exporting firms’ behavior and
performance, but outcomes such as selection into exporting and domestic firm performance may
also be impacted by this event.

This avenue of research has several broader implications. First, it provides evidence that global
banking is more relevant for countries that have been EU members for a longer period of time.
Croatia is the newest member and the multinational affiliate coefficient remains almost unchanged
upon controlling for borrowing behavior. Hungarian and Estonian firms display a noticeable drop
in the “multinational advantage" having been members of the EU since 2004 rather than joining in
2013. This indicates a more long-term benefit to foreign investment and banking. Second, while
Havrylchyk and Jurzyk (2011) find that international acquisition of domestic banks in Central
Europe improved bank profitability and market share, the recent trend in the other direction could
have detrimental effects on exporting firms’ loans. Although the prevailing sentiment of the
Hungarian government is to turn banks domestic, this could be actually be problematic for firms
that borrow from these banks.

46

APPENDIX

47

Variable

MNA

Number of Employees

Profitability

Total Revenue

Export Revenue

Domestic Revenue

Capital City

Cash Flows

Debt

Total Assets

Tangible Fixed Assets

Table 2.1 Summary Statistics- Firm Characteristics

Full Sample

MNA

Domestic

Hungary Estonia Croatia

0.11
(0.32)
57.20
(453.73)

5.95
(16.80)
11,092
(102,921)

4,794
(81,064)
6,298
(51,615)

0.43
(0.50)
0.90
(40.00)
1,137
(7,465)
9,068

(164,776)

3,062
(43,038)

144.14
(667.83)

3.52
(15.26)
41,504
(259,546)
23,866
(229,290)
17,637
(91,976)

0.55
(0.50)
3.69

(117.23)
3,333
(15,507)
33,486
(446,335)

7,983
(50,863)

45.99
(416.89)

6.27
(16.97)
7,170
(56,000)
2,334
(24,193)
4,836
(43,565)

0.42
(0.49)
0.53
(5.71)
853

(5,585)
5,919
(69,803)
2,427
(41,880)

0.16
(0.37)
167.90
(892.23 )

3.88
(12.34)
36,249
(207,897)
16,947
(167,224)
19,301
(99,716)

0.39
(0.49)
2.91
(83.45)
2,818
(12,682)
29,087
(339,697)

9,305
(83,904)

0.11
(0.31)
15.57
(62.86)
6.56
(18.22)
2,865
(21,517)
1,273
(17,332)
1,592
(11,126)

0.49
(0.50)
0.21
(1.78)
248

(1,323)
1,886
(17,659)

665

(12,622)

0.10
(0.29)
31.12
(207.32)

6.57
(17.60)
4,218
(28,833)
1,124
(8,333)
3,094
(26,112)

0.41
(0.49)
0.36
(3.28)
937

(6,318)
4,085
(34,868)
1,629
(20,128)

Number of Observations

190,581

21,773

168,808

43,596

63,797

83,188

48

Table 2.2 Summary Statistics- Banking Characteristics

Variable

Full Sample MNA Domestic Hungary Estonia Croatia

Same Country
Foreign Borrower (excl.)
Foreign Borrower (comb.)
Domestic Borrower (excl.)
# of Non-Bank Advisors
Number of Banks
Austrian Bank
Belgian Bank
Chinese Bank
Croatian Bank
Danish Bank
Estonian Bank
French Bank
Finnish Bank
German Bank
Hungarian Bank
Italian Bank
Korean Bank
Latvian Bank
Dutch Bank
Russian Bank
Swedish Bank
US Bank

0.02
0.63
0.10
0.05
1.09
1.21
0.29
0.06
0.00
0.05
0.03
0.00
0.01
0.00
0.01
0.15
0.34
0.00
0.00
0.01
0.02
0.11
0.02

0.16
0.76
0.06
0.02
1.72
1.26
0.26
0.05
0.00
0.02
0.05
0.00
0.03
0.00
0.05
0.11
0.34
0.00
0.00
0.04
0.02
0.13
0.06

0.00
0.61
0.11
0.05
1.01
1.20
0.30
0.06
0.00
0.05
0.02
0.00
0.00
0.00
0.01
0.15
0.34
0.00
0.00
0.01
0.02
0.10
0.02

0.04
0.55
0.28
0.16
3.06
1.73
0.27
0.24
0.00

0.03

0.06
0.44
0.35
0.01

0.04
0.04

0.09

0.01
0.34
0.01
0.00
0.80
0.48

0.08
0.01

0.00

0.00

0.00

0.32

0.02
0.88
0.09
0.02
0.29
1.50
0.53

0.11

0.11
0.59

0.03

Number of Observations

190,581

21,773

168,808

43,596

63,797

83,188

49

Table 2.3 List of Banks–Hungary

Bank

HQ Country

China
France
Hungary
Italy
USA
Germany
Germany

Bank of China
BNP Paribas
Budapest Bank
CIB Bank
Citi Bank
Commerzbank
Deutsche Bank
Erste & Steiermärkische Austria
Gránit Bank
Hungary
Netherlands
ING Bank
Belgium
K&H Bank
South Korea
KDB
Kinizsi Bank
Hungary
Hungary
Központi Bank
Hungary
MagNet Bank
Merkantil Bank
Hungary
Hungary
MKB
Austria
Oberbank
Hungary
OTP
Polgári Bank
Hungary
Austria
Raiffeisenbank
Russia
Sberbank
UniCredit Bank
Italy

50

Table 2.4 List of Banks–Estonia

Bank

HQ Country

AS Citadele
Latvia
Big Bank
Estonia
Coop Pank
Estonia
Danske Bank
Denmark
LHV Pank
Estonia
Luminor Bank
Sweden
Marfin Pank
Estonia
OP Corporate
Finland
SEB Bank
Sweden
Svenska Handelsbanken Sweden
Sweden
SwedBank
Tallinna Aripank
Estonia
Italy
UniCredit Bank

Table 2.5 List of Banks–Croatia

HQ Country
Bank
Austria
Addiko
Croatia
Banka Kovanica
Croatia Banka
Croatia
Erste & Steiermärkische Austria
Hrvatska Po˘stanska
Croatia
Istarska Kreditna Banka Croatia
Karlova˘ca Banka
Croatia
Hungary
OTP
Podvravska Banka
Croatia
Croatia
Primorska Banka
Italy
Privrenda Bank
Raiffeisenbank
Austria
Russia
Sberbank
Hungary
Splitska Bank
UniCredit Bank
Italy

51

Table 2.6 Most frequent foreign owners – Exporters

Croatia

Estonia

Hungary

1. Slovenia
2. Germany
3. Austria
4. Italy
5. USA
6. B&H
7. UK
8. Switzerland
9. Hungary
10. Serbia

1. Finland
2. Sweden
3. Latvia
4. USA
5. UK
6. Germany
7. Lithuania
8. Norway
9. Russia
10. Denmark

1. Germany
2. USA
3. France
4. Switzerland
5. Italy
6. Austria
7. UK
8. Japan
9. Sweden
10. Netherlands

Notes:Countries in red represent a bordering country (by land)

Countries in blue represent a maritime border

52

Table 2.7 Most Frequent Banks for Multinational Affiliates

Croatia

Estonia

Hungary

1. Raiffeisenbank (Austria)
2. UniCredit (Italy)
3. Privredna Banka (Italy)
4. Erste & Steiermärkische (Austria)
5. Splitska Banka (Hungary)
6. Addiko Bank (Austria)
7. Sberbank (Russia)
8. Hrvatska Po˘stanska (Croatia)
9. Istarska Kreditna Banka (Croatia)
10. Podravska Banka (Croatia)

1. Swedbank (Sweden)
2. SEB Pank (Sweden)
3. Luminor Bank (USA)
4. Danske Bank (Denmark)
5. LHV Pank (Estonia)
6. Svenska Handelsbanken (Sweden)
7. Coop Pank (Estonia)
8. OP Corporate Bank (Finland)
9. AS Citadele (Latvia)
10. Tallinna Aripank (Estonia)

1. UniCredit (Italy)
2. Citibank (USA)
3. K&H (Belgium)
4. Raiffeisenbank (Austria)
5. ING Bank (Netherlands)
6. BNP Paribas (France)
7. OTP Bank (Hungary)
8. CIB Bank (Italy)
9. Deutsche Bank (Germany)
10. Commerzbank (Germany)

53

Table 2.8 Firm Industry (NACE Main)

Industry

Frequency Percent

A - Agriculture, forestry and fishing
B - Mining and quarrying
C - Manufacturing
D - Electricity, gas, steam and air conditioning supply
E - Water supply; sewerage, waste management and remediation activities
F - Construction
G - Wholesale and retail trade; repair of motor vehicles and motorcycles
H - Transportation and storage
I - Accommodation and food service activities
J - Information and communication
K - Financial and insurance activities
L - Real estate activities
M - Professional, scientific and technical activities
N - Administrative and support service activities
O - Public administration and defence; compulsory social security
P - Education
Q - Human health and social work activities
R - Arts, entertainment and recreation
S - Other service activities
Total

4,591
620
49,542
437
1,271
11,667
53,203
17,431
1,983
13,384
768
2,206
22,240
7,119
16
716
649
1,167
1,568
190,578

2.41
0.33
26.00
0.23
0.67
6.12
27.92
9.15
1.04
7.02
.40
1.16
11.67
3.74
0.01
.38
0.34
.61
.82
100

54

Table 2.9 Export Results for All Firms

Dependent Variable: log(Exports)

(1)

(2)

(3)

(4)

0.733***
(0.0530)

0.706***
(0.0537)

0.798***
(0.0466)

0.793***
(0.0458)
0.272***
(0.0871)

0.746***
(0.0332)
0.0108***
(0.000530)

0.0532
(0.0377)

0.740***
(0.0327)
0.0108***
(0.000529)

0.0515
(0.0378)

-0.117***
(0.0159)
-0.250*
(0.138)

-0.116***
(0.0157)
-0.250*
(0.138)

0.279***
(0.0822)
0.177**
(0.0862)
0.291***
(0.0587)
0.729***
(0.0345)
0.0109***
(0.000529)

0.0477
(0.0380)
0.0682***
(0.0151)
-0.116***
(0.0158)
-0.250*
(0.138)

0.00657***
(0.00137)

0.00659***
(0.00138)

0.00630***
(0.00130)

N
Y
Y
Y

N
Y
Y
Y

N
Y
Y
Y

MNA

“Bank" Indicator

Same Country

Foreign Borrower (c)

Foreign Borrower (e)

Size

Profitability

Capital City

# of Non-Bank Advisors

Tangible Asset Share

Debt/Sales Ratio

Cash Flows

Individual Bank Controls
Industry Effects
Firm Country Effects
Year Effects

0.172*
(0.0905)
0.0816
(0.0763)
0.263***
(0.0521)
0.724***
(0.0344)
0.0109***
(0.000529)

0.0305
(0.0385)
0.0678***
(0.0145)
-0.114***
(0.0158)
-0.250*
(0.138)

0.00528***
(0.00112)

Y
Y
Y
Y

Observations
Number of bvdid
Robust standard errors, clustered at the NACE 2 Digit Industry level, in parentheses

190,581
48,063

190,581
48,063

190,581
48,063

190,581
48,063

*** p<0.01, ** p<0.05, * p<0.1

55

Table 2.10 Export Results for Hungarian Firms

Dependent Variable: log(Exports)

(1)

(2)

(3)

(4)

0.537***
(0.0863)

0.537***
(0.0859)
0.728***
(0.162)

MNA

“Bank" Indicator

Same Country

Foreign Borrower (c)

Foreign Borrower (e)

# of Non-Bank Advisors

Individual Bank Controls
Industry Effects
Year Effects

N
Y
Y

N
Y
Y

0.412***
(0.100)

0.342***
(0.0970)

0.461***
(0.110)
0.283***
(0.110)
0.530***
(0.0728)
0.0267*
(0.0140)

N
Y
Y

0.294**
(0.132)
0.137
(0.101)
0.391***
(0.116)
0.0260*
(0.0138)

Y
Y
Y

Observations
Number of bvdid
Robust standard errors, clustered at the NACE 2 Digit Industry level, in parentheses

43,596
9,147

43,596
9,147

43,596
9,147

43,596
9,147

*** p<0.01, ** p<0.05, * p<0.1

Note: All columns include firm characteristics (Size, Profitability, and Capital City Indicator) and

financial characteristics (Tangible Asset Share, Cash Flows, and Debt/Sales Ratio).

56

Table 2.11 Export Results for Estonian Firms

Dependent Variable: log(Exports)

(1)

(2)

(3)

(4)

0.850***
(0.107)

0.839***
(0.104)
0.228*
(0.127)

MNA

“Bank" Indicator

Same Country

Foreign Borrower (c)

Foreign Borrower (e)

# of Non-Bank Advisors

Individual Bank Controls
Industry Effects
Year Effects

N
Y
Y

N
Y
Y

0.742***
(0.109)

0.733***
(0.109)

-0.00642
(0.189)
0.222
(0.173)
0.0426
(0.118)
0.236***
(0.0356)

N
Y
Y

-0.00413
(0.193)
0.211
(0.744)
0.0308
(0.106)
0.231***
(0.0407)

Y
Y
Y

Observations
Number of bvdid
Robust standard errors, clustered at the NACE 2 Digit Industry level, in parentheses

63,797
16,474

63,797
16,474

63,797
16,474

63,797
16,474

*** p<0.01, ** p<0.05, * p<0.1

Note: All columns include firm characteristics (Size, Profitability, and Capital City Indicator) and

financial characteristics (Tangible Asset Share, Cash Flows, and Debt/Sales Ratio).

57

Table 2.12 Export Results for Croatian Firms

Dependent Variable: log(Exports)

(1)

(2)

(3)

(4)

0.903***
(0.0592)

0.903***
(0.0591)
-0.0811
(0.0839)

MNA

“Bank" Indicator

Same Country

Foreign Borrower (c)

Foreign Borrower (e)

# of Non-Bank Advisors

Individual Bank Controls
Industry Effects
Year Effects

N
Y
Y

N
Y
Y

0.892***
(0.0629)

0.885***
(0.0592)

0.0806
(0.100)
0.00171
(0.0819)
0.0721
(0.0766)
0.144***
(0.0366)

N
Y
Y

0.0665
(0.100)
0.0259
(0.107)
-0.0654
(0.0878)
0.159***
(0.0365)

Y
Y
Y

Observations
Number of bvdid
Robust standard errors, clustered at the NACE 2 Digit Industry level, in parentheses

83,188
22,442

83,188
22,442

83,188
22,442

83,188
22,442

*** p<0.01, ** p<0.05, * p<0.1

Note: All columns include firm characteristics (Size, Profitability, and Capital City Indicator) and

financial characteristics (Tangible Asset Share, Cash Flows, and Debt/Sales Ratio).

58

Table 2.13 Export Results for Small/Medium Enterprises (≤ 250 Employees)

Dependent Variable: log(Exports)

(1)

(2)

(3)

(4)

MNA

“Bank" Indicator

Same Country

Foreign Borrower (c)

Foreign Borrower (e)

# of Non-Bank Advisors

Individual Bank Controls
Industry Effects
Firm Country Effects
Year Effects

0.831***
(0.0525)

0.825***
(0.0516)
0.280***
(0.0877)

N
Y
Y
Y

N
Y
Y
Y

0.768***
(0.0611)

0.743***
(0.0619)

0.244**
(0.0959)
0.204**
(0.0840)
0.277***
(0.0578)
0.0819***
(0.0153)

N
Y
Y
Y

0.165
(0.103)
0.0565
(0.0763)
0.227***
(0.0532)
0.0820***
(0.0149)

Y
Y
Y
Y

Observations
Number of bvdid
Robust standard errors, clustered at the NACE 2 Digit Industry level, in parentheses

183,665
47,226

183,665
47,226

183,665
47,226

183,665
47,226

*** p<0.01, ** p<0.05, * p<0.1

Note: All columns include firm characteristics (Size, Profitability, and Capital City Indicator) and

financial characteristics (Tangible Asset Share, Cash Flows, and Debt/Sales Ratio).

59

Table 2.14 Export Results for Large Firms (> 250 Employees)

Dependent Variable: log(Exports)

(1)

(2)

(3)

0.129
(0.118)

0.130
(0.118)
0.930**
(0.402)

MNA

“Bank" Indicator

Same Country

Foreign Borrower (c)

Foreign Borrower (e)

# of Non-Bank Advisors

Individual Bank Controls
Industry Effects
Firm Country Effects
Year Effects

N
Y
Y
Y

N
Y
Y
Y

0.0541
(0.120)

0.401**
(0.168)
0.164
(0.208)
0.542***
(0.209)
-0.0180
(0.0235)

N
Y
Y
Y

(4)

0.0133
(0.125)

0.237
(0.190)
0.328
(0.237)
0.638**
(0.276)
-0.0135
(0.0246)

Y
Y
Y
Y

Observations
Number of bvdid
Robust standard errors, clustered at the NACE 2 Digit Industry level, in parentheses

6,916
1,384

6,916
1,384

6,916
1,384

6,916
1,384

*** p<0.01, ** p<0.05, * p<0.1

Note: All columns include firm characteristics (Size, Profitability, and Capital City Indicator) and

financial characteristics (Tangible Asset Share, Cash Flows, and Debt/Sales Ratio).

60

Table 2.15 Domestic Results for All Firms

Dependent Variable: log(Domestic Revenue)

(1)

(2)

(3)

MNA

“Bank" Indicator

Same Country

Foreign Borrower (c)

Foreign Borrower (e)

# of Non-Bank Advisors

Individual Bank Controls
Industry Effects
Firm Country Effects
Year Effects

0.0152
(0.0734)

-0.00179
(0.0731)
0.801***
(0.0864)

N
Y
Y
Y

N
Y
Y
Y

-0.0466
(0.0707)

0.133**
(0.0672)
0.636***
(0.0541)
0.420***
(0.0526)
0.179***
(0.0129)

N
Y
Y
Y

(4)

-0.0146
(0.0702)

0.0727
(0.0696)
-0.150***
(0.0376)
-0.0256
(0.0638)
0.165***
(0.0123)

Y
Y
Y
Y

Observations
Number of bvdid
Robust standard errors, clustered at the NACE 2 Digit Industry level, in parentheses

183,095
46,657

183,095
46,657

183,095
46,657

183,095
46,657

*** p<0.01, ** p<0.05, * p<0.1

Note: All columns include firm characteristics (Size, Profitability, and Capital City Indicator) and
financial characteristics (Tangible Asset Share, Cash Flows, and Debt/Sales Ratio).

61

Table 2.16 Domestic Results by Country

Dependent Variable: log(Domestic Revenue)

Hungary

(2)

(1)

Estonia

(4)

(3)

Croatia

(6)

(5)

MNA

Same Country

0.0351
(0.0731)

Foreign Borrower (c)

Foreign Borrower (e)

# of Non-Bank Advisors

Individual Bank Controls
Industry Effects
Year Effects

N
Y
Y

0.0197
(0.0666)
-0.0335
(0.0534)
-0.0801**
(0.0400)
-0.0152
(0.0496)
0.0921***
(0.00876)

Y
Y
Y

0.00503
(0.103)

N
Y
Y

-0.204**
(0.0947)
-0.0927
(0.141)
-0.182
(0.264)
0.377***
(0.0601)
0.389***
(0.0290)

Y
Y
Y

-0.00251
(0.0626)

N
Y
Y

0.0159
(0.0724)
0.215
(0.147)
-0.227***
(0.0633)
-0.744***
(0.0633)
0.406***
(0.0273)

Y
Y
Y

Observations
Number of bvdid

43,201
9,119

43,201
9,119

60,204
15,848

60,204
15,848

79,690
21,690

79,690
21,690

Robust standard errors, clustered at the NACE 2 Digit Industry level, in parentheses

*** p<0.01, ** p<0.05, * p<0.1

Note: All columns include firm characteristics (Size, Profitability, and Capital City Indicator) and
financial characteristics (Tangible Asset Share, Cash Flows, and Debt/Sales Ratio).

62

Table 2.17 Domestic Results by Firm Size

Dependent Variable: log(Domestic Revenue)

SME (≤ 250 Employees) Large Firms (> 250 Employees)

(1)

(2)

(3)

(4)

MNA

Same Country

0.0171
(0.0779)

Foreign Borrower (c)

Foreign Borrower (e)

# of Non-Bank Advisors

Individual Bank Controls
Industry Effects
Firm Country Effects
Year Effects

N
Y
Y
Y

-0.0225
(0.0755)
0.0725
(0.0780)
-0.187***
(0.0417)
-0.0423
(0.0682)
0.184***
(0.0144)

Y
Y
Y
Y

-0.0448
(0.0586)

N
Y
Y
Y

-0.0336
(0.0660)
-0.0261
(0.138)
0.0985
(0.129)
0.133
(0.125)
0.0295*
(0.0159)

Y
Y
Y
Y

Observations
Number of bvdid

176,240
45,820

176,240
45,820

6,855
1,381

6,855
1,381

Robust standard errors, clustered at the NACE 2 Digit Industry level, in parentheses

*** p<0.01, ** p<0.05, * p<0.1

Note: All columns include firm characteristics (Size, Profitability, and Capital City Indicator) and
financial characteristics (Tangible Asset Share, Cash Flows, and Debt/Sales Ratio).

63

Table 2.18 Difference in Differences Results for All Firms

Dependent Variable:

Loans

Loans

Log(Exports) Log(Exports)

MKB or Budapest Bank Borrower

0.361**
(0.155)
-0.392***
(0.0935)
0.0481
(0.210)

N

0.339**
(0.155)
-0.379***
(0.0936)
0.0692
(0.211)

Y

1.23

1.23

43,931

43,931

Robust standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

Post Acquisition

Post x MKB or BB

Industry Controls

Outcome Mean

Observations

-0.312***
(0.0341)
-0.0541*
(0.0295)
0.0257
(0.0530)

N

6.47

43,966

-0.313***
(0.0324)
-0.0554**
(0.0272)
0.0514
(0.0496)

Y

6.47

43,966

Note: This table contains the estimated coefficients of Equation 8 where y = Loans (in millions of
USD) or Log(Exports) using the entire sample of firms. All columns include firm characteristics
(Size, Profitability, and Capital City Indicator) and financial characteristics (Tangible Asset Share,
Cash Flows, and Debt/Sales Ratio).

64

Table 2.19 Difference in Differences Results for Domestic Firms

Dependent Variable:

Loans

Loans

Log(Exports) Log(Exports)

MKB or Budapest Bank Borrower

0.381***
(0.105)
-0.158***
(0.0546)
-0.268*
(0.146)

N

0.378***
(0.103)
-0.162***
(0.0538)
-0.274*
(0.144)

Y

1.05

1.05

36,890

36,890

Robust standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

Post Acquisition

Post x MKB or BB

Industry Controls

Outcome Mean

Observations

-0.170***
(0.0358)
-0.0675**
(0.0327)
0.0336
(0.0559)

N

6.22

36,921

-0.194***
(0.0340)
-0.0676**
(0.0305)
0.0634
(0.0525)

Y

6.22

36,921

Note: This table contains the estimated coefficients of Equation 8 where y = Loans (in millions
of USD) or Log(Exports) for domestically owned firms in the sample. All columns include firm
characteristics (Size, Profitability, and Capital City Indicator) and financial characteristics (Tangible
Asset Share, Cash Flows, and Debt/Sales Ratio).

65

Table 2.20 Difference in Differences Results for Multinational Affiliates

Dependent Variable:

Loans

Loans

Log(Exports) Log(Exports)

MKB or Budapest Bank Borrower

1.675
(1.278)
-1.097***
(0.425)
2.373
(2.191)

N

2.153*
(1.284)
-1.139***
(0.432)
1.925
(2.092)

Y

2.21

2.21

7,041

7,041

Robust standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

Post Acquisition

Post x MKB or BB

Industry Controls

Outcome Mean

Observations

-0.380***
(0.119)
0.00165
(0.0628)
0.122
(0.186)

N

7.81

7,045

-0.106
(0.103)
0.00334
(0.0540)
0.0431
(0.152)

Y

7.81

7,045

Note: This table contains the estimated coefficients of Equation 8 where y = Loans (in millions of
USD) or Log(Exports) for multinational affiliates. All columns include firm characteristics (Size,
Profitability, and Capital City Indicator) and financial characteristics (Tangible Asset Share, Cash
Flows, and Debt/Sales Ratio).

66

Figure 2.1 Example From the Data

Note: This figure shows an example of the list of banks used by two large multinational affiliates in Hungary (Lidl and Louis Vuitton).

67

Figure 2.2 Event Study results for Exports (All Firms)

Notes: This figure plots the estimated coefficients from Equation 10 where y=Log(Exports) using

the entire sample of firms. The coefficient when T=0 is omitted and therefore equal to zero.

68

-.2-.10.1Coefficient on Log(ExportRevenue)-5-4-3-2-112TimeParallel Trends Test for Exports (All Firms)Figure 2.3 Event Study results for Loans (All Firms)

Notes: This figure plots the estimated coefficients from Equation 10 where y=Loans (in millions
of 2016 USD) using the entire sample of firms. The coefficient when T=0 is omitted and therefore

equal to zero.

Figure 2.4 Event Study results for Exports (Domestic)

Notes: This figure plots the estimated coefficients from Equation 10 where y=Log(Exports) for

domestic firms. The coefficient when T=0 is omitted and therefore equal to zero.

69

-1-.50.51Loans (Millions of USD)-5-4-3-2-112TimeParallel Trends Test for Loans (All Firms)-.2-.10.1Coefficient on Log(ExportRevenue)-5-4-3-2-112TimeParallel Trends Test for Exports (Domestic Firms)Figure 2.5 Event Study results for Loans (Domestic)

Notes: This figure plots the estimated coefficients from Equation 10 where y=Loans (in millions
of 2016 USD) for domestic firms. The coefficient when T=0 is omitted and therefore equal to zero.

Figure 2.6 Event Study results for Exports (MNA)

Notes: This figure plots the estimated coefficients from Equation 10 where y=Log(Exports) for

MNAs. The coefficient when T=0 is omitted and therefore equal to zero.

70

-1-.50.51Loans (millions of USD)-5-4-3-2-112TimeParallel Trends Test for Loans (Domestic Firms)-.6-.4-.20.2.4Coefficient on Log(ExportRevenue)-5-4-3-2-112TimeParallel Trends Test for Exports (Multinational Affiliates)Figure 2.7 Event Study results for Loans (MNA)

Notes: This figure plots the estimated coefficients from Equation 10 where y=Loans (in millions

of 2016 USD) for MNAs. The coefficient when T=0 is omitted and therefore equal to zero.

71

-15-10-505Loans (Millions of USD)-5-4-3-2-112TimeParallel Trends Test for Loans (Multinational Affiliates)Table 2.21 Size and History of Banks- Hungary

HQ Country # Branches (2015) Established
China
France
Hungary
Italy
USA
Germany
Germany
Austria
Hungary
Netherlands
South Korea

1
1
95
95
10
1
1
129
2
1
1
210
16

Bank
Bank of China
BNP Paribas
Budapest Bank
CIB bank
Citibank
Commerzbank
Deutsche Bank
Erste & Steiermärkische
Gránit Bank
ING
KDB
Kereskedelmi és Hitelbank Zrt Belgium
Hungary
Kinizsi Bank
Központi Bank
Hungary
Hungary
MagNet Bank
Hungary
Merkantil Bank
MKB (MBK)
Hungary
Austria
Oberbank
Hungary
OTP Bank
Hungary
Polgári Bank
Raiffeisen Bank Zrt
Austria
Russia
Sberbank
UniCredit
Italy

2002 (Rebranded in 2015)
1991
1986
1979
1985
1993
1995
1998
1985 (Latest rebranding in 2010)
1991 (Rebranded 2008 after financial crisis)
1989
1986
savings bank (1958); commercial bank (2007)
1993
1995
1988
1950
2007
1949
1972 (Latest rebranding in 2015)
1986
Austrian Volksbank (1993) (Became Sberbank in 2012)
1990

14
1
81
8
388
22
68
30
56

72

Table 2.22 Size and History of Banks- Estonia

HQ Country
Latvia
Estonia
Estonia
Denmark
Estonia
Finland/Norway

Bank
AS Citadele banka
Bigbank
Coop Pank
Danske Bank
LHV Pank
Luminor Bank
Marfin Pank (Versobank) Estonia
OP Corporate Bank
Finland
Sweden
SEB Pank
Sweden
Svenska Handelsbanken
Sweden
Swedbank
Tallinna Aripank
Estonia
Italy
UniCredit Bank

Market Share (2016) Established
0.71%
1.81%
1.45%
6.16%
6.83%
14.16%
1.16%
2.56%
23.42%
0.58%
39.43%
0.92%
Closed in 2013

2010
1992
1992
2007 (Acquired Sampo Bank in 2000)
1999
2017 (Formerly Nordea (1995) and DNB (2006))
Versobank (1999) (Renamed Marfin in 2008)
1991
1998
1995-2002 (Reopened in 2006)
1991 (as Hansabank) (Rebranded in 2009)
1991 (oldest commercial bank in Estonia)
N/A

73

Table 2.23 Size and History of Banks- Croatia

HQ Country Market Share (2016) Established
Bank
Austria
Addiko Bank d.d.
Croatia
Banka Kovanica
Croatia
Croatia Banka
Erste & Steiermärkische
Austria
Hrvatska Po˘stanska Banka (HPB) Croatia
Istarska Kreditna Banka
Croatia
Karlova˘ca Banka
Croatia
Hungary
OTP Bank
Croatia
Podravska Banka
Primorska Banka
Croatia
Italy
Privredna Banka (Intesa)
Austria
Raiffeisenbank
Sberbank
Russia
Hungary
Splitska banka d.d.
Zagrebacka Banka (UniCredit)
Italy

2015 formerly Hypo Alpe-Adria-Bank (1990s)
1997
1989
1997-2002 (through acquisitions of 4 smaller banks)
1991
1956
1856
2005
1872
2001
1966, privatized in 1999 merged with Sanpaolo in 2007
1994
Austrian Volksbank (1997) became Sberbank in 2012
1965, purchased by UniCredit in 2002, and OTP in 2017
1914, purchased by UniCredit in 2002

5.32%
0.29%
0.62%
14.15%
4.88%
0.84%
0.55%
4.92%
0.82%
0.14%
18.27%
7.92%
2.23%
6.82%
26.51%

74

Old country New country Firms affected Percentage

US
DE
US
AT
N/A
N/A

FI/DK
DK
RU
IT
CY/GR

HU
HU
AT
RU
HU
CN

US
EE
EE
N/A
EE

US
RU

4591
4592
2978
1178
71
49

2535
2283
99
47
13

7196
1897

14.22%
14.23%
9.23%
3.65%
0.24%
0.15%

3.07%
2.77%
0.12%
0.06%
0.02%

10.83%
2.85%

Bank
Hungary
Budapest Bank
MKB
Citibank
Sberbank
Polgári Bank
Bank of China

Change

Purchased by Hungarian Government
Purchased by Hungarian Government
Withdrew from Hungarian retail market
Purchased Volksbank
Opened
Opened in Hungary

Table 2.24 Mergers and Acquisitions

Year of Change Old owner

2015
2014
2015
2012-2013
2015
2015

GE Capital
Bayern LB
Citibank
Volksbank (Austria)
N/A
N/A

New owner

Hungary
Hungary
Erste Bank
Sberbank
Polgari Bank
Bank of China

Estonia
Luminor Bank
Danske Bank
Coop Pank
UniCredit
Marfin (Versobank) Purchased by Estonian Governemnt

Acquired branches of Nordea and DNB
Capital shares acquired by LHV
Purchased by Inbank and Coop Pank (formerly Eesti Krediidipank)
Closed

2017
2016
2017
2013
2012

Nordea and DNB
Dankse Bank
Bank of Moscow
Unicredit
Bank of Greece/Cyprus Marfin Popular Bank Estonia

Blackstone Group
LHV
Coop Pank
N/A

Croatia
Addiko Bank
Sberbank

Purchased by Advent International (80%) and ERBD (20%)
Purchased Volksbank

2014
2012-2013

Hypo Alpe-Adria-Bank International A.G.
Volksbank (Austria)

Advent International/ERBD AT
Sberbank
AT

75

Table 2.25 Robustness Check:Total Revenue Results for All Firms

Dependent Variable: log(Total Revenue)

(1)

(2)

(3)

(4)

MNA

“Bank Indicator"

Same Country

Foreign Borrower (c)

Foreign Borrower (e)

# of Non-Bank Advisors

Individual Bank Controls
Industry Effects
Firm Country Effects
Year Effects

0.338***
(0.0224)

0.329***
(0.0217)
0.523***
(0.0379)

N
Y
Y
Y

N
Y
Y
Y

0.295***
(0.0194)

0.305***
(0.0197)

0.117***
(0.0318)
0.399***
(0.0274)
0.331***
(0.0217)
0.122***
(0.00819)

N
Y
Y
Y

0.0391
(0.0348)
-0.0504**
(0.0233)
0.0840**
(0.0338)
0.116***
(0.00784)

Y
Y
Y
Y

Observations
Number of bvdid
Robust standard errors, clustered at the NACE 2 Digit Industry level, in parentheses

190,583
48,063

190,583
48,063

190,583
48,063

190,583
48,063

*** p<0.01, ** p<0.05, * p<0.1

Note: All columns include firm characteristics (Size, Profitability, and Capital City Indicator) and
financial characteristics (Tangible Asset Share, Cash Flows, and Debt/Sales Ratio).

76

Table 2.26 Robustness Check:Total Revenue Results by Country

Dependent Variable: log(Total Revenue)

Hungary

(2)

(1)

Estonia

(3)

(4)

(5)

Croatia

(7)

MNA

Same Country

0.239***
(0.0384)

Foreign Borrower (c)

Foreign Borrower (e)

# of Non-Bank Advisors

Individual Bank Controls
Industry Effects
Year Effects

N
Y
Y

0.192***
(0.0394)
0.00363
(0.0431)
0.0353
(0.0301)
0.171***
(0.0323)
0.0572***
(0.00492)

Y
Y
Y

0.375***
(0.0331)

N
Y
Y

0.228***
(0.0295)
-0.0900
(0.0926)
-0.0329
(0.134)
0.182***
(0.0330)
0.334***
(0.0240)

Y
Y
Y

0.362***
(0.0395)

N
Y
Y

0.383***
(0.0436)
0.0753
(0.0619)
-0.104**
(0.0455)
-0.440***
(0.0398)
0.294***
(0.0174)

Y
Y
Y

Observations
Number of bvdid

43,598
9,147

43,598
9,147

63,797
16,474

63,797
16,474

83,188
22,442

83,188
22,442

Robust standard errors, clustered at the NACE 2 Digit Industry level, in parentheses

*** p<0.01, ** p<0.05, * p<0.1

Note: All columns include firm characteristics (Size, Profitability, and Capital City Indicator) and
financial characteristics (Tangible Asset Share, Cash Flows, and Debt/Sales Ratio).

77

Table 2.27 Robustness Check:Total Revenue Results by Firm Size

Dependent Variable: log(Total Revenue)

SME (≤ 250 Employees) Large Firms (> 250 Employees)

(1)

(2)

(3)

(4)

MNA

Same Country

0.345***
(0.0244)

Foreign Borrower (c)

Foreign Borrower (e)

# of Non-Bank Advisors

Individual Bank Controls
Industry Effects
Firm Country Effects
Year Effects

N
Y
Y
Y

0.309***
(0.0217)
0.0358
(0.0371)
-0.0938***
(0.0237)
0.0617*
(0.0357)
0.131***
(0.00694)

Y
Y
Y
Y

0.0844***
(0.0278)

N
Y
Y
Y

0.0782***
(0.0246)
-0.00974
(0.0587)
0.155**
(0.0743)
0.149**
(0.0680)
0.00616
(0.00750)

Y
Y
Y
Y

Observations
Number of bvdid

183,667
47,226

183,667
47,226

6,916
1,384

6,916
1,384

Robust standard errors, clustered at the NACE 2 Digit Industry level, in parentheses

*** p<0.01, ** p<0.05, * p<0.1

Note: All columns include firm characteristics (Size, Profitability, and Capital City Indicator) and
financial characteristics (Tangible Asset Share, Cash Flows, and Debt/Sales Ratio).

78

Table 2.28 Robustness Check: Export Results for All Firms (Controlling for Bank Country)

Dependent Variable: log(Exports)

(1)

(2)

(3)

(4)

MNA

“Bank" Indicator

Same Country

Foreign Borrower (c)

Foreign Borrower (e)

# of Non-Bank Advisors

Bank Country Controls
Industry Effects
Firm Country Effects
Year Effects

0.798***
(0.0466)

0.793***
(0.0458)
0.272***
(0.0871)

N
Y
Y
Y

N
Y
Y
Y

0.733***
(0.0530)

0.712***
(0.0553)

0.170*
0.279***
(0.0920)
(0.0822)
0.107
0.177**
(0.0849)
(0.0862)
0.328***
0.291***
(0.0587)
(0.0563)
0.0682*** 0.0670***
(0.0151)
(0.0141)

N
Y
Y
Y

Y
Y
Y
Y

Observations
Number of bvdid

190,581
48,063

190,581
48,063

190,581
48,063

190,581
48,063

Robust standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

Note: All columns include firm characteristics (Size, Profitability, and Capital City Indicator) and

financial characteristics (Tangible Asset Share, Cash Flows, and Debt/Sales Ratio).

79

Table 2.29 Robustness Check: Export Results for All Firms (Different MNA Definition)

Dependent Variable: log(Exports)

(1)

(2)

(3)

(4)

0.869***
(0.0631)

0.864***
(0.0621)
0.268***
(0.0867)

MNA

“Bank" Indicator

Same Country

Foreign Borrow (c)

Foreign Borrow (e)

# of Non-Bank Advisors

Individual Bank Controls
Industry Effects
Firm Country Effects
Year Effects

N
Y
Y
Y

N
Y
Y
Y

0.812***
(0.0715)

0.775***
(0.0719)

0.166*
(0.0871)
0.180**
(0.0867)
0.284***
(0.0582)
0.0666***
(0.0154)

N
Y
Y
Y

0.0738
(0.0962)
0.0777
(0.0766)
0.253***
(0.0520)
0.0662***
(0.0147)

Y
Y
Y
Y

Observations
Number of bvdid
Robust standard errors, clustered at the NACE 2 Digit Industry level, in parentheses

190,581
48,063

190,581
48,063

190,581
48,063

Note: All columns include firm characteristics (Size, Profitability, and Capital City Indicator) and

financial characteristics (Tangible Asset Share, Cash Flows, and Debt/Sales Ratio).

*** p<0.01, ** p<0.05, * p<0.1

190,581
48,063

80

CHAPTER 3

BANKRUPTCY AND INTERNATIONAL INTERVENTION: THE CASE OF ADDIKO

BANK

3.1 Introduction

The Balkans region is no stranger to turmoil and change. One aspect of the region that is often
neglected or forgotten, but is just as tumultuous as Balkan politics, is the financial sector. Following
the fall of the Iron Curtain and the dissolution of Yugoslavia, there was a major restructuring of the
banking sector in all Balkan countries. This was done during the economic transition to open market
economies. In the period after the transition, there has been financial restructuring following the
Great Recession, and agin during a country’s EU candidacy. These changes have largely been done
with the goal of enriching the country’s financial sector hoping to attract more foreign investment
and improving open market operations in the Balkans.

This paper exploits a change to the Croatian banking sector in 2014. The corrupt Austrian
bank Hypo Group Alpe Adria went bankrupt and was purchased and rebranded as Addiko bank.
When Hypo Group Alpe Adria went under, they had a large presence in many Balkan countries (not
only Croatia). This prompted the European Bank for Reconstruction and Development to assist
in the bank purchase in order to aid in the financial stabilization of the region. I am interested in
whether or not this turnover had positive effects on firms in Croatia, in effect justifying the need for
intervention from an international development bank.

Using a panel of exporting firms in Croatia, I find that while firm borrowing falls initially
following the turnover, that effect is short-lived and diminishes over time. Additionally, I find that
total revenue is largely unaffected by the bank purchase and domestic revenue increases slightly.
There is some additional evidence that firms responded to the bank turnover by switching from
exporting to domestic operations as a means of recovering from a shock to their loan amount.

81

The remainder of this paper proceeds as follows: section two provides a review of literature,
section three describes the data, section four specifies an empirical design, section five presents the
results, and section six concludes and discusses avenues for further research. Tables and figures
are presented in section seven.

3.2 Review of Literature and Background

Croatia is the most recent EU member, having joined in 2013 after a lengthy candidacy which
began in 2003 and became official in 2004. Following a similar timeline as Estonia and Hungary,
Croatia declared it’s independence from Communist Yugoslavia in 1992, and had already started the
privatization process in 1991. There was already an established banking system in place (Reininger
and Walko (2005)) with several domestic banks with establishment dates in the late 19th and early
20th century (see Table 2.23). This is a phenomenon unique to Croatia relative to other Central and
Eastern European countries. There are also several present-day banks that were established during
the transition to a market economy in the late 1990s. A few banks went through restructuring and
mergers after the global financial crisis, and many of the internationally-owned banks are from
countries in close geographic proximity.

3.2.1 Multinational Banking

The impact that international banks have can be ambiguous. Historically, international banks
opening, or acquiring, existing banks in less financially developed countries has led to inefficiently
run banks. In a 2010 paper, Weller argues that an increase in the number of multinational banks
in Poland after 1991 actually led to a negative effect on business investments due to lower credit
supplies by all Polish banks. This was an unintended consequence immediately following the
transition process. One goal of the transition process in CEEs was to achieve more financial
liberalization and in the long run develop a more sophisticated financial sector.

Chang, Hasan, and Hunter (2010) find that multinational banks are more inefficient than

82

domestic banks, however, their study is conducted using the US banking sector. One would
not expect the same dynamic to hold in less financially developed countries. Another paper by
Lensink, Meesters, and Naaborg (2008) studies the relationship between foreign ownership and bank
efficiency in Central Europe’s post-transition period (1991-2003). The authors measure inefficiency
in the standard way as expenses divided by revenue. During this time period, they studied over 2000
banks in over 100 countries scattered globally, including countries in Asia and Oceania. While they
find inefficiency in foreign-owned banks is present, this inefficiency diminishes with the quality
of host country governance and with the similarity between home and host country. Croatia is a
good example of an area in which this inefficiency would be reduced due to its close proximity to
Western Europe and EU membership goals during this period.

Additionally, De Haas and Van Lelyveld (2010) find multinational banks weather local financial
crises better than domestic banks. The internal capital markets afforded to them by financially strong
parent banks enable the subsidiaries to face minimal changes during periods of local crisis1. Another
paper by Havrylchyk and Jurzyk (2011) shows that Central and Eastern European banks acquired
by multinational banking groups experience increases in market share and profitability following
the turnover. These banks experience a transmission of knowledge from parent to subsidiary and
are thus at an advantage over banks that remain domestically owned. A similar phenomenon is
found between multinational enterprise parent companies and their affiliates.

3.2.2 The EBRD

The European Bank for Reconstruction and Development was founded in 1991 when several Eastern
and Central European countries were in the process of transitioning from a planned economy to a
market-based economy. Initially, the bank was founded to assist these Eastern Bloc countries with
their transition process. In modern times, the scope of the bank has grown and reached Central
Asia and other transitioning economies outside of Europe (EBRD.com). The EBRD also collects
1These results do not extend to global crises (2008). De Haas and Van Lelyveld (2014) show

that spillover effects existed between parent bank and subsidiary during the Great Recession.

83

and publishes data on bank performance in transition economies that aid in research on transition
ecnomies and particularly their financial sector2.

There are several criteria that must be met by a country before it can receive funds from the
EBRD. Most importantly, a country must be wholeheartedly committed to “multiparty democracy,
pluralism, and market economies" (Liu 2012). Liu argues in their 2012 paper that the conditions
mandated by the EBRD are inherently politically motivated, however they are “a form of good
international co-operation rather than that of willful or arbitrary intervention". This lends credence
to the assumption that the EBRD assisting in the purchase of Addiko bank was done in good faith
with “noble" intentions to improve the banking sector in Croatia.

In another paper by Shields, he argues that the EBRD has led to an increased presence of
neoliberalism in Central Europe. In effect, he claims that the conditions for EBRD assistance, and
their framework are inherently neoliberal (Shields 2020). When Hypo Group Alpe Adria went
under, they were a corrupt bank operating mainly in the Balkans. Therefore, the more neutral
neoliberal framework of the EBRD would be a welcome change of pace in a corrupt bank.

This paper assumes that the EBRD assisted in the purchase of Addiko Bank to replace a “bad
bank" and help increase the financial stability of the Balkan region, particularly Croatia as a recent
EU member. This assumption is supported by existing research on the EBRD and their own
mission statement. Enriching a transition economy’s financial sector has positive outcomes for
both households and firms in the country3, and is thus very important to study.

3.3 Data

The data in this analysis is from ORBIS, provided by Bureau van Dijk, and is comprised of exporting
firms from Croatia. Croatia is part of the AMADEUS (European) subset of ORBIS, and the panel
consists of the years 2010-2017. The firm-level data is collected annually from businesses through
websites and various business registries and reports. ORBIS is a well-known datasource used in

2For example Haselmann and Wachtel (2007) and Bjornskov and Potrafke (2011)
3Havrylchyk and Jurzyk (2011)

84

several papers on FDI, credit constraints, and multinational enterprises4. Croatia joined the EU in
2013, towards the beginning of the panel.

There are several advantages to this dataset. First, ORBIS provides detailed balance sheet data
for firms of all sizes (including unlisted firms). Examples of these variables are: cash flows, debt,
and fixed assets. These measures have all been used in previous papers to construct proxies for
financial constraints5. Additionally, ORBIS provides detailed ownership information on each firm
including the country of the global ultimate owner, and an independence indicator showing how
much autonomy a multinational affiliate has. Other useful variables from the dataset included in
my analysis are: NACE industry classification6, total employees, profitability, and city as these
provide other controls for the firm that might also affect firm performance. Summary statistics for
the key variables for the Croatian exporting firms are presented below in Table 3.1. For clarity,
I present the means for key variables for the full sample, then divide by firm ownership status,
then finally present the means for Addiko Borrowers (the treated group), and Non-Borrowers (the
control group). Overall, apporiximately 10% of firms are considered multinational affiliates by
my definition, and 11% are Addiko-Borrowers. It its evident that the Addiko borrowers are larger
in terms of most financial variables (e.g. debt, total assets, revenue). Therefore it is necessary to
control for financial characteristics in all analyses.

Another important feature of ORBIS is that it provides banking data for each firm. This is
presented as a list of “advisors". Advisors can be banks, private individuals, or other means of
financing that a firm uses. Table 3.2 presents the list of all possible banks in Croatia from which
exporting firms can borrow.

4For example Buch et al. (2014) uses ORBIS linked with German customs data to analyze the

effects of credit constraints on firm’s FDI behavior.

5See Buch et al. (2014)
6See Table 3.3 for a detailed breakdown of industry across all firms

85

3.4 Empirical Specification

When Hypo Group Alpe Adria went under in 2014, they had a large presence in the Balkans. This
prompted the EBRD to assist in the bank purchase in order to aid in the financial stabilization of
the region. Many countries in the Balkans are in the process of joining the EU and are following in
the footsteps of the successful Central European Economies. Therefore, sustaining strong financial
institutions in those countries is extremely important. In the dataset of Croatian exporters, 11%
borrow from Addiko bank. These firms serve as the treated group, while the other firms fall in the
control group.

To study the impact this turnover in bank ownership had on exporting firms in Croatia, I use

the following difference-in-differences model:

Yf it = α f + δt +

(cid:88)

γkD f k + ϕi + 
 f it

(3.1)

k

Which I estimate as:
y f i = β1 AddikoBorrower f + β2PostPurchase f + β3PostPurchase f ∗ AddikoBorrower f
+ βX f + γFin f ϕi + 
 f i
(3.2)
Where y f it is the measure of firm outcome: Log(TotalRevenue f i jt), Log(ExportRevenue f i jt),
Log(DomesticRevenue f i jt), or Loans f i jt; AddikoBorrower f is an indicator that takes a value
1 if firm f borrowers from Addiko Bank; PostPurchase f is a dummy variable that equals 1 if the
observation for firm f is in the period after 2014. β3 represents the causal effect of the turnover
to Advent International and EBRD on firm revenues and loans and is therefore the coefficient of
interest.

In order for the difference-in-difference estimation method to be valid, the parallel trends
assumption must be satisfied. Specifically, one must test whether the firms borrowing from Addiko
Bank had common trends in loan amount and revenues before the turnover as the unaffected firms
(non-Addiko-Bank borrowers). I can use the following related model to analyze pre and post trends
and test this assumption:

86

Yf it = α f + δt +

−1(cid:88)

t=m

g(cid:88)

t=1

βtT f t +

y f it = α f + δt +

βτ1(τ = t) ∗ AddikoBorrower f +

Which I estimate as:

−1(cid:88)

τ=m

γtK f t + ϕi + 
 f it

(3.3)

g(cid:88)

τ=1

βτ1(τ = t) ∗ AddikoBorrower f

+ϕi + 
 f it

(3.4)

Figure 3.1 through Figure 3.4 plot the estimated βs from estimating Equation 4 across all firms, then
separating multinational affiliates from domestic firms. As seen in and ?? and ??, the coefficients
before the purchase are statistically indistinguishable from zero, lending credence to the parallel
trend assumption for the exporting firms. One can therefore interpret the findings as causal.

3.5 Results

The main results for the paper are presented in Table 3.4 and Table 3.5. Table 3.4 shows the results
for the effect of the bank turnover on loans taken out by firms. I find that the turnover of Addiko
bank led to a $260,000 decrease in loans taken out by firms. However, this effect seems to occur
immediately after the turnover, and vanishes over time7. $260,000 is a non-trivial amount, since the
mean amount of loans taken out by firms is $240,000. This phenomenon could stem from the fact
that a significant restructuring of the bank was required after purchase. During and immediately
after restructuring, it would have been difficult to maintain the same lending portfolio for affected
firms.

Table 3.5 breaks the results for revenue down across total revenue, exporting revenue, and
domestic revenue. There is no significant effect of the bank turnover on total revenue or exports, yet
there is a slightly positive effect on domestic revenue. It is interesting to see that there is a slightly
positive impact of this bank turnover, but only on domestic firm performance. The turnover had a

7See Figure 3.1

87

small positive impact (or at least non-negative impact) on firm outcomes, which falls in line with
the story of the “noble" intentions of the EBRD to increase financial stability in Croatia, which
should ultimately benefit firms.

3.5.1 Firm Switching

One explanation for the increase in domestic revenue could be that firms are switching operations
from exporting to domestic activities. Since the amount of firm financing (loans) fell, firms would
need to switch to “cheaper" operations. It is well-known that exporting requires more financing
than domestic operations, both in fixed costs, and variable costs8. To analyze whether or not this is
occurring, I estimate the following equation:

ExportShare f i = β1 AddikoBorrower f + β2PostPurchase f
+ β3PostPurchase f ∗ AddikoBorrower f + βX f + γFin f ϕi + 
 f i

(3.5)

Where the effect on a firm’s export share would capture a firm switching from exporting to
domestic activities. A positive effect on export share indicated a firm exporting more following the
turnover whereas a negative coefficient on export share shows more domestic activity following the
bank purchase.

Table 3.6 presents the results for Equation 5. The coefficient on Post ∗ Addiko ( β3) is weakly
negative which provides some evidence that firms are switching out of exporting following the
purchase of Addiko bank. More analyses should be done to further explore this hypothesis.

3.6 Conclusion

When Hypo Group Alpe Adria went under, they had a large presence in the Balkans. This prompted
the EBRD to assist in the bank purchase in order to aid in the financial stabilization of the region.
The bank was then rebranded as Addiko bank. Many countries in the Balkans are in the process of

8Melitz (2003) presents the seminal framework for this phenomenon

88

joining the EU and are following in the footsteps of the successful Central European Economies.
Therefore, sustaining strong financial institutions in those countries is extremely important.

In this paper, I study the effect of the purchase of Hypo Group Alpe Adria on exporting firms’
financing (loans) and performance (revenue).
I find that the turnover of Addiko bank led to a
$260,000 decrease in loans taken out by firms. Importantly, this decrease occurs immediately after
the turnover, and then vanishes over time. Furthermore, I find no effect on firm exports, and total
revenue, and there is also a small increase in the firms’ domestic revenue after the purchase. I also
find weak evidence that firms are switching from exports to domestic activities when their loans
fall after the purchase of Addiko bank. These results indicate that after an initial period of turmoil,
the intervention by EBRD and Advent International had no lasting negative, and slightly positive
domestic effects on firm outcomes. The key contribution of this paper is the unique circumstances
of the bank turnover. Advent International and EBRD wished to operate the bank efficiently and
help stabilize the banking sector in the Balkans, and appeared to have been successful in Croatia.
There are several broad implications from this paper, the first is that a more sophisticated (and
consequently less corrupt) Croatian banking sector can lead to more foreign investment and an
increased presence of Croatia both in the EU and on a larger global scale. Additionally, the EBRD
intervention had a positive impact in Croatia. The “noble intentions" of this organization arguably
led to a small positive impact (or at least non-negative impact) on firm performance, contrasted with
the negative outcomes observed after the nationalistic Hungarian bank turnovers (Gabriel 2020).
This project creates several avenues for further research. Firstly, It would be interesting to see
how this bank turnover affected purely domestic operating firms, not just exporters. Secondly, more
work needs to be done to study whether or not firms change their behavior when they see a shock
to the amount of financing they receive.

89

APPENDIX

90

Variable

MNA

Number of Employees

Profitability

Total Revenue

Export Revenue

Domestic Revenue

Capital City

Cash Flows

Debt

Total Assets

Tangible Fixed Assets

Number of Banks

# of Non-Bank Advisors

Table 3.1 Summary Statistics- Firm Characteristics

Full Sample

MNA

Domestic

Addiko Borrowers Non-Addiko Borrowers

.10
(0.29)
31.12
(207.32)

6.57
(17.60)
4,219
(28,834)
1,124
(8,333)
3,094
(26,113)

0.41
(0.49)
0.36
(3.28)
973

(6,318)
4,086
(34,869)
1,629
(20,129)

1.50
(0.74)
0.29
(0.63)

62.95
(192.00)

2.71
(17.11)
11,324
(33,992)
4,319
(20,654)
6,915
(22,938)

0.53
(0.49)
.97
(5.58)
2,889
(15,174)
10,815
(3,617)
7,983
(18,927)

1.46
(0.69)
0.30
(0.62)

27.73
(208.61)

6.98
(17.60)
3,472
(28,126)

784

(5,496)
2,687
(26,396)

0.40
(0.49)
0.30
(2.93)
730

(4,382)
3,369
(32,517)
1,417
(20,241)

1.50
(0.74)
0.29
(0.63)

0.08
(0.26)
76.53
(458 )
4.97
(15.70)
10,717
(67,180)
2,022
(12,654)
8,695
(63,737)

0.32
(0.47)
0.81
(5.30)
2,170
(9,962)
10,788
(79,147)
4,318
(46,623)

2.11
(0.74)
0.48
(0.80)

0.10
(0.30)
24.95
(141.61)

6.79
(17.83)
3,336
(18,022)
1,003
(7,550)
2,333
(14,760)

0.42
(0.49)
0.30
(2.90)
770

(5,624)
3,175
(22,873)
1,264
(12,799)

1.42
(0.69)
0.26
(0.59)

Number of Observations

83,188

8,008

75,180

9,949

73,239

91

Table 3.2 List of Banks

HQ Country
Bank
Austria
Addiko
Croatia
Banka Kovanica
Croatia Banka
Croatia
Erste & Steiermärkische Austria
Hrvatska Po˘stanska
Croatia
Istarska Kreditna Banka Croatia
Karlova˘ca Banka
Croatia
Hungary
OTP
Podvravska Banka
Croatia
Croatia
Primorska Banka
Italy
Privrenda Bank
Austria
Raiffeisenbank
Sberbank
Russia
Hungary
Splitska Bank
UniCredit Bank
Italy

92

Table 3.3 Firm Industry (NACE Main)

Industry

Frequency Percent

A - Agriculture, forestry and fishing
B - Mining and quarrying
C - Manufacturing
D - Electricity, gas, steam and air conditioning supply
E - Water supply; sewerage, waste management and remediation activities
F - Construction
G - Wholesale and retail trade; repair of motor vehicles and motorcycles
H - Transportation and storage
I - Accommodation and food service activities
J - Information and communication
K - Financial and insurance activities
L - Real estate activities
M - Professional, scientific and technical activities
N - Administrative and support service activities
O - Public administration and defence; compulsory social security
P - Education
Q - Human health and social work activities
R - Arts, entertainment and recreation
S - Other service activities
Total

1,572
620
49,542
437
1,271
11,667
53,203
17,431
1,983
13,384
768
2,206
22,240
7,119
16
716
649
1,167
1,568
83,188

1.89
0.24
25.78
0.17
0.69
3.96
27.03
8.23
1.58
9.57
.16
.54
14.79
3.47
0.01
.34
0.28
.55
.73
100

93

Table 3.4 Difference in Differences Results for Loans

Dependent Variable:

Loans

Loans

Addiko Bank Borrower

Post Turnover

Post x Addiko

Other Controls

Outcome Mean

0.899***
(0.108)

-0.0695***
(0.0147)

-0.235*
(0.121)

N

.24

0.606***
(0.0922)
-0.0270*
(0.0156)

-0.262**
(0.116)

Y

.24

Observations

Robust standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

83,188

83,188

Note: This table contains the estimated coefficients of Equation 2 where y = Loans (in millions of

2016 USD) for the full sample. Other controls are: Industry, Size, Profitability, Capital City

Indicator, Tangible Asset Share, Cash Flows, and Debt/Sales Ratio.

94

Table 3.5 Difference in Differences Results (Revenue)

Dependent Variable:

Log(Total) Log(Total) Log(Exports) Log(Exports) Log(Domestic) Log(Domestic)

0.570***
(0.0408)
-0.167***
(0.0184)

0.0282
(0.0553)

N

4.20

0.00619
(0.0335)
-0.107***
(0.0152)

-0.0151
(0.0448)

Y

4.20

1.059***
(0.0355)
-0.338***
(0.0171)

0.111**
(0.0475)

N

5.80

0.275***
(0.0219)
-0.247***
(0.0122)

0.0527*
(0.0287)

Y

5.80

Addiko Bank Borrower

Post Turnover

Post x Addiko

Other Controls

Outcome Mean

Observations

0.893***
(0.0293)
-0.267***
(0.0129)

0.151***
(0.0136)
-0.189***
(0.00648)

0.0419
(0.0395)

-0.00858
(0.0181)

Y

6.46

N

6.46

83,188

83,188
Robust standard errors in parentheses

83,188

*** p<0.01, ** p<0.05, * p<0.1

83,188

79,690

79,690

Note: This table contains the estimated coefficients of Equation 2 for the full sample where y = Log(Revenue), Log(Exports), and Log
(Domestic). Other controls are: Industry, Size, Profitability, Capital City Indicator, Tangible Asset Share, Cash Flows, and Debt/Sales
Ratio.

95

Table 3.6 Difference in Differences Results for Export Share

Dependent Variable:

Export Share Export Share

Addiko Bank Borrower

Post Turnover

Post x Addiko

Other Controls

Outcome Mean

-0.0642***
(0.00521)
0.0218***
(0.00269)

-0.0119*
(0.00695)

N

-0.0379***
(0.00487)
0.0149***
(0.00254)

-0.00821
(0.00644)

Y

.33

.33

Observations

Robust standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

83,188

83,188

Note: This table contains the estimated coefficients of Equation 5 where y = Export Share for the
full sample. Other controls are: Industry, Size, Profitability, Capital City Indicator, Tangible Asset

Share, Cash Flows, and Debt/Sales Ratio.

96

Figure 3.1 Event Study results for Loans

Notes: This figure plots the estimated coefficients from Equation 4 where y=Loans (in millions of
2016 USD) using the entire sample of Croatian Exporters. The coefficient when T=0 is omitted

and therefore equal to zero.

Figure 3.2 Event Study results for Exports

Notes: This figure plots the estimated coefficients from Equation 4 where y=Log(Exports) using
the entire sample of Croatian Exporters. The coefficient when T=0 is omitted and therefore equal

to zero.

97

-.6-.4-.20.2.4.6Loans (millions of USD)-3-2-11234TimeParallel Trends Test for Loans-.2-.10.1.2Coefficient on Log(ExportRevenue)-3-2-11234TimeParallel Trends Test for ExportsFigure 3.3 Event Study results for Total Revenue

Notes: This figure plots the estimated coefficients from Equation 4 where y=Log(TotalRevenue)
using the entire sample of Croatian Exporters. The coefficient when T=0 is omitted and therefore

equal to zero.

Figure 3.4 Event Study results for Domestic Revenue

Notes: This figure plots the estimated coefficients from Equation 4 where

y=Log(DomesticRevenue) using the entire sample of Croatian Exporters. The coefficient when

T=0 is omitted and therefore equal to zero.

98

-.15-.1-.050.05.1.15Coefficient on Log(TotalRevenue)-3-2-11234TimeParallel Trends Test for Total Revenue-.15-.1-.050.05.1.15Coefficient on Long(DomesticRevenue)-3-2-11234TimeParallel Trends Test for Domestic RevenueTable 3.7 Size and History of Banks- Croatia

HQ Country Market Share (2016) Established
Bank
Austria
Addiko Bank d.d.
Croatia
Banka Kovanica
Croatia
Croatia Banka
Erste & Steiermärkische
Austria
Hrvatska Po˘stanska Banka (HPB) Croatia
Croatia
Istarska Kreditna Banka
Karlova˘ca Banka
Croatia
Hungary
OTP Bank
Croatia
Podravska Banka
Croatia
Primorska Banka
Privredna Banka (Intesa)
Italy
Austria
Raiffeisenbank
Russia
Sberbank
Splitska banka d.d.
Hungary
Italy
Zagrebacka Banka (UniCredit)

2015 formerly Hypo Alpe-Adria-Bank (1990s)
1997
1989
1997-2002 (through acquisitions of 4 smaller banks)
1991
1956
1856
2005
1872
2001
1966, privatized in 1999 merged with Sanpaolo in 2007
1994
Austrian Volksbank (1997) became Sberbank in 2012
1965, purchased by UniCredit in 2002, and OTP in 2017
1914, purchased by UniCredit in 2002

5.32%
0.29%
0.62%
14.15%
4.88%
0.84%
0.55%
4.92%
0.82%
0.14%
18.27%
7.92%
2.23%
6.82%
26.51%

99

BIBLIOGRAPHY

100

BIBLIOGRAPHY

[1] Adarov, Amat, and Robert Tchaidze. “Development of Financial Markets in Central Europe:

the Case of the CE4 Countries." IMF Working Papers (2011): 1-34.

[2] Alfaro, Laura, and Maggie X. Chen. “Surviving the Global Financial Crisis: Foreign Ownership
and Establishment Performance." American Economic Journal: Economic Policy 4, no. 3
(2012): 30-55.

[3] Antras, Pol, Mihir A. Desai, and C. Fritz Foley. “Multinational Firms, FDI Flows, and Imperfect

Capital Markets." Quarterly Journal of Economics 124 (2009): 1171-1219.

[4] Bevan, Alan A., and Saul Estrin. “The Determinants of Foreign Direct Investment into European

Transition Economies." Journal of Comparative Economics 32 (2004): 775-787.

[5] Bilir, Kamran L., Davin Chor, and Kalina Manova. “Host-Country Financial Development and

Multinational Activity." NBER Working Paper 20046 (2014).

[6] Bjornskov, Christian, and Niklas Potrafke. “Politics and privatization in Central and Eastern

Europe." Economics of Transition 19, no.2 (2011): 201-230.

[7] Buch, Claudia M., Iris Kesternich, Alexander Lipponer, and Monika Schnitzer. “Financial
Constraints and Foreign Direct Investment: Firm-Level Evidence." Review of World Economics
150 (2014): 393-420.

[8] Carluccio, Juan, and Thibault Fally. “Global Sourcing Under Imperfect Capital Markets."

Review of Economics and Statistics 94, no.3 (2012): 740-763.

[9] Carstensen, Kai, and Farid Toubal. “Foreign Direct Investment in Central and Eastern European
Countries: a Dynamic Panel Analysis." Journal of Comparative Economics 32 (2004): 3-22.
[10] Chang, C. Edward, Iftekhar Hasan, and William C. Hunter. “Efficiency of multinational banks:

an empirical investigation." Applied Financial Economics 8, no.6 (1998): 689-696.

[11] Cravino, Javier, and Andrei A. Levchenko. “Multinational Firms and International Business

Cycle Transmission." Quarterly Journal of Economics 132 (2017): 921-962.

[12] De Haas, Ralph, and Iman Van Lelyveld. “Foreign banks and credit stability in Central and
Eastern Europe. A panel data analysis." Journal of Banking and Finance 30, no.7 (2006):
1927-1952.

[13] De Haas, Ralph, and Iman Van Lelyveld. “Multinational Banks and the Global Financial
Crisis: Weathering the Perfect Storm?" Journal of Money, Credit and Banking 46, no.1 (2014):
333-364.

[14] Desai Mihir A., C. Fritz Foley and Kristin Forbes. “Financial Constraints and Growth: Multi-
national and Local Firm Responses to Currency Depreciations." Review of Financial Studies
21, no. 6 (2008): 2857-2888

101

[15] Desai Mihir A., C. Fritz Foley and James R. Hines Jr. “Domestic Effects of the Foreign
Activities of US Multinationals." American Economic Journal: Economic Policy 1, no.1
(2009): 181-203.

[16] Desai Mihir A., C. Fritz Foley and James R. Hines Jr. “A Multinational Perspective on
Capital Structure Choice and Internal Capital Markets." Journal of Finance 59, no.6 (2004):
2451-2487.

[17] Eesti Pank. “Financial Stability Review." Eesti Pank (2013): 1-47.

[18] Feenstra, Robert C., Zhiyuan Li, and Miaojie Yu. “Exports and Credit Constraints Under In-
complete Information: Theory and Evidence from China." Review of Economics and Statistics
96, no. 4 (2014): 729-744.

[19] Figlio, David N. and Bruce A. Blonigen. “The Effects of Foreign Direct Investment on Local

Communities." Journal of Urban Economics 48, no.2 (2000): 338-363.

[20] Fons-Rosen, Christian, Sebnem Kalemli- Ozcan, Bent E. Sorensen, Carolina Villegas-
Sanchez, and Vadym Volosovych. "Foreign Investment and Domestic Productivity: Identifying
knowledge spillovers and competition effects." NBER Working Papers No. 23643 (2018).

[21] Haselmann, Rainer, and Paul Wachtel. “Risk Taking by Banks in the Transition Countries."

Comparative Economic Studies 49 (2007): 411-429.

[22] Havrylchyk, Olena, and Emilia Jurzyk. “Inherited or earned? Performance of foreign banks
in Central and Eastern Europe." Journal of Banking and Finance 35, no.5 (2011): 1291-1302.

[23] Helpman, Elhanan. “A Simple Theory of International Trade with Multinational Corpora-

tions." Journal of Political Economy 92 (1984): 451-471.

[24] Helpman, Elhanan, Marc J. Melitz, and Stephen R. Yeaple. “Export Versus FDI with Hetero-

geneous Firms." American Economic Review 94 (2004): 300-316.

[25] Hericourt, Jerome, and Sandra Poncet. “FDI and Credit Constraints: Firm-Level Evidence

from China." Economic Systems 33, no.1 (2009): 1-21.

[26] Janicki, Hubert P., and Phanindra V. Wunnava. “Determinants of foreign direct investment:
empirical evidence from EU accession candidates." Applied Economics 36 no.5 (2004): 505-
509.

[27] Lansbury, Melanie, Nigel Pain, and Katerina Smidkova. “Foreign DIrect Investment in Central
Europe since 1990: and Econometric Study." National Institute Economic Review 156, no.1
(1996): 104-114.

[28] Lensink, Robert, Aljar Meesters, and Ilko Naaborg. “Bank Efficiency and Foreign Ownership:

Do Good Institutions Matter?" Journal of Banking and Finance 32, no.5 (2008): 834-844.

[29] Lipsey, Robert E. “Home and Host-Country Effects of Foreign Direct Investment," NBER

Chapters in: Challenges to Globalization: Analyzing the Economics (2004): 333-382.

102

[30] Liu, Yin. “Intervention vs. Co-Operation: Legitimacy of the EBRD Conditionality." Frontiers

of Law in China 7, no.2 (2012): 314-328.

[31] Manova, Kalina. “Credit Constraints, Heterogeneous Firms, and International Trade." Review

of Economic Studies 80 (2013): 711-744.

[32] Manova, Kalina, Shang-Jin Wei, and Zhiwei Zhang. “Firm Exports and Multinational Activity

Under Credit Constraints." Review of Economics and Statistics 97, no. 3 (2015): 574-588.

[33] Markusen, James R., and Anthony J. Venables. “Multinational Firms, and the New Trade

Theory." Journal of International Economics 46 (1998): 183-203.

[34] Melitz, Marc J. “The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry

Productivity." Econometrica 71, no.6 (2003): 1695-1725.

[35] Minetti, Raoul, and Susan C. Zhu. “Credit Constraints and Firm Export: Microeconomic

Evidence from Italy." Journal of International Economics 83 (2011): 109-125.

[36] OECD. “Estonia: Review of the Financial System" OECD Paris (2011): 1-49.

[37] Paravisini, Daniel, Veronica Rappoport, Philipp Schnabl, and Daniel Wolfenzon. “Dissecting
the Effect of Credit Supply on Trade: Evidence from Matched Credit-Export Data." Review of
Economics Studies Limited 82, no.1 (2014): 1-26.

[38] Reininger, Thomas, and Zoltan Walko. “The Croatian Banking System." OeNB Financial

Stability Report 9 (2005): 110-126.

[39] Shields, Stuart. “The EBRD, Fail Forward Neoliberalism and the Construction of the European

Periphery." The Economic and Labour Relations Review 32, no.2 (2020): 230-248.

[40] Vachudova, Milada Anna. “EU Leverage and National Interests in the Balkans: The Puzzles
of Enlargement Ten Years On." Journal of Common Market Studies 52, no.1 (2014): 122-138.
[41] Varhegyi, Eva “Hungary’s Banking Sector: Achievements and Challenges." EIB Papers 7,

no.1 (2002): 75-89.

[42] Vissak, Tiia, and Tonu Roolaht. “The Negative Impact of Foreign Direct Investment on the

Estonian Economy." Problems of Economic Transition 48, no.2 (2005): 43-66.

[43] Weller, Christian E. “Financial Liberalization, Multinational Banks and Credit Supply: The

case of Poland." International Review of Applied Economics 14, no.2 (2010): 193-211.

103