Introduction
This chapter examines the theory of non-military internal balancing through the use of categorical data analysis (CDA). The specific type of CDA used here is the multinomial logit model (MNLM), which is highly valuable because it determines the probable likelihood that an individual outcome will occur (relative to a defined base outcome). Such models also illustrate how the change in one unit of a given independent variable x (such as economic nationalism) will result in an increased (or decreased) probability of a given outcome y (such as bounded intervention). This type of model is highly flexible and provides a comprehensive picture of the relationship between the individual independent variables and each of the considered outcomes.1
Toward this end, four MNLMs are used to evaluate the hypotheses on which this theory rests. MNLM I tests the primary hypothesis that government intervention into foreign takeovers chiefly depends on geopolitical and economic nationalist concerns in the host state of the target company (state A), controlling for the alternative explanations of economic competition concerns and interest group pressure. In order to explain the puzzle of why such intervention still occurs within security communities, MNLM II restricts the analysis of the primary hypothesis to those deals that take place within the security community context. MNLM III then tests the hypothesis against those cases that have occurred outside of such an environment. A comparison of all three models should thus provide deeper insight into the explanation behind the puzzle. Finally, MNLM IV evaluates the secondary hypothesis that the outcome of cross-border deals is clearly affected by government intervention on the part of state A.
The Variables
Each model uses the database of 209 cross-border M&A transactions specified in the previous chapter. In order to understand the dynamics behind these transactions and their outcomes, the database comprises variables that seek to approximate the concepts put forward in the primary hypothesis as closely as possible. It is important to understand, however, that cross-national and yearly data on these variables are not always available or complete. In certain instances, therefore, it has been necessary to use proxy variables to estimate the desired theoretical concept as closely as possible. Figure 12 explains how this was achieved.
| Conceptual Variables | Proxy Variables | Type | Source | |
|---|---|---|---|---|
| Explanatory Variables | Geopolitical Competition | Security Community | Dummy | CIA World Factbook 2001–07 |
| Relative Military Power | Scale | SIPRI Military Expenditure Database | ||
| Resource Dependency | Scale | International Energy Agency (IEA): International Energy Balances of OECD Countries 2006 and International Energy Balances of Non-OECD Countries 2006 | ||
| Inward FDI | Scale | IMD World Competitiveness Yearbook (WCY) 2001–07 | ||
| Economic Nationalism | Nationalism | Scale | World Values Survey (WVS) 2001–04 | |
| Pro-Globalization Sentiment | Scale | IMD WCY 2001–07 | ||
| Controls | Economic Competition | Economic Competitiveness Index | Scale | IMD WCY 2001–07 |
| Interest Group Presence | Interest Group Position Index | Scale | IMD WCY 2001–07 |
Figure 12 Variable sources
Independent Variables
Geopolitical Competition
The concept of geostrategic concern was measured across four dimensions.
1. Security Community: A dummy variable was created to represent whether state A and state B were part of the same security community at the time of the deal in question. A security community is generally defined here in accordance with Adler and Barnett (Reference Adler and Barnett1998), who argue that the formation of a common identity between states can eventually lead to the development of a norm of non-violence among those states. The result is a community in which participating states resolve their disputes with one another solely through the use of non-violent means, causing a sense of “we-ness” to develop (Adler & Barnett Reference Adler and Barnett1998, 7).2 For the purposes of coding the database, security communities are more strictly defined as those in which this norm of non-violence has been demonstrated to be historically robust, whether through the existence of the highest level of mutual security agreements, such as the North Atlantic Treaty Organization (NATO), through the existence of treaties providing for the sharing of the highest levels of intelligence, such as the American, British, Canadian, Australian and New Zealand Multilateral Master Information Exchange Treaty (ABCANZ), or through membership in a highly integrated political and economic union, such as the EU. These communities are uncommon, and may not always take exactly the same form.
2. Relative Military Power: This variable roughly approximates the relative military power differential between both states involved in the transaction. Its purpose is to shed light on the extent to which state A might have felt threatened by state B, and to indicate when military power differentials might have played a role in causing a given state to seek alternative forms of balancing. It represents the ratio of the military expenditure of state B to that of state A.
3. Resource Dependency: This variable measures the general resource dependency of state A.3 It was calculated in accordance with the method used by the International Energy Agency (IEA),4 utilizing the yearly ratio of state A's energy imports to its energy supply.
4. Inward Foreign Direct Investments (IFDI): This variable is meant to proxy the relative economic power position of state A by representing its IFDI. This measure is used because many political economists argue that it is the rapid or sudden increase in IFDI, rather than changes its net FDI position, that causes a state to react negatively – through protectionism or other means – to conspicuous foreign investment in its country (see e.g., Tyson Reference Tyson1992).
Economic Nationalism
1. Nationalism: A variable measuring the level of nationalism in state A has been included for two reasons. First, nationalism is often a strong component of economic nationalism. The author strongly agrees with Helleiner and Pickel “that economic nationalism, like political or cultural nationalism, can be understood only if it is analyzed in the context of nation and nationalism in general, rather than as just another economic doctrine or ideology (the conventional economistic view)” (Helleiner & Pickel Reference Helleiner and Pickel2005, 12, emphasis in original). Second, comparable cross-national data on economic nationalism per se do not yet exist for many countries outside of the EU, and even within the EU, data are neither yearly nor complete. This variable, therefore, helps to round out an understanding of the theoretical concept of economic nationalism, while providing truly comparable data. Specifically, it represents the percentage of those survey respondents in a given state who said that they were “very proud” of their nationality (see WVS 2001–04).5
2. Pro-Globalization Sentiment: This represents the level of pro-globalization sentiment in state A. As high levels of anti-globalization feeling in a state are often associated with higher levels of economic nationalism, this variable acts as an excellent proxy for the latter concept. Furthermore, anti-globalization attitudes are often identified with the particular form of state action examined in this study, as state intervention into foreign takeovers is often associated with a desire to protect so-called “national champions.” This variable thus represents the level to which respondents (business elites in state A) believed that “attitudes towards globalization [were] generally positive in [their] economy” (IMD 2007a, 6).6
Economic Competition
1. Economic Competition I: In the primary hypothesis, this variable is meant to control for the possibility that a foreign takeover may be blocked on competition, rather than national security, grounds. Data were not readily available for this variable on a case-by-case basis for all countries and within the limits of the author's resources. Moreover, it is a concept for which the creation of a single dummy variable would be inadequate to represent the complexities of the competition review process. The author thus leaves the detailed examination of this concept to the case study section, where it is more appropriately applied.
2. Economic Competition II: This variable seeks to control for a different theoretical concept, which also falls under the rubric of the word “competition.” This is the extent to which state A is believed to be “competitive” in the international market as a state open to liberal economic business practices and foreign investment. Given the nature of the hypothesis, this variable offers another important statistical control to the data being examined here. Thus, an index of the perceived economic competitiveness of state A was made in order to create this variable. This index was formed by taking the average of twelve different measures of economic competitiveness provided by the IMD World Competitiveness Yearbook, which obtains these rankings by conducting a large-n survey of business elites worldwide on a yearly basis (see IMD 2007b, 2016). The IMD measures indexed for the purposes of this dataset are presented in Figure 13.
| IMD Measures of Economic Competitiveness | |
|---|---|
| Relocation Threats of Production | “Relocation of production is not a threat to the future of your economy.” |
| Relocation Threats of R&D Facilities | “Relocation of R&D facilities is not a threat to the future of your economy.” |
| Policy Direction of the Government | “Policy direction of the government is consistent.” |
| Legal and Regulatory Framework | “The legal and regulatory framework encourages the competitiveness of enterprises.” |
| Adaptability of Government Policy | “Adaptability of government policy to changes in the economy is high.” |
| Government Decisions | “Government decisions are effectively implemented.” |
| Political Parties | “Political parties do understand today's economic challenges.” |
| Transparency | “Transparency of government policy is satisfactory.” |
| Public Service | “The public service is independent from political interference.” |
| Bureaucracy | “Bureaucracy does not hinder business activity.” |
| Bribing and Corruption | “Bribing and corruption do not exist in your economy.” |
| Values of Society | “Values of society support competitiveness.” |
Note: Variable definitions sourced from IMD (2007a).
Figure 13 Measures of economic competitiveness sourced from the IMD database
Interest Group Presence
1. Interest Group Position Index: The case study method is used in the following chapters to determine the presence of interest groups, as well as their effectiveness in blocking foreign takeovers, in each of the critical cases. Yet, such an investigation is not practical for the dataset, due to the lack of readily available and comparable data. The index variable used here, however, provides a useful approximation of the concept of interest group presence by measuring the general influence of relevant interest groups within state A. This was achieved by taking the average value of two separate measures of interest group position within these societies, again from the IMD World Competitiveness Yearbook (see IMD 2007b). These measures, outlined in Figure 14, arguably address two of the most critical interest groups involved in cross-border mergers and acquisitions. These are labor unions and shareholders: labor unions because they tend to be one of the more effective groups in voicing their opposition to takeovers and opposing them in costly ways, i.e., through strikes or negotiations, and shareholders for their ability to directly influence the outcome of public takeovers.
| IMD Measures of Interest Group Presence/Position in Society | |
| Labor Relations | “Labor relations are generally productive” |
| Shareholder's Rights | “Shareholder's Rights are sufficiently protected” |
| Interest Group Position Variable used in Dataset | |
| Interest Group Position Index | (Labor Relations + Shareholder's Rights)/2 |
Note: Variable definitions for Labor Relations and Shareholder's Rights sourced from IMD (2007a).
Figure 14 Measures of interest group presence and position from the IMD database
Specification of the Models and Expected Results
MNLM I
The first MNLM tests the primary hypothesis that state A's intervention in the foreign takeovers of companies in its national security sectors will depend primarily on geopolitical competition and economic nationalism, controlling for economic competition and interest group presence. As previously specified, the independent variables tested as part of MNLM I are: security community, relative military power, resource dependency, nationalism, pro-globalization sentiment, economic competitiveness, inward FDI, and interest group position. The dependent variable in MNLM I represents the type of intervention, i.e., the form of non-military internal balancing, that state A chooses to utilize vis-à-vis the foreign takeover in question. This dependent variable, labeled intervention type (Y1), has four value categories, as defined in Figure 15.
| Intervention Type (Y1) | ||
|---|---|---|
| Conceptual Value | Numerical Value | Definition |
| No Intervention | 0 | Those cases in which no apparent government intervention into the foreign takeover in question occurred. |
| Bounded Intervention – Low | 1 | Those cases in which government intervention into the foreign takeover takes the form of lower levels of mitigation intended to result in changes to the deal in question. Low levels of mitigation include, but are not limited to, the creation of a standard national security agreement between the acquiring company and state A, as well as other forms of compliance agreements within the existing national security laws of state A, which are viewed as everyday standard operating procedures within the industry in question. |
| Bounded Intervention – High | 2 | Those cases in which government intervention into the foreign takeover takes the form of higher levels of mitigation intended to result in changes to the deal in question. High levels of mitigation include, but are not limited to, the use of severe national security agreements, the forced creation of proxy boards,* the forced divestiture of strategic assets, and/or the use of novel measures such as the “evergreen clause.”† |
| Unbounded Intervention | 3 | Those cases in which state A (or, in rare cases, a third-party state whose national security interests are also threatened by the potential takeover) either formally or effectively attempts to block a foreign takeover. This may be done through the use of multiple tools, ranging from formal vetoes to high-level government statements or the passage of new laws that prevent the deal from occurring. |
* A proxy board is set up by a proxy agreement (PA) before the takeover can be finalized. See Chapter 1, note 38 for an explanation of PAs in this context.
† See Chapter 1, p. 52 for the definition of an “evergreen clause.”
Figure 15 Values of dependent variable Y1
It should also be noted here that the category of bounded intervention has been divided into two parts: high-bounded and low-bounded intervention (see Figure 15). Parsimony demands that the broader category of bounded intervention is retained within the theory, because governments use both these sub-categories to mitigate rather than block foreign takeovers. However, these sub-categories of bounded intervention utilize different methods, or in some cases the same method to different degrees, to achieve the goal of bounded intervention and modify a specified transaction in state A's favor. Thus, while there are conceptual benefits to maintaining the broader concept of bounded intervention as a whole, these sub-categories are distinct enough to offer valuable insights into the explanation of the puzzle when examined statistically. In addition, a Hausman Test of the independence of irrelevant alternatives (IIA) confirms that all four categories of the dependent variable are statistically distinct and useful.7
It should be the case that the variables representing economic nationalism and geopolitical competition concerns play a significantly larger role than those controlling for interest group presence and economic competition concerns in MNLM I.
MNLM II and MNLM III
MNLM II restricts the cases analyzed to those in which the cross-border M&A deal takes place within the context of a security community. MNLM III restricts the number of cases tested to those that occur outside of that same context. It should be noted here that the independent variable of “resource dependency” drops out of MNLM III for purely statistical reasons: namely, there is insufficient variance on the variable within the population of cases under consideration.8
The purpose behind these additional tests is to determine whether or not the behavior of state A varied significantly under these diverse conditions. If it did, then these models should indicate which independent variables are associated with a higher probability that state A will pursue these different types of intervention vis-à-vis their closest allies. Such information would help explain the puzzle behind this work to a great extent.
MNLM IV
The MNLM IV tests the second hypothesis that the outcome of a foreign takeover is largely determined by the form that state A's intervention takes. The dependent variable of the first three models (intervention type) thus becomes the independent variable in this model. The dependent variable of MNLM IV, labeled deal outcome (Y2), represents the actual outcome of the deal in question, and takes on the values noted in Figure 16. The results of MNLM IV are expected strongly to support the secondary hypothesis.
| Deal Outcome (Y2) | ||
|---|---|---|
| Conceptual Value | Numerical Value | Definition |
| Deal | 0 | Those cases in which the foreign takeover in question was allowed to occur apparently unaltered. |
| Changed Deal | 1 | Those cases in which the foreign takeover in question was (or will be) allowed to go through, but with alterations. This includes deals that were completed, announced, or pending. |
| No Deal | 2 | Those cases in which the foreign takeover in question was not allowed to take place. This category includes both those deals that were withdrawn after a formal announcement and those that were quashed at the stage of a verified market rumor.9 |
Figure 16 Values of dependent variable Y2
Results
The dataset of 209 cases utilized in MNLMs I–IV represents the population of cases of the type of cross-border deal specified in Chapter 1, rather than a random sample. It is interesting to note, therefore, that 158 of these cases (or 75.6%) occurred within the confines of a security community, and only 51 (or 24.4%) occurred outside of the bounds of such a relationship (see Figure 17). Thus, while those deals that crossed the borders of a security community are not rare, they do occur with a much lower frequency. These numbers are extremely important, because they illustrate the extent to which the globalization phenomenon is still largely confined to those states with close military and cultural ties. The notion of a truly globalized international economy, where global financiers can act unimpeded by the borders of such communities, has clearly not yet been realized.
Figure 17 Cross-border deal breakdown: the security community context 2001–07
This differential in “cross-border” deal type further evidences the fear that states experience when confronted by a potential takeover in a sensitive industry from a company that does not reside in an allied country. For these numbers do not include the large number of deals never even attempted by investment bankers, who shied away from economically viable acquisitions as a result of the domestic barriers created by governments to discourage takeovers originating from undesirable sources.
The fear of the acquisition of national champions or sensitive industrials by companies residing in non-allied countries has been intensified by recent concern over the rise of sovereign wealth funds (SWFs) and government investment corporations (GICs). Many political, economic, and market analysts view these funds as not only increasingly powerful within the world economy, but also as potential vehicles for the political goals of the states that control them. The “debate [is] grow[ing] among politicians and policymakers in the US and Europe who are increasingly fearful that some of these investors, including powerful [SWFs], are being driven by political motives, rather than purely financial ones” (Chung Reference Chung2007a). At the same time, a report by the McKinsey Global Institute claims that these SWFs are among the “‘power brokers’ that are having an increasing impact on the world's capital markets,” along with “Asian central banks, hedge funds, and private equity” (Chung Reference Chung2007b).
Thus, it is not at all surprising that states may be more likely to intervene in a given deal in order to protect their national security interests when that deal does not take place within the environment of a well-developed security community. States may also be more concerned about those cases in which states A and B are engaged in intense geopolitical competition with one another, whether that competition is economic, political, or both. The statistical evidence, as demonstrated later, clearly supports such a trend and shows that economic nationalism is of greater importance in determining intervention type among allies.
As almost three-quarters of cross-border deals do take place within the context of a security community, it is not surprising that many deals (151, or 62.7%) have occurred without any apparent government intervention at all since 9/11 (see Figure 18). The remaining 37.3% of transactions, however, did face some form of government intervention, including a number of deals that took place within security communities such as the EU or NATO. The explanation for this behavior may be found in the results of the first three models, reported in Figure 19.
Figure 18 Cross-border deal breakdown by intervention type 2001–07
| Y1: MNLM I (All Deals) | Y1: MNLM II (Deals Within Security Communities) | Y1: MNLM III (Deals Outside Security Communities) | ||||
|---|---|---|---|---|---|---|
| Variable | Coefficient | (S.E.) | Coefficient | (S.E.) | Coefficient | (S.E.) |
| Bounded Intervention Low/No Intervention | ||||||
| Security Community | −0.632 | (0.452) | ||||
| Relative Military Power | −0.000 | (0.015) | −0.003 | (0.014) | −0.313 | (0.254) |
| Resource Dependency | 0.866 | (0.702) | 0.411 | (0.910) | ||
| Nationalism | 2.374 | (1.438)* | 1.824 | (1.916) | 15.459 | (8.623)* |
| Pro-Globalization Sentiment | 0.101 | (0.374) | −0.003 | (0.504) | 1.833 | (1.336) |
| Economic Competitiveness | −0.062 | (0.391) | −0.247 | (0.565) | 0.076 | (1.216) |
| Inward FDI | 0.005 | (0.005) | −0.003 | (0.006) | 0.057 | (0.030)* |
| Interest Group Position | 0.212 | (0.548) | 0.870 | (0.768) | −5.244 | (3.467)† |
| Constant | −4.757 | (1.811)*** | −7.152 | (2.710)*** | 8.680 | (7.589) |
| Bounded Intervention High/No Intervention | ||||||
| Security Community | −0.035 | (0.690) | ||||
| Relative Military Power | −0.105 | (0.071)† | −0.154 | (0.090)* | −0.367 | (0.155)** |
| Resource Dependency | 1.632 | (1.013)† | 2.351 | (1.337)* | ||
| Nationalism | 3.280 | (1.613)* | 2.961 | (1.718)* | 7.552 | (4.387)* |
| Pro-Globalization Sentiment | −0.152 | (0.410) | −0.247 | (0.470) | 0.858 | (1.227) |
| Economic Competitiveness | 0.048 | (0.469) | −0.201 | (0.518) | −0.006 | (1.119) |
| Inward FDI | −0.010 | (0.007)† | −0.014 | (0.010) | −0.006 | (0.008) |
| Interest Group Position | −0.381 | (0.466) | −0.312 | (0.604) | −1.095 | (0.885) |
| Constant | −0.235 | (2.064) | 1.033 | (2.898) | −2.987 | (3.427) |
| Unbounded Intervention/No Intervention | ||||||
| Security Community | −1.810 | (0.700)** | ||||
| Relative Military Power | 0.021 | (0.012)* | −0.003 | (0.010) | 0.089 | (0.036)** |
| Resource Dependency | 0.849 | (1.155) | 2.678 | (1.828)† | ||
| Nationalism | −2.488 | (3.133) | -2.609 | (3.801) | 1.637 | (6.284) |
| Pro-Globalization Sentiment | −1.006 | (0.418)** | −1.861 | (0.688)*** | 0.346 | (0.799) |
| Economic Competitiveness | 0.752 | (0.731) | 1.138 | (0.902) | 0.458 | (0.990) |
| Inward FDI | 0.013 | (0.009) | −0.024 | (0.019) | 0.037 | (0.015)** |
| Interest Group Position | −0.122 | (0.646) | 0.135 | (0.656) | −1.353 | (0.843)† |
| Constant | 1.311 | (1.825) | 0.956 | (4.140) | −1.668 | (3.257) |
| No Intervention/Unbounded Intervention | ||||||
| Security Community | 1.810 | (0.700)*** | ||||
| Relative Military Power | −0.021 | (0.012)* | 0.003 | (0.010) | −0.089 | (0.036)** |
| Resource Dependency | −0.849 | (1.155) | −2.678 | (1.828)† | ||
| Nationalism | 2.488 | (3.133) | 2.609 | (3.801) | −1.637 | (6.284) |
| Pro-Globalization Sentiment | 1.006 | (0.418)** | 1.861 | (0.688)*** | −0.346 | (0.799) |
| Economic Competitiveness | −0.752 | (0.731) | −1.138 | (0.902) | −0.458 | (0.990) |
| Inward FDI | −0.013 | (0.009) | 0.024 | (0.019) | −0.037 | (0.015)** |
| Interest Group Position | 0.122 | (0.646) | −0.135 | (0.656) | 1.353 | (0.843)† |
| Constant | −1.311 | (1.825) | −0.956 | (4.140) | 1.668 | (3.257) |
† p < 0.15; * p < 0.1; ** p < 0.05; *** p < 0.01
| Y1: MNLM I (All Deals) | Y1: MNLM II (Deals Within Security Communities) | Y1: MNLM III (Deals Outside Security Communities) | |
|---|---|---|---|
| Log pseudolikelihood = | −186.937 | −125.355 | −41.517 |
| Number of Observations = | 203 | 152 | 51 |
| Wald chi2 = | 61.75 | 50.67 | 65.36 |
| Prob > chi2 = | 0.000 | 0.000 | 0.000 |
| McFadden's Pseudo R2 = | 0.107 | 0.157 | 0.292 |
Figure 19 Multinomial logit model results: intervention 2001–07
MNLM I
The results of the first MNLM show clear support for the argument that economic nationalism and geopolitical competition are the primary motivations behind non-military internal balancing of this type. The impact of these variables on the probability of state A engaging in either bounded or unbounded intervention – versus the base outcome of no intervention at all – does vary, however.
For example, it can be said with 90% confidence that state A is significantly more likely to use a higher or lower form of bounded intervention, rather than no intervention, as the level of nationalism within state A increases. In general, this finding supports the idea that higher levels of economic nationalism in state A will lead to that state's desire to protect its national interests through such measures. An increase of one standard deviation in nationalism increases the probability that state A will engage in low-bounded intervention by 11.6% and high-bounded intervention by 7.9%. As nationalism increases from its minimum to maximum value, the chance of being in these categories increases by 16.7% and 11.7%, respectively. Figure 20 summarizes the effect of these variables on probability change in MNLM I (see Appendix B for the complete marginal effect tables).
| Independent Variable | Average Change | 0 | 1 | 2 | 3 | |
|---|---|---|---|---|---|---|
| Security Community | 0→1 | 9.11% | 16.67% | −6.86% | 1.55% | −11.36% |
| Relative Military | Min.→Max. | 23.13% | −26.75% | −6.65% | −12.86% | 46.27% |
| Power | +/− s.d. | 13.99% | 19.06% | 4.39% | −27.98% | 4.54% |
| Resource Dependency | Min.→Max. | 21.42% | −42.83% | 13.47% | 25.85% | 25.85% |
| +/− s.d. | 7.49% | −14.98% | 6.39% | 6.90% | 1.69% | |
| Nationalism | Min.→Max. | 14.20% | −19.70% | 16.72% | 11.67% | −8.69% |
| +/− s.d. | 9.71% | −14.17% | 11.57% | 7.85% | −5.24% | |
| Pro-Globalization | Min.→Max. | 28.77% | 41.45% | 15.96% | 0.14% | −57.54% |
| Sentiment | +/− s.d. | 5.48% | 6.09% | 4.88% | −1.67% | −9.29% |
| Economic | Min.→Max. | 9.05% | −11.45% | −6.64% | 0.15% | 17.95% |
| Competitiveness | +/− s.d. | 3.01% | −3.44% | −2.59% | 0.25% | 5.78% |
| Inward Foreign | Min.→Max. | 11.71% | −5.47% | 12.12% | −17.96% | 11.31% |
| Direct Investment | +/− s.d. | 6.64% | −4.51% | 7.17% | −8.77% | 6.11% |
| Interest Group | Min.→Max. | 7.06% | 1.50% | 12.63% | −12.14% | −1.98% |
| Position | +/− s.d. | 3.12% | −0.13% | 6.23% | −5.11% | −0.99% |
Figure 20 Probability change in MNLM I
Three indicators of geopolitical competition between states A and B – relative military power, resource dependency, and IFDI – approach significance in heightening the probability that state A will engage in high-bounded intervention. Increases in the IFDI and resource dependency of state A may therefore play a role in determining state behavior. Interestingly, so may a decrease in the relative power of state B. The results of MNLMs II and III, discussed in the next two sections, indicate that, under certain conditions (namely within security communities), state A may feel more comfortable imposing domestic barriers to foreign takeovers when it is in an advantaged power position versus state B. Yet, it is also clear that when state A is in a position of weakness versus state B, state A will be more likely to engage in unbounded intervention.
Thus, when trying to understand what factors generally motivate states to engage in bounded forms of intervention, nationalism appears to be the most important determining factor. Geopolitical factors may also play a role, though their impact is less clearly defined.
In regard to unbounded intervention, MNLM I indicates that geopolitical competition and nationalism are the chief motivating factors in determining the behavior of state A. In fact, it can be said with 95% confidence that those deals occurring outside of the security community context were significantly more likely to result in state A engaging in unbounded intervention, rather than no intervention at all. For example, when states A and B are members of the same security community, the probability of state A engaging in unbounded intervention decreases by 11.4%. Furthermore, it can be said with 90% confidence that as the military power of state B increases relative to state A, state A is significantly more likely to employ a tool of unbounded intervention to block a foreign takeover. An increase of one standard deviation in the relative military power of state B increases the probability of state A using this category of intervention by 4.5%, but as relative military power moves from its minimum to maximum value, the probability of A engaging in unbounded intervention is raised to 46.3%. Together, these figures provide positive evidence that the motivations behind the behavior of state A will vary in accordance with its alliance relationship to state B.
In terms of economic nationalism, MNLM I also illustrates that a decrease in pro-globalization sentiment will significantly increase the likelihood of state A engaging in unbounded intervention, rather than no intervention at all. An increase of one standard deviation in the value of pro-globalization sentiment decreases the probability of unbounded intervention by 9.3%. In other words, as anti-globalization sentiment increases, the probability of unbounded intervention into foreign takeovers rises by 57.5%.
On the whole, therefore, elements of geopolitical competition and economic nationalism are found to play a clear role in motivating states to engage in this type of non-military internal balancing. Notably, none of the control variables registered as having a significant impact on any level of intervention that state A was likely to choose, indicating that the alternative explanations of interest groups and economic competition concerns cannot necessarily explain state behavior in this model.
MNLM II
Interestingly, none of the variables explored in MNLM II provide insight into why a state would be likely to pursue lower forms of bounded intervention. This phenomenon is likely to be explained by two factors. First, many cases of even low-bounded intervention are classified. Second, low-bounded interventions that occur within the confines of a security community are often not leaked to, or mentioned in, the press. This is because M&A activity is so frequent and regularized between these countries that the companies involved in these transactions see the government's actions as standard operating procedures, and may not feel the need to publicize them of their own accord.
It can be said at the 90% confidence level, however, that high-bounded interventions are significantly more likely to occur when the resource dependency of the target state is high and its levels of nationalism are high. This outcome suggests that elements of economic nationalism and geopolitical competition are again motivating factors for state behavior. The marginal effect of higher levels of resource dependency in state A resulting in high-bounded intervention is 13.2%, and a move from that variable's minimum to maximum value leads to a 44.4% increase in the probability of this form of non-military internal balancing. For nationalism, these numbers are 15.6% and 8.9%, respectively (see Figure 21; for the complete marginal effect tables, see Appendix B). Notably, as in MNLM I, relative military power is significant in the negative direction, indicating that stronger target states are likely to feel they have sufficient power to impose such mitigation measures on their allies.
| Independent Variable | Average Change | 0 | 1 | 2 | 3 | |
|---|---|---|---|---|---|---|
| Relative Military | Min.→Max. | 8.91% | 17.82% | −2.88% | −14.80% | −0.14% |
| Power | +/− s.d. | 19.69% | 34.10% | 5.09% | −39.39% | 0.20% |
| Resource | Min.→Max. | 25.55% | −49.14% | −1.95% | 44.40% | 6.70% |
| Dependency | +/− s.d. | 6.25% | −12.51% | 1.96% | 9.45% | 1.10% |
| Nationalism | Min.→Max. | 10.23% | −19.20% | 11.60% | 8.86% | −1.26% |
| +/− s.d. | 7.00% | −13.27% | 8.04% | 5.97% | −0.73% | |
| Pro-Globalization | Min.→Max. | 36.01% | 60.14% | 11.74% | 0.14% | −72.02% |
| Sentiment | +/− s.d. | 2.77% | 4.71% | 0.83% | −2.80% | −2.74% |
| Economic | Min.→Max. | 8.38% | 10.33% | −12.76% | −4.01% | 6.44% |
| Competitiveness | +/− s.d. | 3.51% | 5.72% | −5.35% | −1.68% | 0.66% |
| Inward Foreign | Min.→Max. | 10.05% | 20.09% | −4.17% | −12.76% | −3.16% |
| Direct Investment | +/− s.d. | 6.14% | 12.28% | −2.90% | −7.87% | −1.51% |
| Interest Group | Min.→Max. | 19.70% | −29.10% | 39.36% | −10.30% | 0.04% |
| Position | +/− s.d. | 9.94% | −15.36% | 19.86% | −4.52% | 0.02% |
Figure 21 Probability change in MNLM II
Furthermore, it can be said with 99% confidence that unbounded intervention into foreign takeovers within security communities is significantly more probable than no intervention when there are lower levels of pro-globalization sentiment in the target state. (A one standard deviation change in this variable decreases the probability of state A engaging in this behavior by approximately 2.7%.) There is a great difference, however, in the value of this variable across countries.10 Thus, it is important to note that as pro-globalization sentiment goes from its minimum to maximum value, the probability of state A engaging in unbounded intervention within a security community decreases by 72.0%. In other words, as anti-globalization sentiment increases (or economic nationalism rises) within state A, it is increasingly likely that state A will utilize a tool of unbounded intervention vis-à-vis a takeover deal emanating from within its own security community. This is in line with the expectation that economic nationalism will play a larger role within security communities, where geopolitical competition is clearly lower.
MNLM III
For the non-security cases examined in MNLM III, nationalism again proves to be a significant factor in increasing the likelihood that a state will engage in either low- or high-bounded intervention. For example, a one standard deviation increase in the value of this variable results in a 13.8% increase in the probability that state A will engage in low-bounded intervention, and a 9.7% increased probability that it will pursue high-bounded intervention (see Figure 22 and Appendix B). A change from the minimum to maximum value of the nationalism variable increases the chances of state A being in these categories by 33.6% and 10.9%, respectively. It is not surprising that this variable, which may be considered a precursor of economic nationalism, plays a role in state behavior here – especially when it is recalled that higher levels of nationalism in state A were associated with a significantly increased likelihood of high-bounded intervention in both MNLMs I and II, and of low-bounded intervention in MNLM I. Clearly, nationalism is associated with states that seek to protect their economic and national security interests by altering or mitigating cross-border deals in their favor.
| Independent Variable | Average Change | 0 | 1 | 2 | 3 | |
|---|---|---|---|---|---|---|
| Relative Military | Min.→Max. | 47.13% | −77.53% | −5.14% | −11.59% | 94.25% |
| Power | +/− s.d. | 43.79% | 62.44% | −23.07% | −64.52% | 25.14% |
| Nationalism | Min.→Max. | 22.80% | −45.59% | 33.60% | 10.90% | 1.10% |
| +/− s.d. | 12.88% | −25.75% | 13.80% | 9.68% | 2.27% | |
| Pro-Globalization | Min.→Max. | 18.70% | −37.39% | 22.09% | 9.27% | 6.03% |
| Sentiment | +/− s.d. | 8.18% | −16.35% | 7.20% | 5.47% | 3.68% |
| Economic | Min.→Max. | 6.81% | −12.96% | 0.24% | −0.67% | 13.39% |
| Competitiveness | +/− s.d. | 3.03% | −5.76% | 0.13% | −0.30% | 5.93% |
| Inward Foreign | Min.→Max. | 40.97% | −74.52% | 45.62% | −7.41% | 36.32% |
| Direct Investment | +/− s.d. | 25.11% | −45.55% | 16.58% | −4.67% | 33.63% |
| Interest Group | Min.→Max. | 49.82% | 96.87% | −99.65% | 1.12% | 1.66% |
| Position | +/− s.d. | 28.45% | 56.89% | −37.68% | −4.87% | −14.34% |
Figure 22 Probability change in MNLM III
Most significantly, however, MNLM III demonstrates the greater importance placed on the geopolitical relationship of states A and B in determining intervention type for non-security community cross-border deals. To begin with, it can be said with 90% confidence that higher levels of IFDI in state A significantly increase the likelihood that state A will utilize low-bounded intervention in non-security community deals, rather than no intervention at all. Indeed, a one standard deviation increase in the level of IFDI in state A increases the probability that state A will use this tool of intervention by 16.6%. This, of course, illustrates the heightened importance that the relative economic power positions of states A and B play in determining even the lesser forms of intervention that occur outside of the security community context.
As with MNLM II, state A is significantly more likely to engage in high-bounded intervention as its military power relative to state B increases. Initially, this seems contradictory. As mentioned already, however, the likely explanation is that as state A's power increases relative to state B, it may feel more comfortable engaging in a form of intervention that simply seeks to impose changes to a deal, rather than blocking it. In such a situation, state A may feel both that it is easier to impose conditions on a transaction that involves a weaker non-allied state and that more severe measures are unnecessary because of the less threatening nature of such a state.
In the category of unbounded intervention, we see that the theory is again borne out. If the acquirer's state (state B) has a higher level of military power relative to the target company's state (state A), then it can be said with 95% confidence that the target state's government is significantly more likely to intervene in this most strenuous of ways, rather than not at all. In fact, a one standard deviation increase in the military power of state B relative to state A increases the probability of unbounded intervention on the part of state A by 25.1%. An increase from the minimum to the maximum value of the military power of state B relative to state A raises the probability of unbounded intervention on the part of state A by an astounding 94.3%.
Relative economic power proves to be an important element of geopolitical competition here as well, as higher levels of IFDI are shown to be significant in determining the probability of this course of action on the part of state A. A one standard deviation increase in the IFDI of state A raises the probability of unbounded intervention by 33.6%. On the whole, therefore, it does seem clear that outside of the security community context, foreign takeovers are more likely to be barred on the basis of geopolitical concerns.
MNLM IV
MNLM IV tests the effect of government intervention on the outcome of cross-border deals. Here, intervention type now acts as the independent variable, and the dependent variable is deal outcome, i.e., deal, changed deal, or no deal. Out of the 209 cases in the database, 104 deals went through with no apparent changes imposed by governments, 48 were mitigated in some way by the state in question, and 57 resulted in “no deal” (see Figure 23).
Figure 23 Cross-border deal breakdown by deal outcome 2001–07
Significantly, the results of MNLM IV show clear evidence that the type of government intervention employed by state A will affect the outcome of the foreign takeover in question (see Figure 24). It may be said with 99.9% confidence that intervention type significantly affects the likely outcome of the dependent variable across each of its categories. Furthermore, as the level of government intervention increases, so do the chances that the deal will be barred or mitigated. For example, a one-unit increase in intervention type – i.e., an increase from no intervention to low-bounded intervention, or from high-bounded intervention to unbounded intervention – decreases the probability of an unaltered deal outcome by 111.3% (see Figure 25). Such a one-unit increase in the level of intervention, then, concurrently increases the probability of a “changed deal” outcome by 53.8% and raises the probability of a “no deal” outcome by 57.5%.
| Variable | Y2: MNLM IV (All Deals) | |
|---|---|---|
| Coefficient | (S.E.) | |
| No Deal/Deal | ||
| Intervention Type | 2.736 | (0.864)*** |
| Constant | −1.600 | (0.286)*** |
| Changed Deal/Deal | ||
| Intervention Type | 2.945 | (0.860)*** |
| Constant | −2.033 | (0.191)*** |
| Deal/No Deal | ||
| Intervention Type | −2.736 | (0.864)*** |
| Constant | 1.600 | (0.286)*** |
| Log pseudolikelihood = | −162.227 | |
| Number of Observations = | 209 | |
| Wald chi2 = | 12.660 | |
| Prob > chi2 = | 0.009 | |
| McFadden's Pseudo R2 = | 0.253 | |
† p < 0.15; * p < 0.1; ** p < 0.05; *** p < 0.01
Figure 24 Multinomial logit model results: deal outcome 2001–07
| Independent Variable | Average Change | 0 | 1 | 2 | |
|---|---|---|---|---|---|
| Intervention Type | Min.→Max. | 49.97% | −74.96% | 44.90% | 30.06% |
| +/− 1 | 74.20% | −111.31% | 53.76% | 57.54% | |
| +/− s.d. | 72.92% | −109.38% | 52.83% | 56.55% |
Figure 25 Probability change in MNLM IV
Conclusion
MNLMs I–III bear out the argument that intervention type is primarily motivated by geopolitical competition concerns and economic nationalism. Increased levels of nationalism (which can serve as a preliminary indicator of economic nationalism) in state A were associated with a significantly higher likelihood of low-bounded intervention in MNLMs I and III and of high-bounded intervention in all three models. This suggests that nationalism may motivate states to protect their interests by seeking to mitigate cross-border deals in their favor. Anti-globalization sentiment, which can be most clearly linked with economic nationalism per se, proved to be highly significant in increasing the probability of unbounded intervention generally, as seen in MNLM I, and even more significant within security communities, as shown in MNLM II. This finding indicates that economic nationalism may be a greater motivating factor for intervention than geopolitical concerns within security communities, where such strategic competition is by definition lower and occurs on a much longer time scale, and, consequently, where elements of economic power may be viewed as more important.
This idea is corroborated by the role played by geopolitical competition (approximated by relative military power) in MNLMs I–III. When state A's military power was greater than that of state B, it was significantly more likely to engage in high-bounded intervention for deals both inside and outside of a security community, most likely because it felt more comfortable mitigating transactions in an advantageous power position. Yet, when the military power of state A was lower than that of state B, it was generally more likely to engage in unbounded intervention, as illustrated in MNLM I, and even more likely to do so outside of the security community context, as seen in MNLM III.
Other elements of geopolitical concern also play a vital role in determining intervention type. Higher levels of resource dependency in state A significantly increased the likelihood of high-bounded intervention within security communities. Furthermore, outside of security communities, inward FDI levels in state A played an important part in determining intervention type. For instance, when states A and B were not part of the same security community, IFDI was a highly significant factor in enhancing the probability that state A would engage in unbounded intervention to internally balance the relative economic power of state B.
It is also important to note that the models do not support the alternative explanations explored in the previous chapter, and controlled for in the primary hypothesis. The only instance in which a control variable neared statistical significance was in MNLM III, where interest group position was shown to approach significance in increasing the likelihood of low-bounded and unbounded interventions. As will become clear in the next chapter, however, interest groups are rarely able to affect a government's decision to intervene in a foreign takeover, even when they actively attempt to do so.
Finally, MNLM IV strongly supports the secondary hypothesis that intervention type significantly affects deal outcomes. As the level of government intervention increased, so did the chances that a deal would be mitigated or blocked. This is vital to the theory presented in this book, and is a result that could not have been as comprehensively tested through the case study method.
The numbers, therefore, clearly support the theory of non-military internal balancing. Moreover, the information provided in these models may offer a preliminary answer to the puzzle of why interventions occur within security communities. Geopolitical competition and nationalism seem to motivate bounded forms of intervention, whatever the security community context. It is now clear, however, that in cases of unbounded intervention, economic nationalism may be a larger motivating factor within security communities, and geopolitical competition a larger motivating factor outside of that context. The case studies that follow should provide additional evidence in support of the theory, and add further detail to these initial insights.
Open access