國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
10
28.94% and 36.74% for 2 and 3 digit SIC codes. Results based on IO tables suggest that the vertical relatedness for M&A activities are 21.67% for the 1%
cut-off point but the percentage drop to 5.31% for the 5% cut-off point. Similar results for cross-border M&A are presented in Panel B of Table 1.2. These results indicate that deals that are considered to be pure vertical are much higher at 1% cut-off point than those at 5% cut-off point.
These findings are consistent with previous empirical literature (Ravenscraft and Scherer, 1987; Markids, 1995; Fan and Goyal, 2006; Herger and McCorriston, 2012). Fan and Goyal (2006) show that IO method provides a more sophisticated measure of vertical integration using industry commodity flows information; therefore, we use the IO method to distinguish vertical and horizontal mergers in the following empirical sections.
Additional results based on SIC codes are provided in the robustness section.
5. Empirical Results
To explore the link between cross-border mergers and market concentration in a vertically related industry, we begin by estimating the following equation:
CrossBorder
it=α+β
1RelativeHHI
i.t+β
2HomeVertical
Targett-1,i.t+β
3RelativeHHI
i.t× HomeVerticalTargett-1,i.t+β
4LogDistance
i.t+β
5LogMktSize_Ratio
i.t+β
6CountrySkill_Ratio
i.t+β
7IndustrySkill_Ratio
i.t+β
8CountrySkill_Ratio
i.t× IndustrySkill_Ratioi.t+ 𝜀𝑖,𝑡(1.1) where subscripts i and t index deal and year, and subscript Target the target of M&A. We use a dummy variable, CrossBorder , to measure cross-border merger activity. RelativeHHI , the relative market concentration ratio, is measured by the ratio of target firm’s HHI divided by acquirer firm’s HHI. We analyze the target’s pre-merger activities by a dummy variable, HomeVerticalTargett-1, which is equal to one when target firms had previously been vertically merged at home market and zero otherwise.
Proposition I suggests that the incentives for cross-border mergers rise with vertical integration when the pre-merger HHI at home is sufficiently low relative to the HHI in the foreign country. We test this proposition using the interaction term between RelativeHHI and HomeVerticalTargett-1 in equation (1.1). Columns (1) in Table 1.3 shows that the coefficient of RelativeHHI is positively significant for the 1% cut off point case, which indicates that higher industry competition ratio of target firm to acquirer firm is associated with higher probability of cross border M&A. The interaction term RelativeHHI ×
HomeVertical
Targett-1) in column (1) is significantly negative at the one percent‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
11
level, which supports proposition I1 that the incentives for cross-border mergers rise with vertical integration when the pre-merger HHI at home is sufficiently low relative to the HHI in the foreign country.
Following Alfaro and Charlton (2009), we also control for international factors such as macroeconomic, country, and industry characteristics in equation (1.1). We proxy trade costs using bilateral distance between target and acquirer countries, Distance. LogMktSize_Ratio is the ratio of GDP per capita of target and acquirer countries. Proxy variable for a country’s unit cost of production is given by the ratio of average years of schooling of target and acquirer countries, CountrySkill_Ratio . The industry-level skill ratio,
IndustrySkill_Ratio, is the logarithm of total workers in a given industry. The
interaction term between country-level and industry-level skill, CountrySkill_Ratio × CountrySkill_Ratio , is used to proxy for the comparative advantage of target and acquirer countries. Results show that both Distance and CountrySkill_Ratio have positive and significant effects on the probability of cross-border merger. These results are consistent with Alfaro and Charlton (2009). We also control for proximity between target and acquirer firms, which equals to the ratio of direct to total requirements coefficients (Proximity) and absolute difference between the four-digit SIC codes between target and acquirer (Closeness) used in the previous literature.
1 See Appendeix 1.B
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
12
Table 1.3 Determinants of Cross-Border M&As in Oligopoly Industry.
We estimate the following logit regression model:
CrossBorderit=α+β1RelativeHHIi.t+β2HomeVerticalTargett-1,i.t +β3RelativeHHIi.t× HomeVerticalTarget
t-1,i.t
+β4LogDistancei.t+β5LogMktSize_Ratioi.t +β6CountrySkill_Ratioi.t+β7IndustrySkill_Ratioi.t +β8CountrySkill_Ratioi.t× IndustrySkill_Ratioi.t+ 𝜀𝑖,𝑡
(1.1) This table presents marginal effects for the logit model of Equation (1.1). The dependent variable CrossBorder equals one for cross-border deal and zero otherwise. Columns (1) and (2) use IO-based 1% cut-off point, while columns (3) and (4) use IO-based 5% cut-off point.
HomeVertical equals one for targets that were vertical merged at domestic market and zero otherwise. RelativeHHI is defined as the ratio (Target_HHI/Acquirer_HHI). Target_HHI is the Herfindahl Index of target’s industry. Acquirer_HHI is the Herfindahl Index of acquirer’s industry.
Our Control variables in this model are: Distance is the bilateral distance between target and acquirer country. MarketSize is the GDP per capita in U.S. dollars. MarketSize_Ratio is defined as the ratio (Target_LogMarketSize/Acquirer LogMarketSize). CountrySkill is the high school enrollment years of schooling per worker. CountrySkill_Ratio is defined as the ratio (TargetCountry_Skill/AcquirerCountry_Skill). IndustrySkill is the number of workers in the industry (Unit: Thousand). IndustrySkill_Ratio is defined as the ratio (TargetIndustry_Skill/AcquirerIndustry_Skill). Country_Skill_Ratio×IndustrySkill_Ratio is defined as the interaction term between CountrySkill and IndustrySkill ratio. Proximity is a ratio of the direct to the total inputs used by the firm. Closeness is the absolute difference if four-digit SIC between target and acquirer. The data sources and definitions of the variables are provided in Appendix 1.A. The marks a, b, and c indicate significance at the 1%, 5%, and 10% level, respectively.
Independent variables
Dependent variable: CrossBorder
IO-Based 1% cutoff IO-Based 5% cutoff
Column (1) Column (2) Column (3) Column (4) Coeff p-val. Coeff p-val. Coeff p-val. Coeff p-val.
HomeVertical 0.067 (0.689) 0.113 (0.497) -0.255 (0.430) -0.234 (0.472) RelativeHHI 0.111a (0.000) 0.128a (0.000) 0.065b (0.016) 0.082a (0.003) HomeVertical×RelativeHHI -0.224a (0.006) -0.223a (0.006) 0.003 (0.984) -0.008 (0.960) LogDistance 0.981a (0.000) 0.983a (0.000) 0.987a (0.000) 0.988a (0.000) LogMaketSize_Ratio 1.040 (0.259) 1.126 (0.234) 1.038 (0.263) 1.121 (0.238) CountrySkill_Ratio 1.076b (0.010) 0.967b (0.022) 1.181a (0.005) 1.059b (0.011) IndustrySkill_Ratio 0.010 (0.223) 0.008 (0.327) 0.010 (0.208) 0.008 (0.313) CountrySkill_Ratio×IndustrySkill_Ratio -0.007 (0.363) -0.005 (0.497) -0.007 (0.347) -0.006 (0.484)
Proximity 1.647a (0.000) 1.717a (0.000)
Closeness -0.164 (0.059) -0.171b (0.047)
Constant -10.18a (0.000) -10.26a (0.000) -10.30a (0.000) -10.35a (0.000)
Observations 3,655 3,655 3,655 3,655
Pseudo R2 0.1627 0.1673 0.1593 0.1644
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
13
Column (2) in Table 1.3 shows that the interaction term of RelativeHHI and HomeVerticalTargett-1 is still significant when we add these control variables in equation (1.1). Columns (3) and (4) in Table 1.3 provide results for equation (1.2) based on IO specifications at the 5 percent cut-off point.
Columns (3) and (4) in Table 1.3 show that the coefficients of RelativeHHI are positively significant; however, the parameters of the term between
RelativeHHI and HomeVertical
Targett-1are negative but insignificant at the five percent level.Proposition II2 indicates that a relatively efficient foreign firm has incentive to acquire an integrated domestic firm when the cost differential is sufficiently large. To test this proposition, we estimate models of the following form:
CrossBorder × HomeVertical
Targett-1,i.t
=α+𝛾
1RelativeHHI
i.t+𝛾
2LogDistance
i.t+𝛾
3LogMktSize_Ratio
i.t+𝛾
4CountrySkill_Ratio
i.t+𝛾
5IndustrySkill_Ratio
i.t+𝛾
6CountrySkill_Ratio
i.t× IndustrySkill_Ratioi.t+ 𝜀i,𝑡(1.2) The dependent variable (RelativeHHI × HomeVerticalTarget
t-1) in equation (1.2) is equal to one when target firms had previously been vertically merged at home market by a foreign firm and zero otherwise. Columns (1) and (3) in Table 1.4 show that the coefficients of RelativeHHI are positively significant for the 1% and 5% cut off point cases, which provides empirical support for proposition II. Columns (2) and (4) in Table 1.4 show that the significant results for the RelativeHHI still hold when we include the proximity variable in equation (1.2).
2 See Appendeix 1.B
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
14
Table 1.4 Determinant of Cross-Border Home Vertical M&As in Oligopoly Industry
We estimate the following logit regression model:
CrossBorder × HomeVerticalTargett-1,i.t
=α+𝛾1RelativeHHIi.t+𝛾2LogDistancei.t+𝛾3LogMktSize_Ratioi.t +𝛾4CountrySkill_Ratioi.t+𝛾5IndustrySkill_Ratioi.t
+𝛾6CountrySkill_Ratioi.t× IndustrySkill_Ratioi.t+ 𝜀i,𝑡
(1.2) This table presents marginal effects for the logit model of Equation (1.2). The dependent variables CrossBorder×HomeVertical is the interaction term between CrossBorder and HomeVertical. CrossBorder is a dummy variable, which equals to one if the M&A is cross-border deal and zero otherwise. HomeVertical is a dummy variable, which equals to one if the target has been vertical integrated at home market and zero otherwise. Columns (1) and (2) use IO-based 1% cut-off point, while columns (3) and (4) use IO-based 5% cut-off point. RelativeHHI is defined as the ratio (Target_HHI/Acquirer_HHI). Target_HHI is the Herfindahl Index of target’s industry. Acquirer_HHI is the Herfindahl Index of acquirer’s industry.
Our Control variables in this model are: Distance is the bilateral distance between target and acquirer country. MarketSize is the GDP per capita in U.S. dollars. MarketSize_Ratio is defined as the ratio (Target_LogMarketSize/Acquirer LogMarketSize). CountrySkill is the high school enrollment years of schooling per worker. CountrySkill_Ratio is defined as the ratio (TargetCountry_Skill/AcquirerCountry_Skill). IndustrySkill is the number of workers in the industry (Unit: Thousand). IndustrySkill_Ratio is defined as the ratio (TargetIndustry_Skill/AcquirerIndustry_Skill). Country_Skill_Ratio×IndustrySkill_Ratio is defined as the interaction term between CountrySkill and IndustrySkill ratio. Proximity is a ratio of the direct to the total inputs used by the firm. Closeness is the absolute difference if four-digit SIC between target and acquirer. The data sources and definitions of the variables are provided in Appendix 1.A. The marks a, b, and c indicate significance at the 1%, 5%, and 10% level, respectively.
Independent variables
Dependent variable: CrossBorder × HomeVerticalTargett-1,i.t
IO-Based 1% cutoff IO-Based 5% cutoff
Column (1) Column (2) Column (3) Column (4) Coeff p-val. Coeff p-val. Coeff p-val. Coeff p-val.
RelativeHHI 0.277a (0.000) 0.290a (0.000) 0.294a (0.000) 0.307a (0.000) LogDistance 0.995a (0.000) 0.989a (0.000) 0.992a (0.000) 0.986a (0.000) LogMaketSize_Ratio -0.951a (0.008) -0.953a (0.008) -1.033a (0.004) -1.040a (0.004) CountrySkill_Ratio 2.478a (0.000) 2.478a (0.000) 2.515a (0.000) 2.517a (0.000) IndustrySkill_Ratio -0.005 (0.358) -0.005 (0.368) -0.005 (0.363) -0.005 (0.374) CountrySkill_Ratio×IndustrySkill_Ratio 0.005 (0.347) 0.005 (0.358) 0.005 (0.358) 0.005 (0.368)
Proximity 1.289a (0.000) 1.313a (0.000)
Closeness 0.0108 (0.772) 0.012 (0.741)
Constant -9.592a (0.000) -9.649a (0.000) -9.522a (0.000) -9.578a (0.000)
Observations 3,655 3,655 3,655 3,655
Pseudo R2 0.1775 0.1792 0.1786 0.1804
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
15