Martine Koonstra s2232960 m.koonstra@student.rug.nl Supervisor: dr. Gjalt de Jong University of Groningen Faculty of Economics and Business International Business & Management June 2013

1

M

ASTER

T

HESIS

The Impact of Political Risk on the Failure of

Cross-Border Mergers & Acquisitions in Central

and Eastern European Countries

Martine Koonstra s2232960 m.koonstra@student.rug.nl

Supervisor: dr. Gjalt de Jong

University of Groningen Faculty of Economics and Business International Business & Management

2 Abstract

A number of studies emphasize the high failure rate of mergers and acquisitions (M&As) in Central and Eastern European (CEE) countries. Given the transition process towards a market economy, foreign investors often have to deal with political turbulence and ambiguity. Prior research finds evidence for the impact of financial and economic risks in explaining M&A failure, however the explanatory power of political risk is mixed. Enhancing the limited literature in this field, we explore the relationship between political risk and M&A failure in CEE countries. Political risk is divided into five indicators which are examined during this paper: government stability, socioeconomic conditions, investment profile, internal and external conflict. We analyze a sample of 5485 cross-border M&A deals between 1997 and 2008. Findings point to the importance of the underlying indicators causing political instabilities. We find significant support for two of the five political risk indicators; a higher level of investment risk and internal conflict increase the failure rate of M&As in CEE countries. Contradictory to our hypothesis, the results indicate that a higher level of government instability decreases the failure rate of M&As. We find no support for the impact of socioeconomic conditions, external conflict, or the composite political risk ratings on M&A failure.

3 TABLE OF CONTENT

1.

INTRODUCTION

5

2.

LITERATURE REVIEW

6

2.1 How can cross-border M&As be defined? 6

2.2 What is the definition of M&A failure and how can it be measured? 7

2.3 What is the definition of political risk and how can it be measured? 10

2.4 What is the relationship between political risk and M&A failure? 11

2.5 Is there any evidence for this relationship? 12

3.

CONCEPTUAL MODEL & HYPOTHESES

13

3.1 Government stability 14 3.2 Socioeconomic conditions 14 3.3 Investment profile 15 3.4 Internal conflict 15 3.5 External conflict 16

4.

RESEARCH METHODS

16

4

5.

EMPIRICAL RESULTS

26

6.

ROBUSTNESS TEST

30

7.

CONCLUSION

31

7.1 Added value of this study 33

7.2 Limitations and future research 34

BIBLIOGRAPHY

36

APPENDICES

39

Appendix 1: CEE countries 39

Appendix 2: Heteroscedasticity 40

Appendix 3: Cronbach’s Alpha 41

Appendix 4: Correlation statistics 42

Appendix 5: Geographical analysis of the normality assumption 43

Appendix 6: Descriptive statistics 44

Appendix 7: Goodness-of-fit tests 46

Appendix 8: Stepwise Rare Event Logit regression results 47

5

1. INTRODUCTION

Since the fall of Communism, cross-border mergers and acquisitions (M&As) are increasingly prevalent in Central and Eastern European (CEE) countries. Trade liberalization policies have attracted foreign direct investment inflows, which has been vital for the privatization process during the transition period (Kornecki & Raghavan, 2011). Multinational enterprises (MNEs) from all over the world have increased their interest into these transition economies. Since the mid 1990s M&As have become a popular business development strategy in Europe, with deals in Europe now accounting for about half of the worldwide M&A volume (Huyghebaert & Luypaert, 2013). However, a number of studies emphasize that a lot of M&As fail. The paper of Craninckx & Huyghebaert (2011), for example, report that about half of their sample of 603 European takeovers fails to create shareholder value during 1997-2006. Similar results were found by Martynova & Renneboog (2011), approximately 50 % of their sample of 155 European M&As between 1997 and 2001 resulted in an operating performance decline for the combined firm. These results are disappointing and asks for further research in order to address the sources of failure in European countries.

Recognizing that even though M&As are most active within the developed world, they are also of increasing importance in the transition countries. However, most of the existing literature reflects the findings of the studies performed in the developed world, particularly the US (Kiymaz, 2004; 2009; Garfinkel & Hankins, 2010). The analysis of M&A activities involving European targets disposes unique contexts as cross-border deals are increasingly prevalent in this region. Little is known about European takeover market, where the privatization process plays an important role. For MNEs entering CEE countries the transition process towards a market economy may create considerable risks and thereby a turbulent environment. Foreign investors often have to deal with significant political turbulence and ambiguity, which is a serious concern due to the potential impact on profitability (Iankova & Katz, 2003).

Compared to domestic M&As, cross-border M&As present greater challenges for a buyer, because of differences in the host country environment that are unfamiliar to a foreign firm. Uncertainty reduces the value of the assets being exchanged and appears as an explanation to the poor performance of acquirers in cross-border M&As (Denis et al., 2002; Moeller & Schlingemann, 2005). Prior research argues that uncertainty is created by country risk, which is composed of economic, financial and political risk. Theoretical evidence indicates that financial and economic risk have a significant impact on the performance of M&As (Erb et al., 1996), however the results of political risk are mixed. It remains an empirical question whether political risk contributes to the failure of M&As.

Considering the lack of empirical studies with a focus on examining the sources of M&A failure in a CEE context, this thesis aims to investigate the relationship between political risk and the failure of M&As in CEE countries. Therefore, the main research question is:

6 In order to answer the research question the following sub questions are used:

1) How can cross-border M&As be defined?

2) What is the definition of M&A failure and how can it be measured? 3) What is the definition of political risk and how can it be measured? 4) What is the relationship between political risk and M&A failure? 5) Is there any evidence for this relationship?

The remainder of the paper is organized as follows. First I will provide a literature review of the theoretical and empirical work on the core concepts and derive the hypotheses. Thereafter the research method is introduced, along with the introduction of the data sample upon which the hypotheses will be tested. Following a discussion of the findings the conclusion is provided.

2. LITERATURE REVIEW

2.1 How can cross-border M&As be defined?

About one third of worldwide M&As combine firms from two different countries. As the world becomes increasingly integrated, cross-border mergers are likely to become even more important in the future (Erel et al., 2012). Cross-border M&As occur when two or more firms, whose headquarters are located in different home countries, join together all or part of their operations (Coyle, 2000). Hereafter, cross-border M&As are referred to as „M&As‟, it will be specifically indicated when M&As are domestic. The reasoning behind M&As is to create shareholders value over and above that of the sum of the two companies. In other words, the main purpose is to create synergy; however this seems to be more difficult in reality than in theory. To give a distinction between mergers and acquisitions, both are defined next. “A merger can refer to any takeover of one company by another, when the businesses of each company are brought together as one, involving a bidder and a target firm. An acquisition occurs when one company acquires from another company either a controlling interest in the company’s stocks or a business operation and its assets” (Coyle, 2000: 2,4). The distinction is important to lawyers, accountants and tax specialists, but less so in terms of economic impact. In business the terms are used interchangeably, so will be done in this thesis.

7 a small physical distance can decrease the costs of merging two firms. Besides, higher economic development and better accounting quality in host countries attract acquirer firms. According to Erel et al. (2012) the valuation effect determines to a large extent whether firms decide to merge, whereby more highly valued firms tend to purchase lower-valued firms. This effect can divided into two factors: currency movements and stock market performance. The former indicates that short-term movements between two countries‟ currencies increase the likelihood that firms in the country with the appreciating currency purchase firms in the country with the depreciating currency. Secondly, the greater the difference in stock market performance between the two countries, the more likely that firms in the superior-performing country purchase firms in a worse performing country (Erel et al., 2012). In sum, country level factors, such as currency appreciation and macro-economic performance, appear to make cross-border mergers more attractive than domestic mergers.

2.2 What is the definition of M&A failure and how can it be measured?

Following an influential study of Erel et al. (2012) the concept of M&A failure refers to withdrawn M&As; M&As that have an announcement date, but not an (expected) completion date. Zephyr, the database used to gather data on M&A deals, defines withdrawn M&As as follows: “the deal does not progress for some reason possibly because the parties cease negotiations or because the acquirer retracts the offer and makes an increased or decreased offer for the target”. The M&A is registered as successful when a completion date is added to the deal. Zephyr records this date when the deal has been formally announced as completed in the media or in certain circumstances has received all approvals to go ahead.

As already outlined the current literature emphasizes M&A performance measured in terms of wealth gains or future expected returns, instead of „the failure‟ of M&As. The focus is more on the success of M&As, however I analyze the failure of M&As as this is particularly interesting due to the fact that a great amount of M&As fail (Kale et al., 2009: 110; McKinsey, 2004). In a European context, evidence also suggest that cross-border takeovers destroy shareholder value (Huyghebaert & Luypaert, 2013; Craninckx & Huyghebaert, 2011; Martynova and Renneboog, 2011).

M&A performance is closely connected to whether M&As are successful or not. Therefore, this paragraph consists of the most discussed sources of a decline in performance (e.g. lower wealth gains or stock returns for acquirers). The sources include geographic and cultural distance, agency problems and other organization-related causes, as well as country risks.

8 2.2.1 Acquisition premiums and organizational causes of failure

Hitt et al. (2009) state that the primary reason for failure is an acquisition premium. An acquisition premium is the price paid for a target firm that exceeds its pre-acquisition market value. Over the past twenty years, the average premium paid has been 40%-50% (Laamanen, 2007). In practice, the value the M&As create together is overestimated and not enough to cover the high premiums paid. A large premium places a major burden on managers of the acquiring firm to recoup those costs and extract sufficient synergies from the merged firm (Hitt et al., 2009).

Furthermore, Hitt et al. (2009) argue that M&As fail, because of integration difficulties. The true value to an acquiring firm can only be captured if the valuable capabilities in the acquired firm are integrated into and absorbed by the acquiring firm. Interorganizational learning is of great importance in the integration process; however it is complex and depends on the absorptive capabilities of the acquiring firm. Selecting the right partner is crucial, since partners without complementary capabilities cannot contribute to the development of sustainable competitive advantages and therefore increases the likelihood of failure.

Moreover, Uhlenbruck and De Castro (2000) study the acquisition of privatized firms as a means of entering CEE countries, they are particularly interested in post privatization performance. One of their findings was that organizational fit between merging firms was negatively related to performance. They argue that acquired SOEs appear to require such fundamental transformation that historical similarities are unimportant. It might also be that Western managers misunderstood some of the organizational systems of SOEs in post-Communist countries and find it difficult to transform their targets efficiently.

2.2.2 Agency theory

Erel et al. (2012) argue that value-decreasing acquisitions are due to agency considerations, meaning that different objectives of the shareholders (principals) and managers (agents) can lead to opportunistic behavior. Agency theory states that managers act in their self-interest, which may result in conflicts with the shareholders (Hoskisson, 2000). Their financial goals are often different, for the manager the prestige of having a larger firm is more important than shareholder growth. Engaging in M&As increases the size of a firm, which is often beneficiary for a top executive‟s compensation and power (Hitt et al., 2009). Moreover, it is one of the reasons why CEOs have a greater incentive to pay high premiums for target firms as discussed earlier (see section 2.2.1). Unfortunately, they are too less concerned with what might happen to the share price on the long term. The agency problem especially increases when combined with the influences from the bankers, lawyers and other assorted advisers who can earn big fees from clients engaged in mergers.

cross-9 border acquisitions of assets located in 128 countries. The results show lower gains to buyers of cross-border versus domestic acquisitions. A large amount of acquirers‟ shareholder wealth was destroyed in a couple of days around the announcement of these M&As. Moeller et al. (2005) explain the former result through agency theory, as buyers involved in cross-border purchases are larger in size and more vulnerable to suffer from agency problems.

Specifically, Moeller et al. (2005) investigate why acquisition announcements in the latest merger wave are costly for acquiring-firm shareholders. The fact that the large loss deal firms create value through acquisitions in the two years before they make the large loss deal is hard to reconcile with the view that highly valued firms make bad acquisitions. Their evidence (Moeller et al., 2005) is consistent with the view of Jensen (2003) that high valuations give management more discretion, so that management can make poor acquisitions if it values growth more than shareholder wealth. With the growth of options as a form of managerial compensation in the 1990s, agency problems have become more severe. Since then managerial wealth is more closely tied to stock prices, presumably making management more conscious of the impact of acquisitions on the stock price and more likely to make acquisitions that increase shareholder wealth (Moeller et al., 2005).

As argued by Hoskisson et al. (2000), in Central and Eastern Europe opportunistic behavior is likely because of the prohibitively high costs of obtaining information for monitoring, difficulties in constructing legal contracts, and shifts in relative bargaining power due to exogenous shocks. Enterprises generally lack managerial skills and knowledge of market-based management. As a result, enterprises are likely to interpret the same objective environment differently, process information differently, and therefore make different strategic responses (Hoskisson et al., 2000).

2.2.3 Cultural & geographic differences

Country factors such as cultural and geographic distance have an influence on M&As as well. Erel et al. (2012) find that geography matters, because the gravity model in international trade shows that physical distance can increase the costs of merging two firms. Furthermore, culture matters as countries have their own cultural identities; the existence of different languages, religions, and values and norms increase the contracting costs associated with combining two firms across borders (Ahern, Daminelli, & Fracassi, 2012 cited by Erel et al., 2012). This finding is consistent with transaction cost theory recognizing that the greater cultural differences between bidder and target, the higher the transaction costs. Hence, the chances for success are hindered if the corporate cultures of the companies are very different, since this makes it more difficult to integrate both firms.

10 2.2.4 Country risks

This paragraph explores the literature about country risk and its impact on M&A performance. An understanding of country risk has become more important, given the increasingly global nature of M&A activity.

Kiymaz (2009) addresses the impact of country risk factors on the wealth gains to large US bidders involved in M&As. The conclusions show that all country risk factors including economic, political, and financial risk ratings play a significant role in explaining the wealth gains to bidders. Specifically, more stable financial markets (lower financial risk) and a more stable political environment (lower political risk) increase the wealth gains of the bidder. Furthermore, higher economic risk leads to higher wealth gains of the bidder, which might be caused by the target firm accepting a lower premium because of the unfavorable economic conditions. Next to this, fewer firms are willing to invest in a country with unfavorable economic conditions.

Girard and Omran (2007) use fundamental risk measures but also country risk scores to explain stock returns when investing in thinly traded emerging markets. They observe that country ratings predict inflation and are correlated with wealth. They also find evidence that higher risk is associated with higher expected returns. Although many of these emerging economies have put considerable effort in economic, financial and political reforms, issues related to financial transparency and political instability are still powerful obstacles to investments in these nascent emerging markets.

Erb et al. (1996) investigate how country risks explain the future expected returns of cross-border M&As. Country risks consists of political, financial and economic risk following the International Country Risk Guide (ICRG). Erb et al. (1996) conclude that some of the ICRG risk measures, economic and financial risk, can predict the cross-section of expected returns. However, political risk only has some marginal explanatory power in emerging equity markets.

The evidence regarding the influence of political risk is scarce in the literature. The paper previously discussed (Erb et al., 1996) finds a weak relationship between political risk and M&A performance in emerging markets, whereas the relationship in developed countries is not significant. However, political risk is expected to have a stronger relationship with M&A failure than the evidence currently suggests. The next paragraph describes political risk in more detail.

2.3 What is the definition of political risk and how can it be measured?

One of the definitions in the literature is from Howell (2004): “Political risk is the possibility that political decisions or political and social events in a country will affect the business climate in such a way that investors will lose money or not make as much as they expected”.

11 nongovernmental organizations. Examples include violence, revolutions, coup d‟états, civil wars, terrorism, ethnic or religious conflicts, etcetera. The authors argue that even in a country with relatively mature political institutions and public interest groups, the political environment is complex. Having a closer look at transition economies, where institutions are still developing, the spectrum of potential risks is even wider (Iankova & Katz, 2003). The transition process towards a market economy may create considerable risks and thereby a turbulent environment. Iankova and Katz (2003) argue that since CEE countries have opened to foreign investment, only a few have implemented political, legal, and regulatory systems that provide a comfortable and predictable business environment. This is in line with research of Ramamurti (2000), who argues that the move away from socialistic systems toward market oriented economic activity, has strong political implications. Governments pursue the transformation gradually with favorable feedback from the first round of decisions strengthening their commitment to further movement along the reform path as favorable outcomes reduce doubts and resistance from vested interests and other forces opposed to reforms (Ramamurti, 2000).

Following studies of Erb (1996), Girard and Omran (2007), Kiymaz (2009), and Erel et al. (2012) political risk is measured by indicators from the International Country Risk Guide (ICRG). The ICRG political risk rating assesses the political stability of countries and includes twelve weighted variables covering both political and social attributes. The following five risk components have been selected as appropriate for this research; government stability, socioeconomic conditions, investment profile, internal conflict and external conflict.

2.4 What is the relationship between political risk and M&A failure?

This thesis tries to connect institutional instability to M&A failure. Institutional theory (IT) can be integrated with transaction cost economics (TCE) and agency theory (AT) in order to explain this relationship. Institutional theory emphasizes the influences of the systems surrounding organizations that shape social and organizational behavior. Institutional forces affect organizations' processes and decision making (Hoskisson et al., 2000). Transaction cost economics studies the firm-environment interface through a contractual or exchange-based approach. Hoskisson et al. (2000) explain TCE in combination with AT, because of its focus on principal-agent problems under uncertainty. The authors suggest several insights for M&As in emerging economies when combining the theories.

12 suspicion of foreign firms have deterred inward foreign direct investment ((Hoskisson et al., 2000: 252). The primary impediment appears to be the lack of well-defined property rights that convey exclusivity, transferability, and quality of title (Devlin et al., 1998). Furthermore, formal rules may change overnight because of political and judicial decisions. As a result, many firms may defer investments where external shocks are frequent and cannot be foreseen. Or where entry would imply high-cost, firms may make irreversible investments. Firms may also defer entry if the creation of asset-specific investments suggests a high probability of opportunistic behavior by an emerging economy government (Hoskisson et al., 2000). Opportunistic behavior is likely because of the prohibitively high costs of obtaining information for monitoring, difficulties in constructing legal contracts, and shifts in relative bargaining power due to exogenous shocks. This is in line with Delios and Henisz (2000) who argue that the writing, monitoring and enforcement of contracts is difficult, especially in countries with high private expropriation hazards. Transaction costs are likely to be higher in emerging economies than in developed market economies, suggesting a preference for more hierarchical governance structures.

Wright et al. (2005) argue that institutional theory has become more enduring in emerging markets as the development of institutions has been slower than anticipated and the pace of political change has not been uniform across the transition economies.

When combining IT, TCE and AT we see that transition countries face many institutional instabilities, resulting in more complex agency relationships and higher transaction costs for M&As. Following the theories and considering the political change in transition countries, it is likely that political instabilities in CEE countries lead to M&A failure. The following paragraph discusses the empirical evidence on a global level, as this relationship has not been explored yet in CEE countries.

2.5 Is there any evidence for this relationship?

Prior research mainly focus on the performance of M&As on a global scale (Kiymaz, 2009, Erb et al., 1996, Girard & Omran, 2007 & Diamonte et al., 1996), these authors examine how country risk factors (political, financial, and economic risk) affect the success of M&As as described in this chapter. This paragraph particularly outlines the evidence about whether or not political risk impacts M&A failure.

Firstly, Kiymaz (2009) finds evidence for the relationship between political risk and M&A performance. They show that a 10% increase in political risk rating leads to a 5.5% increase in wealth gains for US bidders. Thus, wealth gains are higher in countries with stable political environments, less internal/external conflicts, less corruption, and higher democratic accountability.

13 equity markets but not in developed markets. The following paragraphs confirm that political risk has a significant impact in developing economies.

Girard and Omran (2007) find that risk factors, in particular political risk, prove significant in explaining the cross-sections of stock returns in thinly traded emerging markets. Their paper further argues that in particular political risk has strong implications for stock market development, especially in emerging markets where political risk hampers growth. Many scholars have suggested that the development of financial markets supports economic growth. Therefore, a great need exists to improve political risk in emerging countries in order to attract more investment and better allocation of resources through stock markets. According to Girard and Omran (2007) this can be achieved through more institutional (stock market) reforms. Hereby tackling the issues of improving the institutional and legal frameworks (accountability, transparency and disclosure, corruption, rule of law and contract enforceability).

Diamonte et al. (1996) study the impact of political risk on emerging and developing markets. Consistent with the paper of Girard and Omran (2007) they find that a change in political risk has a larger effect on the returns in a developing country than on the returns in a developed country. However, there has been a trend which shows that the political risk in developed countries and in developing countries is converging.

3. CONCEPTUAL MODEL & HYPOTHESES

The conceptual model (figure 1) shows the different concepts involved in this thesis, the dependent variable is the failure of M&As in CEE countries. The independent variable is political risk, which consists of five indicators: government stability, socioeconomic conditions, investment profile, internal and external conflict. These indicators are expected to significantly contribute to the explanation of the high failure rate of M&As. The research units are failed M&As which are compared to successful M&As in CEE countries in the univariate analysis.

14

Proposition: “A lower level of political stability leads to a higher failure rate of M&As in CEE countries’’

3.1 Government stability

The first political risk factor examined here is the government stability of the host country. A stable government is able to carry out its declared program and credibly commit to its policies and regulatory regime (North, 1990). Government stability consist of three subcomponents: government unity, legislative strength and popular support. Uhlenbruck and De Castro's study (2000) emphasizes that foreign firms entering transition economies through acquisition need to take government interference into account, since they find that continued government involvement is negatively related to post acquisition performance. We expect that if a government is more stable, there will be less unexpected changes in the country conditions. This will lead to less friction and thus a less risky environment for a firm to operate in. This led to the following hypothesis:

H1: A lower level of government stability leads to a higher failure rate of M&As in CEE countries

3.2 Socioeconomic conditions

Secondly, socioeconomic factors which are comprised in a society could constrain governmental actions or could cause social dissatisfaction (ICRG, 2012). Socioeconomic factors consist of three subcomponents; first, the level of unemployment within a certain country. Second, the level of consumer confidence is taken into account in rating the political risk within a country. Third, the level of poverty is being considered. According to the ICRG, socioeconomic factors contribute for

Failures of cross-border M&As in CEE countries

Financial risk Economic risk _{Political risk}

Government stability

Socio-economic conditions

Investment profile

Internal conflict

15 a significant extent to the level of political risk of a certain country, compared to other factors. Country factors, and especially unexplained country risk factors lead intuitively to a significant representation of political risk (Click, 2005). Country risk factors such as a high level of unemployment, a low level of consumer confidence, and a high level of poverty leads to social dissatisfaction which in turn could be a barrier for M&As. The bidder will divest in an acquired firm which is located in a country with a high level of uncertainty and a high level of social dissatisfaction. From this perspective no high profits seem to be feasible, which could result in the failure of M&As. Since socioeconomic activities shape the social and economic environment of a certain country, this factor is an important measure of the level of political risk of this specific country (Click, 2005). This leads to the following hypotheses: H2: Less favorable socioeconomic conditions lead to a higher failure rate of M&As in CEE countries

3.3 Investment profile

Thirdly, the investment profile involves several factors that reduce the gains of a firm such as expropriation, profit repatriation and payment delays. Sovereigns of countries in which the M&A is located can reduce the profits of the firms by expropriation of resources for their benefits. If a country has a high expropriation risk, a bidder might choose to invest in a project with a negative net present value. Firms have the incentive to reduce their income to avoid the attention of politicians. So the chance that a firm will be subject of expropriation by politicians will be reduced as well (Stulz, 2005). Because the state possesses a legal monopoly on coercion and is present in the background of every economic transaction, it poses a threat to the revenue streams of all private firms (Delios & Henisz, 2000: 307). This threat may take the form of regulatory or tax policy shifts or, at the extreme, outright expropriation of private sector assets. Delios and Henisz (2000) argue that MNEs face more exposure to these expropriation hazards than its host country competitors. The authors provide two reasons: their inferior knowledge of host country factor markets and their disadvantage to adapt in a manner that reduces the costs of a given expropriation. The latter refers to the higher opportunity costs the MNE faced of transforming the assets to their next best use. Their profitability may be affected by the information disparities described above and their lack of flexibility to adapt to changes in the political environment. This leads to the following hypothesis:

H3: A higher level of investment risk leads to a higher failure rate of M&As in CEE countries

3.4 Internal conflict

16 characterized by a country entangled in an on-going civil war. In the past decades independency pressures created civil unrest in the Western part of the Balkan. It even came to a civil war in Yugoslavia. In this region M&As face high political uncertainty, consequently it is hypothesized that: H4: A higher level of internal conflict leads to a higher failure rate of M&As in CEE countries

3.5 External conflict

The final indicator of political risk is the degree of external conflict a country faces, the three subcomponents; war, cross-border conflict, and foreign pressures, are used to assess this indicator. External conflict indicates the risk to the incumbent government from foreign action, which could take form of non-violent to violent external pressures (ICRG, 2012). Elements of non-violent external pressures are diplomatic pressures, withholding of aid, trade restrictions, territorial disputes and sanctions. Violent external pressure range from cross-border conflicts to all-out war. According to the ICRG (2012) external conflicts can adversely affect M&As in many ways; their businesses can be constrained through restrictions on operations, trade and investment sanctions, distortions in the allocation of economic resources, and/or violent change in the structure of society. This leads to the final hypothesis:

H5: A higher level of external conflict leads to a higher failure rate of M&As in CEE countries

4. RESEARCH METHODS

In this section the sample is introduced which is used for the empirical test of the proposition (overall political risk proxy) and the hypotheses (components) specified above. Next the operationalization of the dependent and independent variables is described. Thereafter the statistical model and estimation method are specified.

4.1 Sample

cross-17 border M&As of both public and private firms that were announced during the period January 1st, 1997 until the December 31st, 2008. The acquirer firms are from all over the world. The target firms are located in 18 CEE countries which are specified in appendix 1. Excluded are the CEE countries: Kosovo, Macedonia, and Bosnia and Herzegovina due to missing data.

CEE countries

The region Central and Eastern Europe is chosen for two reasons: they have a high level of FDI and it is expected that this region has a high level of political risk. First, Stephan & Jindra (2005) argue that FDI plays a particularly important role for relatively more backward regions. These countries are in transition from centrally-led government-owned economies to market-based nation states. CEE countries have experienced a strong FDI inflow due to the liberalization of trade policies, the mass privatization of state-owned companies and an increasing opening-up of markets resulting from EU integration (Stephan & Jindra, 2005). MNEs from all over the world enter CEE markets by M&As in order to enhance their competitive performance. The CEE countries are interesting for investors, because of their long history of industrialization and a relatively well educated work force. Besides, their location is in proximity to large EU markets. In CEE economies it is expected that political risk has a large impact on the failure rate, as these countries experience great changes in their political systems. Since the start of the transition to a market economy, the majority of these countries have not managed to implement political, legal, and regulatory systems that provide a comfortable and predictable business environment (Iankova & Katz, 2003). Hence, foreign investors often encounter political turbulence in CEE countries, which may have a direct impact on the performance of M&As.

4.2 Dependent variable

The dependent variable is the failure of M&As. The data is divided into two groups, namely the successful M&As and the failed M&As. The M&As are considered to be failed if the deal status is withdrawn. I acknowledge that this definition actually refers to deal incompletion, which is a rough proxy of M&A failure. This measure involves failures in the initial process of the deal formation (failure of negotiations, due diligence, etc.) and ignores post M&A failure. M&As are considered to be successful if they are completed. The variables collected include the acquirer‟s and target‟s name, the target and acquirer country codes, the announcement, completion and withdrawn dates. Also collected is the deal status (completed or withdrawn) and the industry in which the bidder operates.

4.3 Independent variables

18 have been selected as appropriate for this research; government stability, socioeconomic conditions, investment profile, internal conflict and external conflict. These five indicators are selected, as they account for 60% of the total political risk score in a country and they are expected to contribute most to the failure of M&As. Each indicator is operationalized individually in three subcomponents which are presented in table 1.

Table 1. The indicators and subcomponents of political risk (ICRG, 2012) Political risk indicators Subcomponents

1. Government stability  Government unity

 Legislative strength

 Popular support

2. Socioeconomic conditions  Unemployment

 Consumer confidence

 Poverty

3. Investment profile  Contract viability/expropriation

 Profits repatriation

 Payment delays

4. Internal conflict  Civil war/coup threat

 Terrorism/political violence

 Civil disorder

5. External conflict  War

 Cross-border conflict

 Foreign pressures

The composite scores, ranging from zero to 100, are broken into categories indicating a very high to very low risk (table 2). In general terms if the points awarded are less than 50% of the total, that component can be considered as very high risk. If the points are in the 50-60% range it is high risk, in the 60%-70% range moderate risk, in the 70-80% range low risk and in the 80-100% range very low risk. When analyzing the scores the political risk of the CEE target countries are taken, it is expected that the host country factors determine the success or failure of M&As.

Table 2. ICRG risk categories

Risk category Composite score range in %

Very high 0.0 - 49.9

High 50.0 - 59.9

Moderate 60.0 - 69.9

Low 70.0 - 79.9

19 In the analysis the composite political risk index and the five indicators are tested separately, as the composite index represents these indicators. The ICRG identifies twelve indicators that measure the composite political risk rating, but only five are used for this analysis. In order to solely test whether the composite score of these five indicators has an effect on the failure of M&As, a new risk index is created. The newly generated variable is calculated by using the sum of the political risk scores on the five indicators and referred to as the „self-designed political risk rating‟ or in the regression as the „sum political risk components‟.

4.4 Control variables

Control variables are included in order to test the validity of this study. Control variables are variables that are held constant in order to assess or clarify the relationship between the dependent and the independent variables. In order to exclude the influence of other risk factors on the failure of M&As several control variables are incorporated. The main control variables are subdivided in two classifications, namely: industry and country level.

First, the industry affiliation of bidders should be taken into account. Each industry has a different level of efficiency and expertise and encompasses different abilities to exploit opportunities which may help explaining the success or failure of M&As. To determine the impact of industry affiliation, a set of dummy variables are constructed representing various industry groups in the acquirer sample following Kiymaz (2009). Seven industries, based on the SIC classification, are used to categorize the M&As: mining and construction SIC1000-1799, manufacturing SIC2000-3999, transportation and communication SIC4000-4999, wholesale and retail SIC5000-5999, finance, insurance & real estate SIC6000-6799, services industry SIC 7000-8999, and public administration SIC 9100-9999. If the bidder is operating in the manufacturing industry, the manufacturing dummy equals 1. If the bidder is not operating in this industry then the manufacturing dummy equals 0. The other industry variables are constructed the same way.

20 the EU and if so, how the reforms developed since the accession. More information on EU membership per CEE country can be found in appendix 1.

In sum, table 3 provides an overview of the dependent, independent and control variables used in this study.

Table 3. Variables and measures

4.5 Statistical model

To test the main proposition the following model with one dependent variable, Y; one independent variable, PR; and control variables will be utilized:

Type of variable

Name of variable

Description

Dependent Deal type Dummy variable that equals 1 if the deal is withdrawn, and 0 if the deal is completed.

Independent Political risk rating

An index assessing the political stability of the countries covered by the International Country Risk Guide (ICRG) on a comparable basis. Risk points are assigned to the political risk components (government stability, socioeconomic conditions, investment profile, internal and external conflict), which together comprise 60% of the aggregate political risk rating. In every case the lower the risk point total, the higher the risk and vice versa (see table 1).

Independent Sum political risk components

A self-designed political risk index generated by the sum of the political risk components: government stability, socioeconomic conditions, investment profile, internal and external conflict.

Independent Government stability

Indicator of political risk, assessing government unity, legislative strength, and popular support.

Independent Socioeconomic conditions

Indicator of political risk, assessing unemployment, consumer confidence, and poverty.

Independent Investment profile Indicator of political risk, assessing contract viability/expropriation, profits repatriation, and payment delays.

Independent Internal conflict Indicator of political risk, assessing civil war/coup threat, terrorism/political violence, and civil disorder.

Independent External conflict Indicator of political risk, assessing war, cross-border conflict, and foreign pressures.

Control Acquirer industry A dummy variable that indicates the industry of the target following Kiymaz (2009). The industries are categorized by using the Standard Industrial Classification (SIC). The seven industries are: mining and construction SIC1000-1799, manufacturing SIC2000-3999, transportation and communication SIC4000-4999, wholesale and retail SIC5000-5999, finance, insurance & real estate SIC6000-6799, services industry SIC 7000-8999, public administration SIC 9100-9999. If the bidder is in the manufacturing industry the manufacturing dummy equals 1, if the bidder is not operating in this industry then the dummy equals 0. The same is done for the other industries.

21

Failure ( y = 0 ) = αi + PRi + + (1)

Where,

y is a dummy variable for the success or failure of an M&A, in which a failure of M&A equals 1 and the success equals 0,

αi is a constant,

β1 is the slope,

PR is political risk,

is a vector with the control variables, εi are the residuals

In order to test hypotheses 2-6, in which the five indicators of political risk are specified, formula (2) is utilized.

( ) (2)

Where,

GSi is the government stability,

SECi are the socioeconomic conditions,

IPi is the investment profile,

ICi is internal conflict,

ECi is external conflict,

Finally, a model is constructed in order to measure the probability (P) that an M&A will fail.

( ₎ (3)

4.6 Estimation method

22 The sample used in this study is unbalanced as the dependent variable, M&A failure, is a rare event. Only 2,6 % of all cases are withdrawn M&As. King and Zeng (1999) have introduced a method that corrects for this bias, called Rare Events Logit Regression (ReLogit). This method estimates the same model as standard Logit regression, but with an estimator that gives lower mean square error in the presence of rare events data for coefficients, probabilities, and other quantities of interest (King & Zeng, 1999).

For the two aggregate political risk ratings (ICRG‟s political risk index and the self-created political risk index, which is comprised of the sum of the five components) separate models are run. The five political risk components are also run in a separate model. However, due to violations of some assumptions which should be satisfied when running a ReLogit analysis, it is not possible to run the five indicators in one model. That is why the problematic components, internal and external conflict, are added apart from each other to the other three political risk components.

The assumptions of a ReLogit regression are discussed in detail in the following section. Even though some assumptions are not satisfied, I will show why ReLogit is a valid estimator in this study.

4.7 Evaluation of method assumptions

According to Wooldridge (2009) the logit model can be derived from an underlying latent variable model which satisfies the classical linear model assumptions. Additionally, he states that those assumptions should not be strictly met in the case of logit models. Despite this we will test whether or not the datasets meet the key assumptions which need to be satisfied in order to have unbiased estimates. In other words, the variables must not contain any problems regarding heteroscedasticity, endogeneity, multicollinearity and normality.

Homoscedasticity

Logit regression assumes that the variance of the error term is constant (homoscedastic) and the same for all observations (i) (var(ei) = σ²). If this assumption is violated and the error variance for

23 Prob > chi2 = 0.0000

chi2(1) = 1865.69

Variables: fitted values of Dummy_failed_MAs Ho: Constant variance

Breusch-Pagan / Cook-Weisberg test for heteroskedasticity . estat hettest

In the sample heteroscedasticity could be expected since successful M&As might have a smaller error variance in political risk scores than failed M&As, simply because there are much more observations for successful M&As than for failed M&As. Appendix 2a plots the residuals, which estimate the errors, against the composite political risk rating. From the scatter plot I conclude that heteroscedasticity is an issue, as the variance is dissimilar for successful and failed M&As. A formal Breusch-Pagan test (see table 4) confirms

this observation and shows significant results as well as an extremely large chi-square. To test whether the five political risk indicators show the same results, separate tests are performed for each

component. Four out of five components are significant, but one (internal conflict) shows an insignificant result when performing the BP test (appendix 2b) indicating a constant variance of the error term. As the majority of tests is significant, there can be concluded that heteroscedasticity is present. According to Wooldridge ( 2009) large samples with rare events will almost surely turn up positive when testing for heteroscedasticity.

Following King and Zeng (1999) an alternative test for heteroscedasticity is performed, the White's test, which gives more correct estimates in the case of rare events, because it tests for homoscedasticity against unrestricted forms of heteroscedasticity. The results (see table 5) are again significant and indicate a clear presence of heteroscedasticity. Note that the chi-square value is much smaller compared to the BP test. The remedy for heteroscedasticity is simple. White

(1980) developed an estimator for standard errors that is robust to the presence of heteroscedasticity. The estimation method used in this study (ReLogit) corrects for this problem by default, as it calculates robust variance estimates. Next, robust standard errors will be discussed.

Robust Standard Errors (RSE)

As outlined earlier, heteroscedasticity causes standard errors to be biased. Logit regression assumes that errors are both independent and identically distributed; RSE relax either or both of those assumptions. Hence, when heteroscedasticity is present, robust standard errors tend to be more Total 402.42 26 0.0000 Kurtosis 172.63 1 0.0000 Skewness 181.05 5 0.0000 Heteroskedasticity 48.74 20 0.0003 Source chi2 df p Cameron & Trivedi's decomposition of IM-test

Prob > chi2 = 0.0003 chi2(20) = 48.74

against Ha: unrestricted heteroskedasticity White's test for Ho: homoskedasticity

Table 4. Breusch-Pagan / Cook-Weisberg Test for heteroscedasticity

24 trustworthy. According to Wooldridge (2009) RSE are justified only as the sample size becomes large (in small sample sizes, the robust t statistics can have distributions that are not very close to the t distribution, and that could throw off the inference). Moreover, in large samples heteroscedasticity tests will almost surely turn up positive, so reporting only the heteroscedasticity-RSE has become a common practice for analyzing cross-sectional data.

To ensure the validity of the study, the RSE are compared to the normal standard errors when performing a logit regression. This comparison reveals that both errors and the corresponding p-values are quite similar. For the independent variables the errors are slightly smaller. Furthermore, the p-values indicate comparable results with and without RSE, except for one country dummy (Latvia). Latvia is significant at the 0.1 level using robust standard errors and insignificant using normal standard errors, however the p-values are not very different (SE: p = .112 and RSE: p = .084). Most important is that the independent variables show the same significance levels, so it does not bias the results. Therefore, I conclude that although heteroscedasticity is present, it is not a problem.

Endogeneity

If the assumption that the error term is uncorrelated with the independent variables is violated, such that xi is correlated with unmeasured variables (ei), we will consistently overestimate the effect of

xi on y, which indicates an endogeneity problem. Additionally simple numerical or statistical tests for

this assumption are not available therefore it has to be satisfied theoretically. Two-stage least-squares (2SLS) regression with instrumental variables is one option to test for and overcome the endogeneity problem and generalized least squares (GLS) is another. In this thesis five indicators represent the composite political risk rating. In addition, the PRS Group uses another seven variables (c

orruption,

military in politics, religious tensions, law and order, ethnic tensions,

and bureaucracy quality) to produce this political risk rating. It could be that the error term is correlated with these seven unmeasured variables. Cronbach‟s alpha is used to test whether the five items measure the underlying construct (political risk). As can be seen in appendix 5 the alpha coefficient for the five items is 0.78, suggesting that the items have an acceptable level internal consistency.

Multicollinearity

Another key assumption is that the independent variables are not perfectly correlated such that “the values of xik are not exact linear functions of other explanatory variables” (Hill et al., 2009: 154).

25 parameters, an issue which might even become more problematic when there is little variation in the explanatory variables. To test for the presence of multicollinearity the variance inflation factor (VIF) is calculated. This index estimates how much the variance of an estimated regression coefficient is inflated due to collinearity. The first test for multicollinearity on the original data indicates a high value (9.84) for one of the political risk components, internal conflict. Having a closer look at the correlation values (appendix 4), indicates a high positive correlation between the two variables internal and external conflict. Besides imprecise parameter estimates, not correcting for high multicollinearity among the political risk indicators results in insignificant model specification. To solve this problem internal and external conflict are run separately in the regression. Table 6a provides the new vif-values of the dummy control variables and the independent variables, excluding external conflict. In table 6b internal conflict is excluded. Both tests indicate that multicollinearity is not problematic any longer as all variables have values well below the cut-off value of 10 recommended by Neter et al. (1985).

Table 6a. Tests for Multicollinearity (excluding external conflict)

Mean VIF 2.88 PublicAdmi~n 1.09 0.920511 Target_AL 1.11 0.898612 Target_ME 1.15 0.867520 Target_BY 1.25 0.798437 Target_MD 1.30 0.768698 WholesaleR~r 1.32 0.755073 Transporta~u 1.42 0.704729 Services_A~y 1.47 0.678511 Target_SI 1.64 0.609003 FinanceIns~i 1.73 0.578965 Manufactur~y 1.90 0.526298 Target_HR 2.17 0.461119 Target_RS 2.26 0.441837 Target_LV 2.43 0.412216 Target_SK 2.55 0.392641 Target_EE 2.76 0.361790 Target_LT 2.96 0.337347 Target_HU 3.48 0.287007 Government~y 3.66 0.273430 Target_BG 3.69 0.271310 Target_UA 3.73 0.268091 Target_RO 3.97 0.251975 Target_CZ 4.91 0.203607 SocioEcono~s 5.25 0.190609 Internalco~t 5.70 0.175478 Target_PL 6.11 0.163655 Investment~e 6.82 0.146735 Variable VIF 1/VIF

Mean VIF 2.48 Target_AL 1.08 0.928298 PublicAdmi~n 1.09 0.917193 Target_BY 1.10 0.911746 Target_ME 1.12 0.893486 Target_MD 1.21 0.824017 WholesaleR~r 1.32 0.755189 Transporta~u 1.42 0.704810 Target_SI 1.44 0.693435 Services_A~y 1.47 0.678745 Target_HR 1.55 0.646117 FinanceIns~i 1.73 0.579047 Manufactur~y 1.90 0.526409 Target_RS 1.92 0.519823 Target_EE 1.93 0.519098 Target_LV 1.96 0.509738 Target_LT 2.23 0.448712 Target_SK 2.24 0.445513 Target_HU 2.71 0.369290 Target_UA 2.76 0.362569 Target_BG 3.01 0.332692 Government~y 3.40 0.293712 Target_RO 3.51 0.285170 Externalco~t 3.63 0.275465 Target_CZ 4.28 0.233702 SocioEcono~s 5.09 0.196278 Target_PL 5.56 0.179896 Investment~e 6.21 0.161089 Variable VIF 1/VIF

26 Normality

The final test which will be carried out is the test for normality. It assumes that the values of the error term are normally distributed about their mean. A violation of the normality assumption might lead to biased p-values and thus affects the test for statistical significance (Hill et al., 2009). A formal statistical test for normality rejects the null hypothesis of normal distribution, which might nevertheless be due to the large sample size. One limitation of the normality tests is that the larger the sample size, the more likely to get significant results. In this thesis, the sample size is large (n=5485) so the significance of the formal tests may only indicate slight deviations from normality. Accordingly, a graphical analysis may be a more appropriate way to determine the level of normality. Appendix 6a-f plots the expected values against the observed values for the political risk rating and its five components. If the dots (actual data) fall exactly on the black line, the distribution is normal, which is the case for the composite political risk rating and one of its components government stability. However, the other four components slightly deviate from the straight black line. Wooldridge (2009) states that the Logit model could still provide correct estimates even when errors are not normal nor independent and identically distributed. As the sample size is reasonably large for each component, it is assumed that the data is approximately normally distributed for each component.

After running these tests it can be concluded that the model does not contain any problems with endogeneity. Furthermore, the ReLogit model uses an estimator for standard errors that is robust to the presence of heteroscedasticity. The multicollinearity assumption will be resolved by running the components internal and external conflict separately in the ReLogit regression. Even though the dichotomous form of the dependent variable and the large sample size, a potential problem might be the normality assumption of four political risk components. But according to Wooldridge (2000) the Logit model is still consistent and can be employed even though it has a non-normal distribution. Hence in the next section we move forward and run the ReLogit regressions.

5. EMPIRICAL RESULTS

5.1 Descriptive statistics

27 failed M&As. The majority of all M&As (30-36 %) occur in the manufacturing industry, followed by the finance, insurance and real estate sector (21 %). Failed M&As occur more in the manufacturing, and the mining and construction industry. Successful M&As occur more in the services, wholesale and retail, and transportation and communication industry.

Table 7 below provides an overview of the most relevant descriptive statistics for failed and successful M&As. For both failed and successful M&As the composite political risk score is approximately 71. Following ICRG‟s rating system, the CEE countries on average have a low level of political risk (table 2). Compared to the other components, socioeconomic conditions show the highest level of risk in the CEE countries. Socioeconomic conditions have an average score of 5 (a score of 12 is the maximum score for each component), in general terms if the points awarded are less than 50% of the total, that component can be considered as very high risk. CEE countries possess low risk levels for government stability and investment profile, and very low risk levels for internal and external conflict. Surprising is the result that CEE countries where M&As fail, on average have slightly lower levels of political risk than CEE countries where M&As are successful. The sum of the political risk components indicates the opposite result, but the mean scores are very close to each other. For the CEE countries where M&As fail socioeconomic conditions, investment profile and internal conflict indicate slightly higher levels of risk, whereas government stability and external conflicts indicate slightly lower levels of risk.

Table 7. Descriptive statistics of successful and failed M&As

Successful M&As Failed M&As

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

Political risk rating 71.17 12.15 0 89 71.81 11.01 0 82.64

Sum political risk components 43.88 6.74 0 52 43.60 6.35 0 51.42

Government stability 8.39 2.03 0 11.5 8.55 1.82 0 11.5 Socioeconomic conditions 5.74 1.60 0 8.5 5.40 1.70 0 7.5 Investment profile 9.51 2.05 0 12 9.37 1.90 0 12 Internal conflict 10.06 1.79 0 12 10.03 1.63 0 12 External conflict 10.18 1.69 0 12 10.26 1.58 0 12 5.2 Regression results

28 Hypotheses 1-5 zoom in on the selected political risk components. Model 3estimates whether these components impact the failure rate of M&As in CEE countries. Two regressions are run for the five independent variables, so that the components internal and external conflict do not appear simultaneously in one model (see the multicollinearity section 4.7). Model 3a adds four political risk components, external conflict is excluded. Government stability shows a positive and highly significant relation to M&A failure (β=0.245; p<0.001). A one-unit increase in government stability, increases the probability of an M&A failure by 0.25. This is contradictory to the predictions1, as it indicates that a more stable government increases the chance of an M&A failing, hence hypothesis 1 is rejected. Next, internal conflict shows a negative and significant relation to M&A failure (β=-0.290; p<0.01). A one-unit increase in internal conflict, decreases the probability of an M&A failure by 0.29. This confirms the fourth hypothesis, as it indicates that a lower level of internal conflict decreases the chance of an M&A failing. Socioeconomic conditions and investment profile are both insignificant, hence hypothesis 2 and 3 are rejected.

Model 3b adds the component external conflict instead of internal conflict. External conflict shows insignificant results, socioeconomic conditions stays insignificant as well. Government stability shows again a positive significant relation to M&A failure, however the significance level changes from high to modest (β=0.129; p<0.1). A one-unit increase in government stability, increases the probability of an M&A failure by 0.13. An interesting result is the negative and significance relationship between investment profile and M&A failure (β=-0.163; p<0. 1). A one-unit increase in investment profile, decreases the probability of an M&A failure by 0.16. This confirms the third hypothesis, indicating that a lower level of investment risk decreases the chance of an M&A failing.

The following results are derived for the control variables. First, the industry classification dummy control variable is discussed. The mining and construction industry is chosen as a control group and the findings for the remaining industry groups are interpreted accordingly. M&As by firms in all industries, except for public administration, decrease the probability of an M&A failure relative to the M&As by firms in the mining and construction industry. The transportation & communication, wholesale & retail, and services industry are statistically significant, indicating that the deal is less likely to fail in these industries compared to the mining and construction industry. Second, we turn to the target country dummy control variable, Russia is chosen as the reference country. All target countries increase the probability of an M&A failure. If an M&A occurs in the target countries Albania, Bulgaria, Czech Republic, Croatia, Hungary, Lithuania, Montenegro, Poland, Romania, Serbia, Slovenia, Slovakia or Ukraine, the deal is more likely to fail relative to M&As that occur in Russia. Overall, it can be said that the results do not confirm the main proposition, H2 and H5. The effects of the composite political risk ratings, as well as the components socioeconomic conditions

1

29 Table 8. Rare Events Logit regression results of the relationship between political risk and M&A failure in Central & Eastern Europe

Model 1 Model 2a Model 2b Model 3a Model 3b Controls Political risk rating Political risk rating Political risk components Political risk components

Control-acquirer industry

Manufacturing -0.239 (0.221) -0.241 (0.221) -0.241 (0.221) -0.223 (0.222) -0.227 (0.221) Transportation & communication -0.771* (0.364) -0.767* (0.364) -0.759* (0.365) -0.682+ (0.365) -0.697+ (0.365) Wholesale & retail -1.107* (0.492) -1.103* (0.492) -1.093* (0.492) -1.031* (0.493) -1.036* (0.493) Finance, insurance, real estate -0.337 (0.257) -0.336 (0.256) -0.329 (0.256) -0.266 (0.259) -0.280 (0.260) Services -1.019* (0.402) -1.018* (0.401) -1.011* (0.402) -0.948* (0.405) -0.953* (0.405) Public administration 0.254 (0.619) 0.257 (0.619) 0.264 (0.619) 0.262 (0.629) 0.299 (0.635) Control-target country Albania 1.788+ (1.063) 1.800+ (1.063) 1.752 (1.066) 3.127** (1.155) 2.100+ (1.102) Bulgaria 1.363*** _{(0.361) 1.415}*** _{(0.373) 1.382}*** _{(0.362) 3.171}*** _{(0.563) 2.260}*** _(0.535) Belarus 0.749 (1.045) 0.728 (1.046) 0.690 (1.043) 1.494 (1.158) 0.496 (1.033) Czech Republic 0.650+ (0.359) 0.751+ (0.397) 0.711+ (0.364) 2.301*** (0.509) 1.397** (0.521) Estonia -0.267 (0.563) -0.216 (0.562) -0.268 (0.574) 1.061 (0.653) 0.186 (0.610) Croatia 1.353** (0.415) 1.410** (0.431) 1.309** (0.411) 2.841*** (0.589) 1.781*** (0.477) Hungary -0.289 (0.522) -0.176 (0.555) -0.222 (0.527) 1.307* (0.640) 0.450 (0.642) Lithuania -0.169 (0.567) -0.127 (0.565) -0.189 (0.576) 1.286+ (0.722) 0.453 (0.672) Latvia -0.536 (0.753) -0.497 (0.753) -0.549 (0.755) 0.851 (0.830) 0.0247 (0.846) Moldova 0.470 (1.034) 0.480 (1.034) 0.389 (1.039) 1.262 (1.109) 0.475 (1.077) Montenegro 1.778+ (1.045) 1.759+ (1.046) 1.650 (1.055) 2.967* (1.253) 2.271* (1.094) Poland 0.352 (0.351) 0.454 (0.393) 0.387 (0.354) 1.963*** (0.531) 1.228* (0.517) Romania 0.422 (0.401) 0.459 (0.408) 0.413 (0.401) 1.921*** (0.563) 1.010+ (0.564) Serbia 0.953+ (0.527) 0.938+ (0.526) 0.841 (0.529) 2.608*** (0.670) 1.545* (0.662) Slovenia 2.046*** _{(0.371) 2.142}*** _{(0.393) 2.132}*** _{(0.375) 3.296}*** _{(0.482) 2.560}*** _(0.440) Slovakia 0.245 (0.522) 0.331 (0.541) 0.318 (0.528) 1.880** (0.654) 1.006 (0.624) Ukraine 0.224 (0.465) 0.236 (0.466) 0.177 (0.466) 1.620* _{(0.646) 0.515 (0.598)} Independent variables

Political risk rating -0.00561 (0.00824)

Sum political risk components -0.0154 (0.0107)