APPENDIX A
A.1 70 MNEs in the Food, beverages and tobacco industry.
1. Altria Group 2. Nestlé 3. Unilever 4. Coca-Cola Entr. 5. PepsiCo 6. British American Tobacco 7. Diageo 8. Anheuser-Busch 9. Groupe Danone 10. Archer Daniels 11. Sara Lee 12.Cadbury Schweppes 13. General Mills 14. ConAgra Foods 15. Hain Foods 16. Nissin Food products 17. Saputo 18. San Miguel 19. Meiji Dairies 20. Bimbo Grupo 21. Jm Smucker 22. Danisco 23. CSM 24. Universal 25. Yamazaki Baking 26. Pilgrim's pride 27. Tiger Brands 28. Molson coors brewing 29. Royal Numico 30. McComirck & Co 31. Brown-Forman 32. Corn products intl 33. Kerry Group 34. Hormel Foods 35. Carlsberg
36. Tate & lyle group
37. Constellation brand 38. Sudzucker 39. Foster's group 40. Scottish & Newcastle 41. Wm Wrigley Jr 42. Associated British Foods 43. Dean Foods 44. Gallaher Group plc 45. Smithfield foods 46. Pernod Ricard 47. Campbell soup 48. Tyson Foods 49. Japan Tobacco 50. Heineken 51. Bunge 52. Altadis 53. RJ Reynolds tobacco 54. SABMiller 55. Kellogg 56. Dole Food Co. 57. HJ Heinz
58. Procter and gamble 59. Chiquita Brands Intl. 60. Seaboard 61. Grupo maseca 62. Kraft foods 63. Grupo modelo 64. del Monte 65. Femsa
66. Burns, Philp & Co 67. RHM
A.2 Industries included in sample provided by the U.S Census Bureau.
NAICS industry codes and titles definition
311 Food manufacturing
- 3112 Grain and Oilseed milling
-3113 Sugar and Confectionery Product Manufacturing
-3114 Fruit and Vegetable Preserving and Specialty Food Manufacturing - 3115 Dairy Product manufacturing
- 3116 Animal Slaughtering and Processing -3118 Bakeries and Tortilla Manufacturing
- 3119 Other Food Manufacturing ( Snacks, coffee, tea, miscellaneous) 312 Beverages and Tobacco Product Manufacturing
-3121 Beverage Manufacturing - 3122 Tobacco Manufacturing
A.3 Semi and Double log methods
When using natural logarithms, the resulting coefficients are different in alternative functional forms.
In the following formulations Y represents the dependent variable, X the independent variable, a is the Y-intercept, b is the slope coefficient, ln(Y) and ln(X) represent the natural logarithm of Y and X, respectively, and e is an error term. In this study, the two cases performed are the following:
Case 1
When assessing the ROE as natural logarithm and the independent variables of operational knowledge, operating margin and board size a semi-log model is used:
ln(Y) = a + bX + e .
Case 2
When assessing as natural logarithms ROE and the independent variables of firm size, and capital intensity, a double-log model is used:
ln(Y) = a + b ln(X) + e
A.4 Interaction terms
The regression model 2 includes:
ln ROEi = β0+ β1ln (Capinten*firm size) + β2 ln(Capinten*operaknowledge)
+ β3 ln (Capinten*operamargin) + β7Boardsize*operamargin + β4 Boardsize*firmsize + β5 Boardsize*operaknowledge + ε
The construction of these interaction terms is shown in table 7.
Table 7
Construction of interaction terms
Interaction term Measure
Capital intensity*firm size Capital intensity x firm size
Capital intensity*operational knowledge Capital intensity x operational knowledge Capital intensity*operating margin Natural logarithm of capital intensity x operating margin
Board size*operating margin Board size x operating margin
Board size* firm size Board size x firm size
APPENDIX B Testing of Assumptions
This study utilizes OLS estimation in order to test how firm level factors affect firm performance in terms of ROE. In addition of the tests described in section V, the remaining assumptions underlying such method are fulfilled by the following test.
Normality
After running descriptive statistics including Skewness and Kurtosis values, normality was tested for the sample of 70 MNEs. By the results presented in table 8, it can be concluded that the sample data used in this study is normally distributed, due that the range of skewness values are in the range within +2 to -2, while the kurtosis are within the +3 to -3 range (few authors use + 4 to - 4).
Table 8
Normality test by assessing Skewness and Kurtosis
Variable Skewness Kurtosis
Constant -.65 2.61 Firm size -.03 -.00 Operational knowledge .36 1.32 Operating margin .91 .53 Capital intensity -.11 1.33 Board size .86 1.07 GDP -.46 -.62 Industry -.60 -1.67
Capital intensity x firm size 1.69 2.66
Capital intensity x operational knowledge 1.84 3.90
Capital intensity x operating margin -.25 -1.66
Board size x operating margin .925 .41
Board size x firm size 1.160 2.10
Board size x operational knowledge 1.317 2.68
No autocorrelation
The assessment of the Durbin-Watson value was used to test for autocorrelation in the study sample. Thus, table 9 was obtained by OLS regression where the results (d= 1.74) obtained show that there is no serial correlation in the sample under study, following that the Durbin-Watson value should be between 1.5 and 2.5 to indicate independence of observations.
Table 9 Model Summary
R R Square Adjusted R Square Durbin-Watson
Homoscedasticity
The transformation of the specific variables to natural logarithms achievesnormal distribution and also provides similar variances. This is due to the fact that the observations themselvesmust come from a population which follows a normal distribution, and different groups of observations must come from populationswhich have the same variance or standard deviation (Bland & Altman, 1996).
Figure 2 Scatter plot
Dependent Variable: LNROE
Regression Standardized Predicted Value
3 2 1 0 -1 -2 -3 R e gr e s s ion S tan dar diz ed R e s idua l 3 2 1 0 -1 -2 -3 -4 -5