Impact of Corruption on Cross-Border Mergers and Acquisitions
MSc International Financial Management
Lixue Zhang
Student Number: S2713608
Supervisor: dr. R.O.S. Zaal
Co-Assessor:
dr. J.O. Mierau
Abstract
This thesis empirically examines the relationship between corruption and cross-border mergers and acquisitions (M&As). It employs a sample of 14,557 cross-border M&As, involving 157 countries over the period of 1997 until 2015. I examined the impact of host country corruption on cross-border M&A activities in terms of volume and value. The absolute difference in corruption level between home and host country is also analyzed to uncover its effect on cross-border M&As. The results indicate that host country corruption has a negative impact on the number of cross-border M&A deals that it attracts. Moreover, the absolute difference in corruption level between home and host country is found to be negatively related to both the volume and the value of cross-border deals. The findings confirm the institutional theory, and add to the psychic (administrative) distance perspective in explaining the impact of corruption on cross-border M&As. Findings also suggest that firms should consider corruption in site selection if they initiate M&As abroad.
1 1. Introduction
Prior studies have found that corruption has critical influence on foreign direct investments (FDI). Corruption is associated with less FDI inflows (Habib & Zurawicki, 2001 & 2002). Corruption may result in a change of the composition regarding FDI’s country of origin (Cuervo-Cazurra, 2006). Brouthers, Gao & McNicol (2008) propose that costs induced by corruption are different for marketing-seeking FDI and resource-seeking FDI, thus, the influence of corruption is also expected to be different for different types of FDI. Despite various studies on corruption and FDI, few researches were conducted to investigate the influence of corruption on cross-border mergers and acquisitions (M&As). It is somewhat surprising to me, since cross-border M&As have been increasing their share in FDI (UNCTAD, 2006). Brakman, Garita, Garretsen, & van Marrewijk (2008) also report in their work that cross-border M&As constitute 78% of total value of FDI in 2000. Given the fact that cross-border M&A is the primary mode of FDI, exploring the impact of corruption on cross-border M&As is thereby of great importance.
2
countries tend to make outbound investments to countries where the corruption levels are lower than their home country. In addition to host country corruption, few studies have investigated the difference in corruption between home and host country and its effect on cross-border M&As. There are findings suggesting that a great number of cross-border M&As target countries with corruption levels that are similar to home country (Dikova, Panibratov, Veselova, & Ermolaeva, 2016). I was motivated by the various findings in terms of the impact of corruption, and intending to answer the following research question through this thesis: How are cross-border M&A activities influenced by the level of corruption in the home and the host countries?
This thesis extends the work by Malhotra, Zhu & Locander (2010), and investigates the relationship between corruption and cross-border M&As with a sample of 14,557 cross-border M&A deals around the globe from 1997 until 2015. This thesis addresses the host country corruption and the absolute difference in corruption level between host and home country, and it attempts to find their impacts on cross-border M&A activities in terms of volume and value. The results suggest that the number of cross-border M&A deals received by host country is negatively related to host country corruption level. Moreover, the absolute difference in corruption level between home and host country is also negatively associated to the volume and value of cross-border M&A deals in host country.
The remainder of this thesis is structured as follows. Next section will discuss corruption and relevant theories that are applied to explore the relationship between corruption and cross-border M&A activities; prior studies will be reviewed, and hypotheses will be developed accordingly. Section 3 will introduce data and methodology employed in this thesis. Empirical findings will be presented in Section 4, and the concluding remarks will be made, and limitations of this thesis will be discussed in the last section.
2. Literature review and hypotheses development
3
that ‘the control of assets and operation is transferred from a local to a foreign company, the former becoming an affiliate of the latter’ (UNCTAD, 2000). According to Brakman, et al. (2008), cross-border M&As constitute 78% of total value of FDI in 2000.
Studies have investigated the effect of corruption on FDI volume (Habib & Zurawicki, 2001 & 2002), on the change in the composition regarding FDI’s country of origin (Cuervo-Cazurra, 2006), and on the choice of different types of FDI (Brouthers, et al., 2008). Given the fact that border M&A is the primary form of FDI, investigating the influence of corruption on cross-border M&As becomes the major concerns for this thesis. This section consists of two fields of literature. Firstly, literature with regards to corruption will be reviewed; secondly, theories behind the effect of corruption on cross-border M&As will be discussed, and hypotheses will be developed accordingly.
2.1 Corruption
Corruption has been defined in various ways with different emphases. This thesis refers to corruption as the abuse of public power for private benefit, since it covers government corruption and firm corruption. As discussed by Cuervo-Cazurra (2016), this definition addresses three characteristics of corruption: first, a person abuses power that is entrusted to him or her by others; second, this entrusted person abuses power to engage in activities that are beyond his or her mandate; and third, the entrusted person gains benefits that are only acquired by him or her rather than by the organization that he or she serves.
4
collectivism, and humane orientation) behind bribery activities, and they also discover that such forces would be mitigated if social institutions are better designed.
By its nature, corruption incorporates unethical behaviors, such as bribery, extortion, cronyism, fraud, side payments, misuse of information, and abuse of discretion (Xie et al., 2017). These behaviors bring significant consequences for firm and for country. Cuervo-Cazurra (2006) describes two opposing views on the impact of corruption on firms. The negative view suggests that corruption as ‘sand in the wheels of commerce’ leads to the wasteful use of resources and inefficient allocation of resources. This view addresses the increasing costs and uncertainties that are associated with corruption. Costs increase as firms bribe and as companies devote time and resources to manage the relationship with government officials. However, there is no guarantee for the expected result as promised by government officials, which will result in further risks and uncertainties. On contrary, the positive view holds that corruption ‘greases the wheels of commerce’. Corruption is often complimented for its role in facilitating transactions and accelerating procedures, which would be more difficult and time-consuming for firms to implement if they fully comply what is required in the book. In this sense, corruption would improve firm’s efficiency by smoothing originally rigid procedures. However, this thesis takes the position of corruption being ‘sand rather than ‘grease’, since paying corrupt transactions does not necessarily result in more efficient operations for firms. It is rather a form of rent-seeking activities (Tanzi, 1998).
5
(1998) contends that corruption distorts markets and allocation of resources by impairing government’s role in the regulatory control and by reducing the legitimacy of market economy. Given these circumstances, foreign investments would not favor corrupt countries as the destinations. Several empirical studies have supported this perspective, which find that countries that are more corrupt would be less appealing for cross-border M&A activities (Malhotra, et al., 2010), FDI inflows (Wei, 2000; Habib & Zurawicki, 2001), and FDI inflows from countries that are against corruption, which are the major foreign investors (Cuervo-Cazurra, 2008).
This thesis views corruption as a national characteristic embedded in a country, which will be used as a variable directly explaining worldwide cross-border M&A activities. The underlying theories of the impact of corruption on cross-border M&As will be reviewed in the next section.
2.2 Impact of corruption on cross-border M&As 2.2.1 Host country corruption and cross-border M&As
6
In addition, Habib & Zurawicki (2002) point out that a corrupt host country tends to offer less open and equal market access to all investors, which consequently has a negative effect on FDI inflows. Weitzel & Berns (2006) adopt the notion of lock-in effect of corruption. They posit that within a corrupt country, partners tend to be tied to each other due to the threat of mutual denunciation. This situation thus limits the exchanges between foreign bidding firms as the outsiders to insiders. In this sense, the access to a corrupt target country would be the primary concern for foreign bidding firms. When entering a corrupt country, foreign bidding firms may face adverse selection challenges of choosing the ‘right’ government officials who accept bribery and be able to fulfill the deal that these foreign firms desire (Cuervo-Cazurra, 2016).
Moreover, Glambosky et al. (2015) argue that corruption in host country creates barriers to information flow, which may result in unfair treatment of the bidding firms. When target firm resides in a corrupt country, bidding firm may perceive the target as vulnerable to acquisition below market value. It, therefore, creates more competitions for the target firm, which could lead to a decline in the likelihood of successful bid (Glambosky et al., 2015). Thus, when the corruption level of host country is high, the probability of successful cross-border takeovers would drop, thereby a fewer number of cross-border takeovers would be. This is empirically supported by Malhotra et al. (2010) that less corrupt countries attract a greater number of cross-border M&As from both China and the United States. From the above arguments, the first hypothesis can be drawn as follows:
Hypothesis 1a: The number of cross-border M&As received by host country is negatively related to host country corruption level.
7
corruption. Malhotra et al. (2010) further argue that initiating large cross-border takeovers usually means taking more risks, and bidding firms would favor locations where these risks can be minimized. As discussed earlier, a corrupt host country is usually associated with a lack of credibility and increasing uncertainties (Hur, Parinduri& Riyanto, 2011). Hence, it is expected that the cross-border M&A deals that a corrupt country receives tend to have lower transaction values. The second hypothesis can be drawn as follows:
Hypothesis 2a: The transaction value of cross-border M&As that host country receives is negatively related to host country corruption level.
2.2.2 Difference between home and host country corruption and cross-border M&As
Despite the effect of host country institutions, the type of institutional environment in home country also can be considered as one of the causes of cross-border M&As. Martynova & Renneboog (2008) discuss that when bidding firm is struggling in poor investment environment (i.e. high level of corruption) in its home country, it would initiate M&A abroad instead of domestically. It can be inferred that when a bidding firm originates from a corrupt country, it tends to target a country with a relatively lower level of corruption to extricate from the poor home environment. Whereas, there are studies that otherwise posit that cross-border M&As are more likely to occur if the difference in corruption level between home and host country is minor (Dikova et al., 2016; Malhotra et al., 2010).
8
bribe request from government official, which is a prevalent custom in host country (normative pressure). However, this situation would be a rather usual practice for bidding firms from a similarly corrupt country, even if the bribe request goes against regulatory pressures. Therefore, it can be inferred that bidding firms are less likely to merge or acquire target firms in countries that differ great from their home countries in terms of corruption level (Dikova, Sahib & Witteloostuijn, 2006; Malhotra et al., 2010).
In addition, from the psychic distance perspective, the underlying assumption is that the lack of knowledge of the foreign market is the major disturbance for firms to internationalize. Firms would target countries that are considered as psychologically close to its home country before entering remote ones, since selection of psychically close market would be less risky (Johanson & Vahlne, 1977). As suggested by Malhotra et al. (2010), the difference in corruption level between home and host country constitutes a distance similar to psychic distance. They argue that firms would prefer a country with the similar level of corruption as their home countries as the destination of cross-border M&A, because the similarity in terms of corruption would result in reduced risks associated with the internationalization process (Malhotra et al., 2010). Habib & Zurawicki (2002) believe that difference in corruption level between home and host country can be considered as a reflection of administrative distance. Administrative distance, together with cultural, geographic and economic distances influence companies in evaluating global expansion opportunities, as these four distance dimensions create barriers to foreign investors (Ghemawat, 2001). Furthermore, psychic closeness also facilitates learning about host country (Kogut and Singh, 1988). Bidding firm would target countries with similar level of corruption, because it has prior experience and knowledge that can prepare itself to handle the corrupt (or less corrupt) investment environment in the host country (Habib & Zurawicki, 2002). Exposure to corruption in home country equips the bidder with ability and knowledge of dealing corruption, which makes the cross-border takeovers more likely in a similar environment. However, when it targets a less corrupt country, the advantage that bidder has knowledge of handling corruption would be considered as redundant. Thus, it can be posited that host country would attract more cross-border M&A deals that originate from countries with similar corruption level. The hypothesis can be thereby formulated as:
9
Furthermore, larger deals are usually more time-consuming, demanding extensive resources in order to overcome the differences in institutional environments between home and host country (Malhotra et al., 2010). This suggests that firms undertaking high-value cross-border M&A deals usually have more at stake. However, uncertainties that these firms encounter would be reduced as the difference in institutional profiles between home and host country gets smaller, since firms’ prior experience and knowledge enable them to be better adjusted to the institutional environment in the host country. Due to the decline in uncertainties, bidding firms would be willing to initiate cross-border M&A deals with larger transaction value. Hence, it can be argued that transaction value of cross-border M&As would become larger, when the absolute difference in the level of corruption in home and host country gets smaller. The last hypothesis can be posited as:
Hypothesis 2b: The transaction value of cross-border M&As that host country receives is negatively related to the absolute difference of corruption between home and host country.
3. Empirical research design
This research will employ multiple regression analysis to examine the impact of corruption on cross-border M&A activities in terms of amount and value. In this section, sample selection will be outlined first, variables used in the model will be properly defined, data collection process will be illustrated, and descriptive statistics of data will be presented. Next, due to the attributes of the data, estimation methods of the models will be highlighted, and the multivariate regression models will be formulated.
3.1 Sample and data 3.1.1 Sample
10
issues. To further limit the sample size, M&A deals that have transaction value below 1,000 thousand US dollars are omitted. With these criteria, the sample ends up with 14,557 cross-border M&A deals with a total transaction value of $ 3,137.53 billion involving 157 countries during the sample period of 19 years.
3.1.2 Variables and data
Dependent variable
Two dependent variables are relevant in this thesis, which describe cross-border M&A activities between home country j and host country i. For each cross-border deal, there is one county pair ji involved. The first dependent variable is the number of cross-border M&A deals (NUMBER'(,*), which indicates the aggregate of the number of M&A deals between a particular country pair ji in year t. The second dependent variable is the annual average transaction value of cross-border deals (VALUE'(,*), which is measured by the aggregate transaction value divided by the aggregate number of M&A deals between a country pair ji in the year t during 1997 to 2015. It is then transformed by taking natural logarithm.
Independent variable
11
|CORR'1(,*| is calculated as the absolute difference of rescaled CPI index between home and host country.
Control variables
Several control variables are included in the models to reflect other determining factors of cross-border M&A activities. First, variables reflecting host country characteristics are taken into account. Wealth of the host country, measured by log-transformed GDP per capital (GDP(,*) is associated with cross-border investment inflows. Dunning (1981) suggests that a country’s international investment position is associated to its wealth. It is empirically supported by Loree & Guisinger (1995) that GDP is one of the major elements attracting investments from abroad. Thus, it is expected in this thesis that the wealth of a host country is positively related to the cross-border M&A activities that it attracts.
A similar relationship is expected between cross-border investment inflows and the economic growth of host country measured by GDP annual growth (∆GDP(,*). Focarelli & Pozzolo (2001) suggest that the development of economic conditions of the host country is one of the driving forces of locating cross-border investments.
Moreover, the size of the host country, measured by log-transformed population (POP(,*), also plays a role. Dunning (1981) indicates that firms are inclined to invest in wealthier countries with larger populations. It is expected in this thesis a positive relation between the size of host country and the cross-border investments that it receives.
Rossi & Volpin (2004) find that the quality of accounting disclosure in a country has an effect on cross-border M&As, such that countries with higher ratings on accounting standards are associated with greater volume of mergers and acquisitions. However, empirical findings are inconsistent over studies. Erel, Liao, & Weisbach (2012) contend that bidders from countries with better governance quality (i.e. better quality of accounting disclosure) tend to target firms in countries with poorer governance quality. In this thesis, accounting standards (ACD(,*) is measured by the index of the quality of accounting disclosure assembled by La Porta, Lopez-de-Silanes, Shleifer & Vishny (1998).
12
import value as a percentage of GDP in host country as its international trade flows (TRA(,*), and expects a positive impact of this variable on cross-border M&As.
Secondly, cross-national differences between acquiring and target country are controlled. Cultural and geographical distance between home and host country tends to hold back international mergers, contended by Buch & Delong (2004). In particular, similar cultures between home and host country make cross-border takeovers more likely (Guiso, Sapienza & Zingales, 2006). In this thesis, cultural distance (CUL(,',*) is measured by integrating Hofstede’s four common cultural dimensions, including power distance, uncertainty avoidance, individualism, and masculinity. It is calculated by taking the natural logarithm of Zhou, Xie & Wang (2016) formula that is expressed as >;?@(S',; − S(,;)= 4.
Geographical distance is controlled for its impeding effect on cross-border deals, as greater distance hinders the transmission of soft information (Uysal, Kedia & Panchapagesan, 2008) and increases transportation costs. Geographic distance (DIS(,',*) is measured by log-transformed kilometers between the capital cities of host and home country.
13
deflated by consumer price index (2010 = 100) to calculate the real stock market return in each country.
Lastly, YEAR is controlled as the dummy variable, expressing the year of cross-border M&A occurrence. Appendix 1 lists descriptions, measurements and sources of each variable that are used in this thesis.
Data collection
Data is exported from the database Zephyr to construct two dependent variables. For each cross-border M&A deal in the sample, the announcement date and completion date, bidder’s and target’s name and country of domicile are acquired. The transaction value for each deal is collected in thousand US dollars. Bidders are from 109 different countries, and target firms are from 152 different countries. In total, this gives 1,908 ordered country pairs and 36,252 country pair-year observations.
14
3.1.3 Descriptive statistics
Table 1 gives an overview of the distribution of 14,557 cross-border M&A deals and total transaction value (in million US dollars) among host countries in the sample. United States,
Target Country Volume Value(in mil.US$) Target Country Volume Value(in mil.US$) Target Country Volume Value(in mil.US$)
United States 2,635 726,765.54 Lithuania 27 1,351.75 Sierra Leone 5 745.32
United Kingdom 1,902 340,290.10 Croatia 26 1,895.16 Barbados 4 98.39
Germany 847 359,416.78 Slovakia 26 1,309.45 Bangladesh 4 127.35
France 692 126,102.48 Uzbekistan 26 694.15 Cambodia 4 356.89
Canada 615 82,958.90 Estonia 24 697.60 Kuwait 4 595.27
China 560 42,849.09 Kazakhstan 24 7,269.74 Madagascar 4 44.83
Netherlands 513 167,110.77 Vietnam 23 1,288.75 Mecedonia 4 12.23
Australia 447 63,215.35 Egypt 22 2,333.02 Oman 4 470.94
Italy 379 105,953.18 Panama 22 5,351.96 Pakistan 4 547.35
Spain 351 89,491.87 Slovenia 20 1,034.09 Qatar 4 645.79
Sweden 336 112,273.29 Nigeria 18 2,697.48 Uganda 4 67.36
Brazil 277 53,665.92 Uruguay 18 1,738.36 Albania 3 168.03
Russia 251 52,945.28 Bosnia and Herzegovina 17 1,271.05 Bahrain 3 2,214.00
India 232 60,477.04 Greece 17 2,294.86 Cameroon 3 115.49
Switzerland 227 138,089.30 Venezuela 17 1,261.42 Gabon 3 368.57
Belgium 222 28,194.93 Ecuador 16 1,102.43 Guinea 3 114.10
Ireland 218 25,798.67 Malta 15 2,382.24 Kyrgyzstan 3 315.00
Norway 184 21,170.35 Morocco 14 3,060.44 Mongolia 3 56.66
Denmark 172 29,965.46 Georgia 14 388.51 Seychelles 3 38.62
Mexico 169 37,208.82 Kenya 14 5,216.11 Azerbaijan 2 50.87
Hong Kong 160 9,920.96 Mauritius 14 1,110.92 Burkina Faso 2 5.36
Singapore 148 24,656.63 Costa Rica 13 1,007.65 Benin 2 1,783.08
Czech Republic 137 13,172.25 Buelarus 11 3,774.25 Botswana 2 84.20
Argentina 135 28,867.40 Guatemala 10 547.90 Guyana 2 8.89
Israel 133 14,707.95 Jordan 10 1,451.36 Liberia 2 205.00
Poland 129 11,782.95 Namibia 10 391.65 Mali 2 331.63
Finland 126 16,947.52 Saudi Arabia 10 1,069.12 Malawi 2 18.03
New Zealand 112 8,296.79 Bolivia 9 202.02 Nicaragua 2 8.00
Indonesia 111 16,064.90 Iceland 9 1,019.55 Rwanda 2 12.50
South Africa 100 9,593.32 Moldova 9 232.07 Suriname 2 30.96
Malaysia 98 3,681.07 Ghana 8 1,052.06 Togo 2 14.99
Austria 90 14,525.38 Montenegro 8 178.78 Trinidad and Tobago 2 112.50
Romania 88 4,566.46 Armenia 7 794.09 Brunei 1 23.50
Chile 85 13,560.58 Algeria 7 219.36 Belize 1 369.17
Ukraine 84 13,122.34 Honduras 7 408.50 Congo 1 80.00
Portugal 80 16,461.41 Jamaica 7 855.68 Cape Verde 1 4.97
Japan 76 15,883.20 Mozambique 7 305.09 Djibouti 1 21.74
Korea, Republic 76 16,786.99 Tunisia 7 1,504.75 Iraq 1 34.48
Turkey 69 14,513.93 Tanzania 7 190.10 Korea, Democratic 1 9.80
Hungary 68 17,747.85 Zambia 7 572.86 Lao 1 1.95
Colombia 62 11,892.72 Zimbabwe 7 51.07 Libya 1 145.53
Cyprus 52 8,293.07 Papua New Guinea 6 139.72 Niger 1 2.72
Luxembourg 50 36,804.35 Sudan 6 2,745.00 Nepal 1 333.89
Peru 50 21,078.60 El Salvador 6 788.70 Puerto Rico 1 43.60
United Arab Emirates 44 7,467.81 Dominican Republic 5 2,555.10 Senegal 1 8.44
Thailand 44 1,780.17 Bahamas 5 1,225.13 Sao Tome and Principe 1 32.00
Bulgaria 43 5,397.47 Côte d'Ivoire 5 222.12 Chad 1 202.50
Taiwan 38 1,993.86 Ethiopia 5 421.71 Tajikistan 1 12.00
Serbia 32 2,466.72 Sri Lanka 5 355.18 Turkmenistan 1 2.00
Latvia 30 785.17 Myanmar 5 912.02 Yemen 1 335.00
Philippines 28 7,363.90 Paraguay 5 1,031.19 Total 14,557 3,137,531.60
15
United Kingdom, and Germany are top three countries which attract the greatest amount of cross-border M&As; they receive more than 36% of 14,557 cross-cross-border M&As that take place globally. These three countries also receive the greatest value of cross-border M&A transactions, which counts for more than 45% of the grand total of transaction value for all deals in the sample. Figure 1 shows the pattern of how the total number and total transaction value of cross-border M&As in the sample develop over the period from 1997 to 2015. The total number and total transaction value of cross-border M&As show. The volume and value of cross-border M&As increase both from 1997 and reach a peak in 2000, following a decline due to the stock market crash in 2000. A subsequent wave started from 2003, and the volume and value develop again until 2007, reaching a new peak in 2007, and then the volume and value drop due to the financial crisis. After 2009, both volume and total value of cross-border M&As recover and begin to fluctuate. In 2015, the total value of cross-border M&A regains its level as that of in 2007. The sample employed in this thesis encompasses two whole waves: 1997 to 2003 and 2003 to 2009, and also includes a wave that is still ongoing (after 2009).
0 200 400 600 0 500 1000 1500 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 T ot al of va lue of c ros s-bor de r M & A s (i n bi ll ion U S$) N um be r of c ros s-bor de r M & A s
Total value of cross-border M&As Number of cross-border M&As
16
Table 2 presents the descriptive statistics of the individual sample of each variable. It is important to note that the distribution of the first dependent variable – the number of cross-border M&As (NUMBER'(,*) – is distinct. Among 36,252 observations (1,908 country pairs times 19 years) of the number of cross-border M&As, there are 29,944 observations that have zero cross-border M&A deals. Among 1,908 country pairs, there are 888 country pairs that have only one cross-border deal throughout 19 years. The great proportion of the number of cross-cross-border M&As being zero suggests a zero-gravity problem that is quite common when dealing with bilateral trade. Additionally, the second dependent variable – the average transaction value of cross-border deals (VALUE'(,*) also has a large proportion of observations being zero, but these zero observations are canceled out when taking the natural logarithm of the value, which results in 6,308 observations available in total.
Table 2 Descriptive statistics of individual sample
Mean Median Max. Min. Std. Dev. N Number of cross-border M&As 0.398 0.000 65.000 0.000 1.951 36,252 Average transaction value (natural logarithm) 11.024 10.915 17.139 6.908 2.057 6,308 Host country corruption level 4.160 4.500 9.600 0.000 2.493 36,252 Absolute difference of corruption level 2.737 2.400 9.000 0.000 1.963 36,252 GDP per capita (natural logarithm) 9.214 9.505 11.667 4.714 1.437 35,905 GDP growth (annual %) 3.338 3.300 12.100 -7.300 3.434 35,921 Population (natural logarithm) 16.837 16.807 21.039 11.256 1.686 35,986 Accounting standards 63.576 64.000 83.000 24.000 11.573 21,090 International Trade (annual %) 0.842 0.707 2.039 0.276 0.455 35,646 Cultural distance (natural logarithm) 2.588 2.694 3.499 0.362 0.486 29,294 Geographic distance (natural logarithm) 8.166 8.444 9.866 4.024 1.093 36,252 Currency movement 0.003 0.002 0.167 -0.165 0.082 30,033 Difference of stock market performance 0.000 0.000 1.471 -1.365 0.256 22,691
3.2 Methodology
3.2.1 Estimation method
17
should be carefully managed in model estimation. In order to cope with this problem, gravity model that was developed by Tinbergen (1962) has been applied in various researches. However, given the big proportion of observations being zero, the often-preferred log-linearized gravity specification makes the zeros undefined, which leaves the problem unsolved. Silva & Tenreyro (2006) further present several approaches that have been developed to deal with the zero-gravity problem, and one is to simply discard the zeros from the dataset and to estimate the log-linearized model by Ordinary Least Squares (OLS) regression. However, Garita & van Marrewijk (2007) argue that using OLS to estimate gravity model could be biased in case of heteroscedasticity. Other researches add a constant to each observation of the dependent variable, and use Tobit regression to estimate the model. This method corrects for the zeros provided the zeros are randomly distributed. Yet, it is often the case that zero values are not randomly distributed, which may lead to selection bias (Brakman, et al., 2008). Silva & Tenreyro (2006) contend that the common log-linearized method to solve the gravity problem is incompatible with the existence of zeros of trade data, and log-linearization may also result in truncation of the sample and nonlinear transformations of the dependent variable.
As discussed above, estimating log-linearization by OLS would be problematic, and therefore it is not considered to be an appropriate estimation method. This thesis adopts Poisson pseudo-maximum-likelihood method as suggested by Garita & van Marrewijk (2007) to handle the zero-gravity problem. The total number of cross-border M&As (NUMBER'(,*) as one of the dependent variables is a discrete count data such that Poisson regression model should be applied. However, it is important to point out that Poisson model imposes a restriction that dependent variable should have equal conditional mean and conditional variance. To test the equality, Poisson model will be estimated in the first place, and the fitted value of the total number of cross-border M&As will be obtained. A specification test should then be conducted to check if the total number of cross-border M&As variable has inequality of conditional mean and variance (indicating over-dispersion). In case of the inequality of conditional mean and variance, negative binomial quasi-generalized pseudo-maximum likelihood method that is corrected for over-dispersion is more appropriate.
18
Table 3 Specification test for over-dispersion of the dependent variable - total number of cross-border M&As
Coefficient Std. Error t-Statistic Estimator – Model 1 1.916* 0.110 17.476
Estimator – Model 2 1.923* 1.923 17.225 Estimator – Model 3 1.938* 0.108 17.917
Estimator – Model 4 1.943* 0.110 17.676
* Significant at 1% level
over-dispersion of the total number of cross-border M&As, indicating that negative binomial quasi-generalized pseudo-maximum likelihood method is more an appropriate estimation method. It allows for mean-variance inequality by correcting the variance using coefficients of the estimators presented in Table 3.
Furthermore, the average transaction value of cross-border M&As between home and host country (VALUE'(,*) as another dependent variable is continuous data. OLS regression can be applied to estimate the model according to Uddin & Boateng (2011). All estimations are implemented using Eviews 9.5.
3.2.2 Regression model and variables
Due to the attributes of the data, negative binomial regression model is used to test hypothesis 1a and 1b, of which the aim is to model the relationship between explanatory variables and the likelihood of count outcome. The regression is run through the equation below.
E (NUMBER'(,*= FGH,I CORR(,*, CORR'1(,*, GDP(,* , ∆GDP(,* , POP(,* , ACD(,* , TRA(,* , CUL(,',* , DIS(,',* , EXR'1(,* , MKT'1(,* , YEAR
= exp (αO+ β@CORR(,*+ β=CORR'1(,* + βRGDP(,*+ β>∆GDP(,*+ βSPOP(,*+ βTACD(,*+ βUTRA(,*
+ βVCUL(,',*+ βWDIS(,',*+ β@OEXR'1(,*+ β@@MKT'1(,*+ β@=YEAR)
The multivariate statistical model to test the hypothesis 2a and 2b is formulated as follows:
VALUE'(,*= αO+ β@CORR(,*+ β= CORR'1(,* + βRGDP(,*+ β>∆GDP(,*+ βSPOP(,*+ βTACD(,*+ βUTRA(,*
+ βVCUL(,',*+ βWDIS(,',*+ β@OEXR'1(,*+ β@@MKT'1(,*+ β@=YEAR + ε
19
i in year t. The definition and measurement of each variable in the models are presented in the earlier Variable and data section.
4. Results
4.1 Correlation matrix
A correlation test is performed to test for multicollinearity, before regressions are run. Table 4 presents the correlation matrix which shows the correlation coefficients of each variable. The significance of these correlation coefficients is obtained through performing t-test. It should be noted that GDP(,* and CORR(,* are highly correlated (with a coefficient of -0.770) at 1% significance level, which could imply a problem of multicollinearity. However, CORR(,* is the dependent variable of primary interest in this thesis, and GDP(,* is an explanatory variable that is commonly controlled in various studies. These two variables are both included in regression models, and multicollinearity is tested again by using variance inflation factor (VIF). VIFs of all independent variables in all models are smaller than the threshold of 5 (see Appendix 2), which indicates the multicollinearity in the dataset is not problematic.
4.2 Regression result
4.2.1 Coefficient interpretation of negative binomial regression
Before analyzing the results, it is important to address how the coefficients of negative binomial regression should be interpreted. The coefficient is interpreted as the difference in the logs of expected counts for a one unit change in the explanatory variable, given other explanatory variables remain constant. It can be expressed as: Y = log ]^_`a − log ]^_ = log (]^_`a ]^_)1.
A negative Y suggests ]^_`a < ]^_, which can be explained as the expected count will drop by
de− 1, when it responds to one unit change of explanatory variable f, while holding other explanatory variables constant. A positive Y suggests ]^_`a > ]^_, which means that the expected count will increase by de− 1 when it responds to one unit change of f while other variables remain constant.
1 where Y is the coefficient of explanatory variable f; ]
^_, the dependent variable, is the expected count and the
20 Table 4 Correlation matrix
NUMBER'(,* represents the total number of cross-border M&As between country pair ji in year t; VALUE'(,* represents the annual average transaction value of cross-border
M&As between country pair ji; CORR(,* represents corruption level in host country i; |CORR'1(,*| represents the absolute difference in corruption level of home country j and
host country i; GDP(,* represents GDP per capita in host country i; ∆GDP(,* represents the annual growth of GDP per capita in host country i; POP(,* represents host country size; ACD(,* represents quality of accounting information disclosure in host country; TRA(,* represents international trade of host country; CUL(,',* represents cultural distance
between home and host country; DIS(,',* represents geographic distance between home and host country; EXR'1(,* represents currency movement between home and host
country; MKT'1(,* represents the difference in stock market return of home and host country. * Significant at 1% level.
NUMBER'(,* VALUE'(,* CORR(,* |CORR'1(,*| GDP(,* ∆GDP(,* POP(,* ACD(,* TRA(,* CUL(,',* DIS(,',* EXR'1(,* MKT'1(,*
21
4.2.2 Corruption and volume of cross-border M&As
Table 5 presents the results of four negative binomial regressions to address the impact of corruption on cross-border M&As. The dependent variable is the number of cross-border M&A deals between a particular country pair. Four regressions all include 15,515 observations.
Model 1 includes all independent variables expect the host country corruption and absolute difference of corruption. Model 1 shows that the coefficients of GDP$,&, ∆GDP$,& and POP$,& are significantly positive; and it can be inferred that the change in the number of cross-border deals is positively influenced by host country’s wealth, economic growth and size. These results are as expected and are consistent with the findings of Loree & Guisinger (1995), Focarelli & Pozzolo (2001) and Dunning (1981). In addition, the coefficient of ACD$,& is significantly positive, giving supportive evidence that host country with better ratings on accounting disclosure quality will attract more foreign mergers and acquisition, which is in line with the findings by Rossi & Volpin (2004). Lastly, the coefficient of TRA$,& is positive, but not statistically significant. It implies that trade flows cannot be used to explain the total number of cross-border M&As in the host country. With regards to cross-border characteristics, Model 1 reveals that host country would receive less cross-border M&As when the cultural and geographical distance between home and host country get larger, which is in line with the findings of Buch & Delong (2004). However, the result does not provide evidence regarding the influence of the difference in exchange rate as well as the difference of stock market return between home and host country on cross-border M&A activities as suggested by Erel et al. (2012). Model 1 is statistically significant (QLR = 4,654, p < 0.001).
Model 2 and Model 3 include host country corruption (CORR$,&) and absolute difference in corruption (|CORR9:$,&|) respectively, in attempt to uncover their effects on cross-border M&A volume separately. The coefficients of CORR$,& and |CORR9:$,&| are all statistically significant and negative under these two models, which provides evidence for Hypothesis 1a and 1b. In Model 2, the coefficient of CORR$,& indicates that when corruption level of host country gets one score higher, the number of cross-border deals that host country receives would drop by 8.15%2.
2 log > ?@AB − log >?@ = −0.085, EF@AB EF@ = G :H.HIJ,EF@AB:EF@ EF@ = G
:H.HIJ− 1 ≈ −8.15%, and similar calculation can
22
Similarly, Model 3 suggests that the number of cross-border M&A deals that flow into host country would decrease by 11.93% as home and host country corruption scores differ one unit. Model 4 further includes both host country corruption and absolute difference in corruption in order to uncover the effect of these two explanatory variables simultaneously. These two variables are not highly correlated, hence the interpretation of Model 4 is appropriate. Model 4 is statistically significant (QLR = 4,821, p < 0.001). The coefficient of CORR$,& implies that as corruption score of host country develops one score higher, the difference in the logs of expected counts would deteriorate by 0.062 while holding other variables constant. It can be further interpreted as one corruption score higher that host country gets resulting in 6.01% fewer number of border M&As that host country receives. This result implies that the number of cross-border M&As attracted by host country is negatively related to its corruption level. Therefore, Hypothesis 1a is confirmed. This is consistent with the findings of Malhotra et al. (2010), which report that less corrupt countries receive greater number of cross-border deals that are originated from China and the United States. The result also adds to their findings, since it involves cross-border M&As deals across 157 different countries.
23
Table 5 The effect of corruption on the volume of cross-border M&As
This table presents the estimates of negative binomial regressions with year effect. The dependent variable is the number of cross-border M&A deals between home country j and host country i in year t. CORR$,& represents
corruption level in host country i; |CORR9:$,&| represents the absolute difference in corruption level of home country j
and host country i; GDP$,& represents GDP per capita in host country i; ∆GDP$,& represents the annual growth of GDP
per capita in host country i; POP$,& represents host country size; ACD$,& represents quality of accounting information
disclosure in host country; TRA$,& represents international trade of host country; CUL$,9,& represents cultural distance
between home and host country; DIS$,9,& represents geographic distance between home and host country; EXR9:$,&
represents currency movement between home and host country; MKT9:$,& represents the difference in stock market
return of home and host country. Standard errors are in parentheses. QLR statistic stands for quasi-likelihood ratio statistic. *, **, and *** denote statistical significance at 10%, 5% and 1% respectively.
Model 1 Model 2 Model 3 Model 4
CORR$,& -0.085*** -0.062*** [0.022] [0.022] |CORR9:$,&| -0.127*** -0.121*** [0.014] [0.014] GDP$,& 0.615*** 0.505*** 0.547*** 0.469*** [0.034] [0.044] [0.035] [0.044] ∆GDP$,& 0.054*** 0.047*** 0.056*** 0.050*** [0.013] [0.013] [0.012] [0.013] POP$,& 0.510*** 0.534*** 0.515*** 0.532*** [0.022] [0.023] [0.022] [0.023] ACD$,& 0.021*** 0.016*** 0.021*** 0.018*** [0.003] [0.003] [0.003] [0.003] TRA$,& 0.050 0.035 0.049 0.039 [0.064] [0.065] [0.063] [0.028] CUL$,9,& -0.975*** -0.961*** -0.852*** -0.847*** [0.038] [0.039] [0.040] [0.040] DIS$,9,& -0.225*** -0.224*** -0.217*** -0.216*** [0.019] [0.019] [0.019] [0.019] EXR9:$,& -0.149 -0.125 -0.226 -0.207 [0.269] [0.270] [0.267] [0.268] MKT9:$,& -0.078 -0.087 -0.062 -0.070 [0.107] [0.108] [0.107] [0.107] Constant -13.540*** -12.344*** -13.093*** -12.241*** [0.630] [0.700] [0.626] [0.694]
Year effect Yes Yes Yes Yes
Adjusted R2 0.241 0.241 0.240 0.240
QLR statistic 4,654*** 4,694*** 4,799*** 4,821***
24
4.2.3 Corruption and transaction value of cross-border M&As
Table 6 presents the results of four OLS regressions, which are intended to reveal the effect of corruption on transaction value of cross-border M&As. The dependent variable is the average transaction value of all cross-border M&As between a country pair. All models include 4,039 observations.
Model 5 shows that among all control variables, the coefficients of GDP$,&, ∆GDP$,& and POP$,& are significantly positive, indicating that host country’s wealth, economic growth and country size positively affect the transaction value of cross-border M&As. These results are as expected and are consistent with the findings of Loree & Guisinger (1995), Focarelli & Pozzolo (2001) and Dunning (1981). Moreover, the coefficient of CUL$,9,& indicates that transaction value would be greater when home and host country have similar culture. Model 5 is statistically significant (F statistic = 17.830 , p < 0.001 ), and 10% of the variation in dependent variable can be explained by this model (adjusted Rb = 0.101).
Model 6 and Model 7 attempt to uncover the impact of host country corruption (CORR$,&) and absolute difference in corruption (|CORR9:$,&| ) on transaction value of cross-border M&A individually. In Model 6, the coefficients of CORR$,& is not statistically significant at any significance level, which implies that transaction value of cross-border takeovers cannot be explained by host country corruption level. Hence, Hypothesis 2a is not supported. However, in Model 7, the coefficients of |CORR9:$,&| is statistically significant and negative, which provides evidence for Hypothesis 2b.
In addition, Model 8 presents how the transaction value of cross-border deals is affected by host country corruption (CORR$,&) and absolute difference in corruption (|CORR9:$,&|) simultaneously. Model 8 is statistically significant (F statistic = 17.004, p < 0.001), and it can also explain 10% of the variation in dependent variable (adjusted Rb = 0.103). The coefficients of CORR
25
(2010). Their findings suggest that Chinese bidders which originate from a relatively corrupt country than U.S. bidders would initiate larger cross-border M&A deals to more corrupt country; whereas U.S. bidders tend to have high-value cross-border deals in host countries with lower corruption level (Malhotra et al., 2010).
To sum up, host country corruption has significant and negative influence on the volume of cross-border M&A deals that host country receives, and it has insignificant influence on the transaction value of these deals. The results also support the negative effect of the absolute difference in corruption level between home and host country on the volume as well as the transaction value of cross-border M&A deals.
4.3 Robustness check
Tests were conducted to check the robustness of the results by re-estimating the regression models using another proxy for corruption level, which is the control of corruption index constructed by World Bank. Control of corruption is a component of worldwide governance indicators, and it captures ‘perceptions of the extent to which public power is exercised for private gain’ (World Bank, 2017). This index measures the control of corruption for 215 countries between 1996 and 2015, ranging from - 2.5 (indicating no control) to 2.5 (indicating full control). It is rescaled by deducting the index from 2.5, so that a higher score represents higher corruption level. Two independent variables of primary interest - CORR$,& and |CORR9:$,&| are re-constructed applying this index. Negative binomial regressions and OLS regressions were conducted, and the results are presented in Appendix 3.
26
Table 6 The effect of corruption on transaction value of cross-border M&As
This table presents the estimates of OLS regressions with year effect. The dependent variable is the log-transformed average transaction value of cross-border M&As between home country j and host country i in year t. CORR$,&
represents corruption level in host country i; |CORR9:$,&| represents the absolute difference in corruption level of
home country j and host country i; GDP$,& represents GDP per capita in host country i; ∆GDP$,& represents the annual
growth of GDP per capita in host country i; POP$,& represents host country size; ACD$,& represents quality of
accounting information disclosure in host country; TRA$,& represents international trade of host country; CUL$,9,&
represents cultural distance between home and host country; DIS$,9,& represents geographic distance between home
and host country; EXR9:$,& represents currency movement between home and host country; MKT9:$,& represents the
difference in stock market return of home and host country. Standard errors are in parentheses. *, **, and *** denote
statistical significance at 10%, 5% and 1% respectively.
Model 5 Model 6 Model 7 Model 8
CORR$,& -0.048 -0.030 [0.034] [0.034] |CORR9:$,&| -0.070*** -0.067*** [0.022] [0.023] GDP$,& 0.600*** 0.538*** 0.554*** 0.517** [0.054] [0.070] [0.056] [0.070] ∆GDP$,& 0.041** 0.036* 0.041** 0.038** [0.020] [0.020] [0.019] [0.020] POP$,& 0.374*** 0.387*** 0.377*** 0.386*** [0.033] [0.034] [0.033] [0.034] ACD$,& -0.003 -0.005 -0.003 -0.005 [0.004] [0.004] [0.004] [0.004] TRA$,& -0.128 -0.137 -0.127 -0.133 [0.097] [0.097] [0.097] [0.097] CUL$,9,& -0.420*** -0.413*** -0.363*** -0.362*** [0.058] [0.058] [0.061] [0.061] DIS$,9,& 0.012 0.012 0.016 0.016 [0.028] [0.028] [0.028] [0.028] EXR9:$,& -0.392 -0.380 -0.432 -0.423 [0.422] [0.422] [0.422] [0.422] MKT9:$,& 0.049 0.043 0.071 0.066 [0.174] [0.174] [0.174] [0.174] Constant -0.169 0.516 0.211 0.622 [0.941] [1.059] [0.947] [1.058]
Year effect Yes Yes Yes Yes
Adjusted R2 0.101 0.101 0.103 0.103
F-statistic 17.830*** 17.268*** 17.586*** 17.004***
27 5. Discussion and conclusion
This thesis is dedicated to examining the relationship between corruption on cross-border M&A activities in terms of volume and value, with a sample of 14,557 cross-border M&A deals from 1997 until 2015. This thesis contributes to previous studies on examining corruption as a separate variable that directly explains cross-border M&A activities, and it fills the gap of the research by Malhotra et al. (2010) that only investigates the effect of corruption on cross-border M&As from one developed and one developing country. The data used in this thesis are retrieved from international statistics of cross-border M&As, aggregated by countries of origin and destination. Hence, the results can be generalized to enrich the current understanding in terms of the pattern of how cross-border M&A activities react to corruption.
28
This thesis views corruption as a national characteristic embedded in a country, which can be seen as ‘sand’ in commerce for firms and countries. This thesis arrives at a conclusion that corruption matters for cross-border M&A activities. The results to some extent confirm the theoretical perspective that corruption, as a reflection of a country’ institution quality, not only acts as a barrier to open and equal market access to these foreign bidders, and it is also associated with increasing uncertainties that hinder foreign bidders exploring location advantages and exploiting complementarities through cross-border M&As. Moreover, the results confirm the institutional theory in explaining the effect of the absolute difference in corruption on cross-border M&As. The similarities of institutional profiles in home and host country smooth the firm’s adjustment to institutional pressures from the host country, which consequently lead to more cross-border deals and larger deal value. The results on the effect of absolute difference in corruption also seem to be consistent with psychic (or administrative) distance perspective. Psychic (administrative) closeness results in more cross-border M&As and larger project values, due to the less risk associated and enriched knowledge in dealing with similar business environments.
29
foreign investments, such as cross-border M&A activities; and the control of corruption should be highlighted when public officials promote their country.
This thesis is subject to some limitations, which can be the starting point for future researches. Firstly, the measurement of corruption (Corruption Perceived Index) is perception-based, which involves the views and direct experience with corruption of individuals being surveyed. To the best of my knowledge, Corruption Perceived Index by Transparency International and Control of Corruption by World Bank are two most commonly used measurements in various researches. However, these two corruption measurements cannot accurately reflect the real corruption in a country, due to the measurement challenges. People involved in a corrupt transaction would not admit being corrupt, and their perceptions of corruption would differ individually.
Secondly, further researches could focus on investigating the effect of corruption on cross-border M&As at firm level. In addition to the influences regarding country-level and cross-national level characteristics, the cross-border takeover decisions are also relied on the specific project, the nature of the industry, strength and expansion motives of the firms that participate in cross-border M&As.
30 References
Bardhan, P. (1997). Corruption and development: A review of issues. Journal of
Economic Literature, 35(3), 1320–1346.
Beets, S. D. (2005). Understanding the demand-side issues of international corruption. Journal of Business Ethics, 57(1), 65-81.
Brakman, S., Garita, G., Garretsen, H., & van Marrewijk, C. (2008). Unlocking the value of cross-border mergers and acquisitions.
Brouthers, L. E., Gao, Y., & McNicol, J. P. (2008). Corruption and market attractiveness influences on different types of FDI. Strategic management journal, 29(6), 673-680.
Buch, C. M., & DeLong, G. (2004). Cross-border bank mergers: What lures the rare animal?
Journal of Banking and Finance, 28(9), 2077-2102.
Control Risks Group Limited and Simmons & Simmons. (2006). International business
attitudes to corruption – survey 2006. Retrieved from:
https://www.controlrisks.com/~/media/Public%20Site/Files/Our%20Thinking/International%20b usiness%20attitudes%20to%20corruption%20survey_2006.pdf
Conyon, M. J., Girma, S., Thompson, S., & Wright, P. W. (2002), The productivity and wage effects of foreign acquisition in the United Kingdom. The Journal of Industrial
Economics, 50: 85–102.
Cuervo-Cazurra, A. (2006). Who Cares about Corruption? Journal of International
Business Studies, 37(6), 807-822.
Cuervo-Cazurra, A. (2008). The effectiveness of laws against bribery abroad. Journal of
International Business Studies, 39, 634–651.
Cuervo-Cazurra, A. (2016). Corruption in international business. Journal of World
Business, 51(1), 35–49.
Dikova, D., Panibratov, A., Veselova, A., & Ermolaeva, L. (2016). The joint effect of investment motives and institutional context on Russian international acquisitions. International
31
Dikova, D., Sahib, P. R., & Witteloostuijn, A. V. (2006). The effect of acquisition experience, institutional context and national culture on cross-border merger abandonment and completion. Academy of Management Best Conference Paper, U1–U6.
DiMaggio, P. J., & Powell, W. (1991). The new institutionalism in organizational analysis. Chicago, IL: University of Chicago Press.
Di Giovanni, J. (2003). What drives capital flows? The case of cross-border M&A activity and financial deepening. Journal of International Economics, 127-149.
Dunning, J. (1981). Explaining the international direct investment position of countries: Towards a dynamic or developmental approach. Weltwirtschaftliches Archiv, 117, 30-64.
Dunning, J. (1988). The Eclectic Paradigm of International Production: A Restatement and Some Possible Extensions. Journal of International Business Studies, 19(1), 1-31.
Erel, I., Liao, R., & Weisbach, M. (2012). Determinants of Cross-Border Mergers and Acquisitions. The Journal of Finance, 67(3), 1045-1082.
Focarelli, D., & Pozzolo, A. F. (2001). The patterns of cross-border bank mergers and shareholdings in OECD countries, Journal of Banking & Finance, 25(12), 2305-2337.
Froot, K., & Stein, J. (1991). Exchange Rates and Foreign Direct Investment: An Imperfect Capital Markets Approach. The Quarterly Journal of Economics, 106(4), 1191-1217.
Garita, G., & van Marrewijk, C. (2007). Macroeconomic determinants of cross border mergers and acquisitions. Department of Economics, Erasmus University and Tinbergen Institute Working Paper.
Ghemawat, P. (2001). Distance still matters. Harvard business review, 79(8), 137-147. Glambosky, M., Gleason, K., & Murdock, M. (2015). Political risk and the factors that affect international bids. Global Finance Journal, 28, 68–83.
Guiso, L., Sapienza, P., & Zingales, L. (2006). Does culture affect economic outcomes?
Journal of Economic Perspective, 20 (2), 23–48.
32
Habib, M., & Zurawicki, L. (2002). Corruption and foreign direct investment. Journal of
International Business Studies, 33(2), 291-307.
Hur, J., Parinduri, R. A., & Riyanto, Y. E. (2011). Cross-Border M&A Inflows and Quality of Country Governance: Developing versus Developed Countries. Pacific Economic Review, 16(5), 638-655.
Hyun, H.-J., & Kim, H. H. (2010), The Determinants of Cross-border M&As: The Role of Institutions and Financial Development in the Gravity Model. World Economy, 33: 292–310.
Johanson, J., & Vahlne, J. (1977). The internationalization process of the firm—A model of knowledge development and increasing foreign market commitment. Journal of International
Business Studies, 8, 23–32.
Kaufmann, D. (2004). Corruption, governance and security: challenges for the rich countries and the world. The World Economic Forum, Global Competitiveness Report 2004-2005, Palgrave Macmillan: New York, 83-102.
Ketkar, S. (2014). The rules of global engagement for developing country firms.
Competitiveness Review, 24(2), 124–146.
Kogut, B., & Singh, H. (1988). The effect of national culture on the choice of entry mode. Journal of international business studies, 19(3), 411-432.
La Porta, R. L., Lopez-de-Silanes, F., Shleifer, A., & Vishny, R. W. (1998). Law and finance. Journal of political economy, 106(6), 1113-1155.
Lambsdorff, J. G. (2003). How corruption affects persistent capital flows. Economics of
Governance, 4, 229–243.
Lee, S. H., & Weng, D. H. (2013). Does bribery in the home country promote or dampen firm exports? Strategic Management Journal, 34, 1472–1487.
Loree, D. W., & Guisinger, S. E. (1995). Policy and non-policy determinants of US equity foreign direct investment. Journal of International Business Studies, 26 (2), 281–299.
33
Martin, K. D., Cullen, J. B., Johnson, J. L., & Parboteeah, K. P. (2007). Deciding to bribe: A cross-level analysis of firm and home country influences on bribery activity. Academy of
Management Journal, 50(6), 1401-1422.
Martynova, M., & Renneboog, L. (2008). Spillover of corporate governance standards in cross-border mergers and acquisitions, Journal of Corporate Finance, 14(3), 200-223.
Mauro, P. (1995). Corruption and growth. The quarterly journal of economics, 110(3), 681-712.
Nocke, V., & Yeaple, S. (2007). Cross-border mergers and acquisitions vs. greenfield foreign direct investment: The role of firm heterogeneity, Journal of International Economics, 72(2), 336-365.
Rodriguez, P., Uhlenbruck, K., & Eden, L. (2005). Government corruption and the entry strategies of multinationals. Academy of Management Review, 30(2), 383–396.
Rossi, S., & Volpin, P. F. (2004). Cross-Country Determinants of Mergers and Acquisitions.
Journal of Financial Economics, 74, 277-304.
Silva, J. S., & Tenreyro, S. (2006). The log of gravity. The Review of Economics and
statistics, 88(4), 641-658.
Tanzi, V. (1998). Corruption around the world: Causes, consequences, scope, and cures. Staff Papers, 45(4), 559-594.
Tinbergen, J (1962). Shaping the World Economy: Suggestions for an International Economic Policy. Twentieth Century Fund, New York.
Uddin, M., & Boateng, A. (2011). Explaining the trends in the UK cross-border mergers & acquisitions: An analysis of macro-economic factors. International Business Review, 20(5), 547-556.
UNCTAD. (2000). World Investment Report, United Nations Conference on Trade and Development: New York and Geneva.
34
Uysal, V. B., Kedia, S., & Panchapagesan, V. (2008). Geography and acquirer returns.
Journal of Financial Intermediation, 17(2), 256-275.
Vorachen, M. (2016). Determinants of Cross-Border Mergers & Acquisitions. Master's thesis. Utrecht University.
Voyer, P. A., & Beamish, P. W. (2004). The effect of corruption on Japanese foreign direct investment. Journal of Business Ethics, 50(3), 211-224.
Wei, S. J. (1997). Why is corruption so much more taxing than tax? Arbitrariness kills (No. w6255). National Bureau of Economic Research.
Wei, S. J. (2000). How taxing is corruption on international investors? Review of
Economic and Statistics, 82, 1–11.
Weitzel, U., & Berns, S. (2006). Cross-border takeovers, corruption, and related aspects of governance. Journal of International Business Studies, 37(6), 786–806.
World Bank. (2017). Retrieved from:
http://databank.worldbank.org/data/reports.aspx?source=worldwide-governance-indicators Xie, E., Reddy, K. S., & Liang, J. (2017). Country-specific determinants of cross-border mergers and acquisitions: A comprehensive review and future research directions. Journal of
World Business, 52(2), 127-183.
35 Appendix 1 Description of variables
Variable Abbr. Measurements Source
The number of cross-border M&A deals
NUMBER9$,& Total number of cross-border M&As for each home-host country pair in year t
Zephyr database
Annual average transaction value
VALUE9$,& Log-transformed average transaction value of all cross-border M&As for each home-host country pair in year t
Zephyr database
Host country corruption
CORR$,& Corruption Perception Index (CPI) of host country. For CPI index before the 2012, it is rescaled by subtracting CPI index from 10; for CPI index after 2012, it is rescaled by dividing CPI index by 100 and then subtracting from 10.
Transparency International https://www.transp arency.org/
Control of corruption index (CCI) by World Bank for a robust analysis. The original score is rescaled by multiplying -1. (No data available for 1997, 1999, and 2001.) http://data.worldba nk.org/data- catalog/worldwide- governance-indicators Difference in
corruption between the host and home country
|CORR9:$,&| Absolute difference of rescaled CPI (or CCI) between the host and home country
Calculation
GDP per capita of host country
GDP$,& Natural logarithm of annual GDP divided by total population of host country (in US dollar)
36
Variable Abbr. Measurements Source
GDP growth of host country
∆GDP$,& Annual growth rate of GDP of host country
World bank
Population of host country
POP$,& Natural logarithm of annual population of host country
World bank
Quality of accounting information disclosure of host country
ACD$,& Index assembled by La Porta, Lopez-de-Silanes, Shleifer & Vishny (1998)
La Porta, Lopez-de-Silanes, Shleifer & Vishny (1998) International trade of
host country
TRA$,& Export and import as percentage of GDP
World bank
Cultural distance CUL$,9,& Distance is calculated by taking logarithm of Zhou et al. (2016) formula, using Hofstede’s four most common cultural
dimensions (power distance, uncertainty avoidance,
individualism, and masculinity) CUL$,9,& = ln ( i S9,h − S$,h b hjk 4) https://www.geert-hofstede.com/ Zhou et al. (2016)
Geographic distance DIS$,9,& Log-transformed geographic distance between the capital cities of host and home country
http://geobytes.com /citydistancetool/
Currency movement between home and host country
EXR9:$,& Difference of the annual real bilateral US dollar exchange rate return between home and host country
37
Variable Abbr. Measurements Source
Difference between the stock market return of home and host country
MKT9:$,& Difference of annual real stock return in between home and host country
Datastream
Appendix 2 Table of variance inflation factor (VIF)
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8
CORR$,& NA 4.251 NA 4.293 NA 4.235 NA 4.370 |CORR9:$,&| NA NA 1.309 1.330 NA NA 1.386 1.430 GDP$,& 2.568 4.311 2.672 4.315 2.592 4.314 2.780 4.357 ∆GDP$,& 2.357 2.414 2.358 2.418 1.675 1.724 1.675 1.726 POP$,& 2.168 2.351 2.169 2.348 2.167 2.349 2.170 2.350 ACD$,& 1.505 1.770 1.505 1.772 1.551 1.784 1.552 1.789 TRA$,& 1.847 1.854 1.854 1.860 1.865 1.872 1.865 1.873 CUL$,9,& 1.065 1.073 1.203 1.204 1.072 1.079 1.177 1.177 DIS$,9,& 1.132 1.132 1.133 1.133 1.167 1.167 1.170 1.170 EXR9:$,& 1.028 1.028 1.029 1.030 1.023 1.023 1.024 1.025 MKT9:$,& 1.047 1.047 1.047 1.047 1.050 1.051 1.052 1.053
Appendix 3 Robustness check results
Table 7 Specification test for over-dispersion of dependent variable - mnopqrst,u Coefficient Std. Error t-Statistic Estimator – Model 1 1.823* 0.111 16.410 Estimator – Model 2 1.833* 0.107 17.078
Estimator – Model 3 1.843* 0.110 16.807
