COUNTRY, INDUSTRY AND FIRM HETEROGENEITY
AS DETERMINANTS OF FIRM PERFORMANCE:
A COMPARISON BETWEEN DEVELOPED
AND DEVELOPING COUNTRIES
Judith Alise Hoogeboom Student Number: 1533819
UNIVERSITY OF GRONINGEN Faculty of Economics and Business
Master Thesis - International Business and Management June 2011
ABSTRACT
Conforming to previous research on variation in firm performance, this study examines country-level, industry-level and firm-level heterogeneity as determinants of firm performance in developed and developing countries. This paper empirically analyses a sample of 47,111 corporations from 27 developed and 14 developing countries. Our Variance Components Analysis (VCA) shows that the relative importance of country, industry and firm as components are similar in developed and in developing countries. Results show that firm heterogeneity accounts for the majority of variation in firm performance, in both developed and developing countries. A comparison between developed and developing countries provides strong evidence that country and country-industry interaction effects exert a stronger impact in developed countries than in developing countries, whereas between-industry and firm variation attributes for more variation in developing countries than in developed countries. In addition to estimating the relative importance of those factors, specific country, industry and firm determinants provide an explanation of variance. Our results show that determinants economic growth, inflation and industry profits are more important for explaining firm performance in developing countries, whilst determinants firm size, capital intensity and industry growth are more important for explaining firm performance in developed countries. Therefore, our results of decomposing variation in firm performance demonstrate that country, industry, and firm are of similar relative importance in developed and in developing countries. The results also offer important insights on specific country, industry, and firm determinants as explanation of firm performance in developed and in developing countries.
TABLE OF CONTENTS Abstract 2 Table of Contents 3 List of Tables 5 1. Introduction 6 2. Theoretical Framework 10
2.1 Background to Theoretical Approaches 10
2.2 Firm and Industry Heterogeneity 11
2.3 Country Heterogeneity as Additional Component 14
2.4 Summary 17
3. Theory and Hypotheses 18
3.1 The Importance of Country-level, Industry-level and Firm-level Heterogeneity 18
in Developed vis-à-vis Developing Countries
3.2 Determinants of Firm Performance 19
3.2.1 Country Determinants 19
3.2.2 Industry Determinants 21
3.2.3 Firm Determinants 22
4. Empirical Analyses 24
4.1 Data 24
4.1.1 Dependent Variable: Firm Performance 24
4.1.2 Independent Variables 26
4.1.3 Sample and Sampling Design 27
5. Empirical Results 32
5.1 Variance Components of Firm Performance 32
5.2 Determinants of Variation in Firm Performance 34
5.2.1 Country Determinants 35
5.2.2 Industry Determinants 37
5.2.3 Firm Determinants 40
5.3 Robustness Check 43
5.3.1 Results for ANOVA 43
5.3.2 Results for Different Samples 46
6. Conclusion 50
References 54
Appendix 59
A. Descriptive Statistics 59
LIST OF TABLES
List of Tables
Table 1. Descriptive Statistics for Key Variables 25
Table 2. Variance Components for Mean Firm Performance 32
Table 3. Country Determinants 36
Table 4. Industry Determinants 39
Table 5. Firm Determinants 42
Table 6. Robustness for Variance Components 43
Table 7. Robustness for Outliers in Variance Components 45
Table 8. Robustness for Outliers in Country Determinants 47
Table 9. Robustness for Outliers in Industry Determinants 48
Table 10. Robustness for Outliers in Firm Determinants 49
Appendix
Table A.1 List of Countries included in this Research 59
Table B.1 Robustness for Country Determinants 61
Table B.2 Robustness for Industry Determinants 62
1. INTRODUCTION
Since the research of Schmalensee (1985) on the effect of industry and corporate factors on firm performance, explaining differences in firm performance has become a central theoretical and empirical issue in the field of international business and strategic management. Two important perspectives provide theory of performance variation between corporations. Initial research suggests that firm effects are negligible and industry effects explain the majority of variation in firm performance, which supports Industrial Organization (IO) theory (Schmalensee, 1985; Rumert, 1991; McGahan and Porter, 1997; Chang and Singh, 2000; Hawawini et al., 2003). Other research indicates that firm effects do exist and are relatively more important than industry effects, which are consistent with the Resource Based View (RBV) theory (Roquebert et al., 1996; Brush et al., 1999; Bowman and Helfat, 2001; Ruefli and Wiggins, 2003). Since the research of Schmalensee (1985) on influence of industry and corporate factors on firm performance, explaining the variation in firm performance by components became a well accepted research topic in the field of international business and strategic management (Roquebert et al., 1996; Ruefli and Wiggins, 2003; Hawawini et al., 2004; Makino et al., 2004; McGahan and Porter, 2005; Chen, 2008; Galbreath and Galvin, 2008).By describing the relative contribution of country-level, industry-level and firm-level effects on firm performance, research in this field seeks to discover whether managerial involvement throughout strategic activities influences firm performance compared to the influence of industry structure or macro-economic conditions. Industry structure and macro-economic conditions are external factors, which are frequently outside the control of corporate management.
cross-country differences in firm performance. An important study in this field by Makino et al. (2004) reveals a pattern in variation of firm affiliates performance among different countries. This research suggests that firm factors are relatively more important for variation in firm performance of MNC affiliates in developed countries (4.8% for small Less Developed Countries (LDCs) and 8.3% for large LDCs) than in developing countries (11.3% for Newly Industrializing Economies (NIE) and 13.4% for Developed Countries (DC)). In addition, country and industry factors account for a larger part of variation in firm performance of MNC affiliates in developing countries (respectively 7.7% and 8.8% for small LDCs and 6.2% and 7.6% for large LDCs) than in developed countries (4.4% and 6.7% for NIE and 3.6% and 5.5% for DC). A possible explanation for these results, which suggest that firm-specific characteristics of MNCs from developed countries play the most important role in explaining firm’ performance, is that developed countries are more integrated in terms of market transactions, infrastructure, institutional rules and enforcement mechanisms (Makino et al., 2004; Chan et al., 2008). Moreover, Makino et al. (2004) and Chan et al. (2008) argue that country and industry factors differ significantly among MNC affiliates in developing countries because of underdeveloped and less integrated market transactions, infrastructure, institutional rules, and enforced mechanisms. Those results indicate that countries and industries differ more when it concerns developing countries than when it concerns developed countries. The main contribution of Makino et al. (2004) is that their research demonstrates that variation in firm performance attributes to country differences.
face the double challenges of growth and globalization in an extremely competitive global market. As a result, our research problem is the lack of research and information about the extent of country, industry, and firm effects in firm performance for developing countries in comparison to developed countries. This paper responds to that research problem and contributes to the existing literature in two explicit ways. First, in contrast with the majority of research of variance decomposition, which used one or more developed countries (Roquebert et al., 1996; Ruefli and Wiggins, 2003; Hawawini et al., 2004; Makino et al., 2004; McGahan and Porter, 2005; Chen, 2008; Galbreath and Galvin, 2008), we use data on corporations from developed countries and developing countries. More specifically, our research describes the relative contribution of country-level, industry-level and firm-level heterogeneity for differences in firm performance, in both developed and developing countries. This research seeks to discover whether managerial involvement throughout strategic activities influences firm performance compared to the influence of industry structure or macro-economic conditions in those countries, which are frequently outside the control of corporate management. Still, estimating the relative importance of country, industry and firm does not provide an explanation of variance. Therefore, our second objective is to analyse specific country, industry, and firm determinants and see whether these determinants explain firm performance. As a result, the following research question covers and leads our research:
How much do country-level, industry-level and firm-level heterogeneity account for variation in performance of corporations in developed and developing countries and what is the quantitative importance of specific country, industry and firm determinants to explaining variation in firm performance in developed and developing countries?
Figure 1. Conceptual Model of this Study
2. THEORETICAL FRAMEWORK
A central question in strategic management literature concerns the source of corporations’ competitive advantage. Two important theoretical approaches that explain performance differences, Industrial Organization and Resource Based View, are discussed. Further, previous research has explained variance in profitability by different components. This Chapter describes a large number of empirical studies, which have examined performance differences on firm and industry effects in addition to country effects. Finally, a summary is provided of this theoretical framework on country-level, industry-level and firm-level heterogeneity in firm performance.
2.1 Background to Theoretical Approaches
A central theoretical and empirical discussion in the field of international business and strategic management concentrates on the extent to which firm performance varies across corporations and industries. From a classical perspective, corporations are seen as single-business units whose performance is primarily a function of structural components of an industry (Bain, 1968). Research that attributes performance differences to industry effects from an Industrial Organization (IO) perspective emphasises the importance of external market factors. IO economists refer mainly to the Structure-Conduct-Performance model (Weiss, 1974), which suggests that the market structure determines firm performance of corporations in that industry. The structure of the industry refers to factors that shape the competitiveness of the market. In this framework, the industry structure in which the corporation operates is the primary determinant of variation in firms’ long-term profitability. As a result of industry structure, IO argues that variation in firm performance is greater between rather than within industries. Mason (1939) explains those structural characteristics of an industry as the source for different conducted strategies, which clarify the differences in firm performance
factors rather than external factors (Wernerfelt, 1984). Therefore, differences in firm performance are greater between corporations than between industries. Within this theory, the focus lies on the unique resources of every corporation individually, which the corporation should deploy in order to benefit from its competitive advantage. Idiosyncratic immobile resources are seen as valuable resources that are neither perfectly imitable nor substitutable without serious attempt (Barney, 1991). Since strategies and structures between corporations differ both between and within industries, the internal resources and capabilities of each corporation differ. As a result of internal resources and capabilities, RBV argues that differences in firm performance are greater between corporations than between industries and determine differences in firm performance (Nelson, 1991).
2.2 Industry and Firm Heterogeneity
focuses on firms’ accounting returns at the level of industry, Wernerfelt and Montgomery (1988) employ Tobin’s q as performance measure and conduct their analysis at the level of firms.
Rumelt (1991) reanalyses the data from FTC for a period from 1974-1977 and uses instead of a fixed-effects model (Schmalensee, 1985) a random-effects model. While Schmalensee (1985) reports that industry factors are main determinants of variance in firm performance, research of Rumelt (1991) differentiates between stable and fluctuating influences. Based on that distinction, Rumelt (1991) argues that stable industry factors only explain a small part of the variance and that stable business-unit factors explain the majority. Those results can be explained by the fact that business units differ more within industries than industries differ among themselves. Like Schmalensee (1985), Rumelt (1991) confirms that firm factors are negligible for explaining variance in firm performance. The initial stream of research in this field is therefore descriptive instead of causal relation testing of developed theories (Schmalensee, 1985; Rumelt, 1991).
Several authors have replicated the studies of Schmalensee (1985) and Rumelt (1991) with different kind of databases and methods. McGahan and Porter (1997) continue research on differences in performance across firms by using comprehensive data and improved statistical methods. Neither Schmalensee (1985) nor Rumelt (1991) include economic sectors as possible determinant of variance in their research. Research of McGahan and Porter (1997) distinct from previous research by using business segments (SIC codes) as unit of analysis rather than business unit from Compustat. McGahan and Porter (1997) support the assumptions of Schmalensee (1985) that industry factors are an important determinant for variation in firm profitability. Their outcomes show that the variance in firm performance is explained by 2% year effects, 19% stable industry effects, 4% stable corporate-parent effects and 32% stable segment-specific effects. However, their research challenges the conclusions of Rumelt (1991) concerning the relatively low impact of stable industry effects.
are insignificant, their research differs from traditional IO theory. Rumelt (1991) suggests that, in the long run, differences between corporations may result in different profits. In contrast with research of McGahan and Porter (1997), research by Roquebert et al. (1996) confirms the suggestion of Rumelt (1991) that variance within industries is greater than within industries. In particular, Roquebert et al. (1996) suggest that the relative importance of corporate effects increases as the average number of lines of business decreases. However, their results suggest that firm effects are more important as source of variation in performance than industry effects.
those results are not the same as estimating the percentage influence of industry and firm effects on firm performance (McGahan and Porter, 1999; 2003). Ordinary Least Squares analysis mainly shows that estimating the relative importance of firm and industry effects on firm performance is a complex issue. Another study by Ruefli and Wiggins (2003) supports the assumption that industry effects persist longer than business and corporate effects.
More recently, a study by Hawawini et al. (2003) on industry and firm effects shows that generally industry effects have little influence on firm performance. Their results indicate that firm effects dictate firm performance irrespective of the measure of performance used. The stable firm effects account for more variance in firm performance than industry effects do. The industry effects explain only 10.7% for Economic Profit divided by Capital Employed, 14.3% for Total Market Value divided by Capital Employed and 11.2% for Return on Assets (ROA) as compared to firm effects, which attribute respectively 27.1%, 32.5%, and 35.8%. However, after excluding the two highest and lowest performers from every industry they alter their assumption. It seems that industry structure is only important for corporations that perform average and not for high-performers and low-high-performers. Previous research from Roquebert et al. (1996), McGahan and Porter (2002), Ruefli and Wiggins (2003) and Hawawini et al. (2003) support the RBV of firms who found that corporate effects determine firm performance (Wernerfelt, 1984; Barney, 1991; Nelson, 1991).
2.3 Country Heterogeneity as Additional Component
In addition to firm and industry effects on firm performance, some research refers to the influence of industry interaction. For example, Porter (1990) refers to the country-industry interaction effects in the Diamond Framework. The Diamond Framework offers a framework to analyse and understand a nation competitive position. Porter’s theory analyses national competitiveness and excludes firm performance. Nevertheless, this framework shows that corporations and industry structure varies across countries and obtain different performances across countries.
different countries by adding country effects as extra determinant of variation in firm performance. While IO and RBV theories initiate discussion of industry versus firm effects of firm performance, exploring country effects are more complex, because of missing theories. Hawawini et al. (2003) mention that there are various mechanisms by which a country influences firm performance: macro-economic conditions, factor endowments and legal, social and cultural distinctiveness. However, discussing all those different mechanisms goes beyond the hypotheses in this paper. For the economic attributions, Ghemawat (2003) argues that semi-globalization, noticeable by significant barriers to the economic integration of countries, matters for firm performance, based on evidence from international trade, foreign direct investments, geographical distribution, price dispersion among countries and international flow of production factors. Although, in economic terms, reducing barriers for economic integration would suggest that country effects become less relevant for differences in firm performance. Theoretically, research by Ghemawat (2003) and Hawawini et al. (2003, 2004) shows that country is a relevant source of variation in firm performance.
Foremost empirical research in this field distinguishes diverse country effects, for example host-country effects (Makino et al., 2004), and home-country effects (Hawawini et al., 2004; McGahan and Victer, 2010). Hawawini et al. (2004) examine the relative importance of home country effects on firm performance in a world of increasing market integration. Hawawini et al. (2004) conduct economic and financial performance measures for 1,305 corporations in six countries (USA, UK, Germany, Netherlands, Belgium, and Luxemburg) from 1993 to 1996. Their results show that performance differences cannot be significantly explained by external factors like country, industry, and year effects. Hawawini et al. (2004) argue that firm effects explain the majority of variation in firm performance for different performance measures (ROA, market value and economic profit). Their research favours the RBV that highlights the importance of internal factors, where the unique resources are a result of managerial capabilities rather than a result of industry and country factors (Hawawini et al., 2004).
country-industry effects in construction, 12.2% in wholesale-retail and 45% in transport. Similar to Brito and Vasconcelos (2006), McGahan and Victor (2006) analyse a sample of Compustat Global counting ROA of 4,551 corporations in 43 countries. Their research includes only country, industry and year effects, as well as the interactions among these effects. Results indicate that those fixed effects only account for 31.36% of the total variation in firm performance. Moreover, these results suggest that excluding fixed firm effects from the analysis may increase the relative importance of country, industry, and year effects for firm performance.
Further, Makino et al. (2004) suggest that country matters for affiliates’ performance in host countries. This finding supports the main argument in conventional international business literature and institutional theory that country factors have an impact on firm strategy and performance. Captivatingly, their results show that country effects and industry effects are more important than internal firm effects in developing countries, whereas firm effects are more important in developed countries (Makino et al., 2004). Another interesting result of their research is the relative large corporate effect compared to previous literature (Brush and Bromiley, 1997; Bowman and Helfat, 2001). A possible explanation for extreme larger corporate effects is that MNCs differ from domestic corporations, which have been included in previous research. McGahan and Victer (2010) support these findings and demonstrate that home-country and industry effects are more important to domestic corporations than to MNCs. Because of these outcomes, they suggest that single-country studies should be interpreted carefully since those corporations are influenced by the home country.
from industry and firm effects, this research incorporates country and country-industry interaction effects as additional components of variation in firm performance (Ghemawat, 2003; Hawawini et al., 2004; Makino et al., 2004; Chen, 2008).
2.4 Summary
3. THEORY AND HYPOTHESES
Considering that performance differences are attributed to country, industry and firms, our research provides the percentage influence of country-level, industry-level and firm-level heterogeneity in a comparison between developed and developing countries. This Chapter describes the development of hypotheses for both variance components of firm performance and specific country, industry and firm determinants to explain firm performance. Hypotheses are developed to guide our research. Our hypotheses are built on the research of Makino et al. (2004) and Chan et al. (2010). Their results suggest that country effects and industry effects are more important for developing countries, whereas internal factors are more important in developed countries.
3.1 The Importance of Country-level, Industry-level, and Firm-level Heterogeneity in Developed vis-à-vis Developing Countries
This thesis seeks to reveal the relative importance of country, industry, and firm effects as components of variation in firm performance for developed countries in comparison to developing countries. Therefore and as a result, of given arguments, we hypothesise:
H1a: Country-level heterogeneity accounts for more variation in firm performance in developing countries than in developed countries.
H1b: Industry-level heterogeneity accounts for more variation in firm performance in developing countries than in developed countries.
Based on those hypotheses, our research expects that firm effects are relatively more important as a source of variation in firm performance in developed countries than in developing countries. By examining the percentages of influence attributed to country-level, industry-level and firm-level heterogeneity in developed and developing countries, our research tests results of Makino et al. (2004) and Chan et al. (2008).
3.2 Determinants of Firm Performance
Apart from estimating the percentage influence of country, industry and firms effects on variation in firm performance, our research provides meaningful content to country-level, industry-level and firm-level heterogeneity. VCA determines the relative percentages of country, industry, and firm effects on the overall firm performance variability. However, VCA only attributes variation to differences between corporations rather than providing an explanation of firm performance. Therefore, our research continues with describing specific country, industry, and firm determinants to explain firm performance.
3.2.1 Country Determinants
between those countries. Second, related to financing; savings are mainly used for internal financing instead of foreign financing. Third, investors’ preference; investors prefer domestic shares instead of foreign shares.
Hence, while certain decisions and actions from corporations can change industries’ competitive structure, so too can changes in macro-economic conditions affect all corporations and industries within a given country. Two important macro-economic conditions are economic growth and inflation. According to classic economic theory ‘supply creates demand’. This theory indicates that economic growth is a result of higher quantity or quality in production of a given country. Economic growth increases society’ wealth and results in higher demand for products. Higher demand on its turn creates opportunities for corporations to extend their supplies. Therefore, economic growth provides corporations opportunities of growth to improve their performance and profitability (Barney, 1986). However, corporations will not have any opportunities of growth to improve their performances before there is a demand for their products. According to Keynes ‘supply does not create its own demand’ but ‘demand creates its own supply’ (Malabre, 1994). Whether demand creates supply first or supply creates demand is irrelevant for this research, since our research analyses the contribution of economic growth to differences in firm performance. Important is that from an economic view suggested by Adam Smith, supply and demand are guided by an invisible hand to a balanced equilibrium. This theory suggests that economic growth results in higher supply and demand. A study by Demirguc-Kunt and Huizinga (1999) argues that high economic growth increases profitability of corporations for a large number of countries. Economic growth captures expansion or contraction in the market because of higher or lower consumer expenditures. Fluctuation in the market impacts profitability of corporations directly through demand and the increase or decrease in competitive pressures.
In line with our arguments of the previous two hypotheses, the following two hypotheses provide complementary information on country determinants of firm performance in developed countries compared to developing countries:
H2a: Economic growth is a more important determinant of firm performance in developing countries than in developed countries.
H2b: Inflation is a more important determinant of firm performance in developing countries than in developed countries.
We test these hypotheses to analyse whether economic growth and inflation are important determinants of firm performance. This research expects that economic growth and inflation are more important for explaining variation in firm performance in developed countries than in developing countries. Thereby we aim to explain performance differences by country factors.
3.2.2 Industry Determinants
In line with our arguments of the previous hypotheses, the following two hypotheses analyse specific industry determinants to explain variation of firm performance between
industries in developed countries compared to developing countries:
H3a: Industry growth is a more important determinant of firm performance in developing
countries than in developed countries.
H3b: Industry profits are a more important determinant of firm performance in
developing countries than in developed countries.
We test these hypotheses to analyse whether industry sales growth and industry profits are important determinants of firm performance. Our research expects that industry growth and industry profits are more important for explaining variation in firm performance in developing countries than in developed countries. In that way, our research seeks to explain performance differences by industry factors.
3.2.3 Firm Determinants
Mueller (1990) and Amato and Amato (2004) counter several firm-specific characteristics in
Given the suggested high importance of internal factors for developing countries (Makino et al., 2004; Chan et al., 2008; Chan et al., 2010) and academic significance of firm effects on firm performance, firm size and capital intensity are incorporated in the following hypotheses. In line with our arguments of the previous hypotheses, the following two hypotheses analyse
specific firm determinants to explain variation of firm performance between corporations in
developed countries compared to developing countries:
H4a: Firm size is a more important determinant of firm performance in developing
countries than in developed countries.
H4b: Capital intensity is a more important determinant of firm performance in
developing countries than in developed countries.
4. EMPIRICAL STRATEGY
After describing hypotheses that guide our research, this Chapter presents data and method used. Starting with the data to provide descriptions of the dependent and independent variables as well as the sample and sampling design. Further, we discuss the method to calculate the percentages of which country, industry, and firm attribute to variation of firm performance. In addition, the process of explaining firm performance by country, industry, and firm determinants is described.
4.1 Data
4.1.1 Dependent Variable: Firm Performance
An important issue for our research is how to measure firms’ performance. For the dependent variable of firm performance, our research follows other researchers by using Return on Assets (ROA) (Rumelt, 1991; McGahan and Porter, 1997; Hawawini, Subramanian and Verdin, 2003; Makino et al., 2004). ROA is calculated as net income to total assets, which indicates how successful a corporation is at generating income from their invested capital and thereby capturing both profitability and efficiency of assets engaged. Data regarding ROA are collected from Orbis that provides financial data for corporations on a global scale.
This research is limited to a cross sectional analysis of last five years, since most developing countries are only experiencing a rapid economic growth during the last decade. The period of five years is chosen to prevent the problem of year-to-year differences (Rumelt, 1991) and to include the total period of a business cycle phase (Rumelt, 1991; McGahan and Porter, 1999). Although, Chang and Singh (1997) suggest that the results of previous research have to be interpreted within the context of their research samples, we compare some statistics of ours with previously found statistics.
Table 1. Descriptive Statistics for Key Variables
Dependent Variable Minimum Maximum Mean Std. Deviation Skewness Kurtosis
ROA (%)
Developed Countries -73.98 90.38 5.244 9.439 0.577 8.431
Developing Countries -65.92 85.95 7.176 13.78 0.878 3.220
Mean -69.95 88.17 6.210 11.61 0.728 5.826
Independent Variables Minimum Maximum Mean Std. Deviation Skewness Kurtosis
Economic Growth (%) Developed Countries (n=27) -0.437 6.195 1.885 1.416 1.193 2.116 Developing Countries (n=14) 0.640 11.38 3.964 2.873 1.347 2.323 Inflation (%) Developed Countries (n=27) 0.011 8.084 2.415 1.444 2.498 9.159 Developing Countries (n=14) 2.639 15.32 6.287 3.652 1.265 1.521 Industry Growth (%) Developed Countries (n=634) -25.78 157.0 9.221 12.47 4.645 40.29 Developing Countries (n=521) -24.21 14,296 126.5 775.6 14.18 232.7 Industry Profits (%) Developed Countries (n=634) -48.71 47.11 4.149 6.612 0.603 16.19 Developing Countries (n=521) -27.54 49.26 8.098 9.616 0.812 2.482
Firm Size (Million US$)
Further, Skewness and Kurtosis Statistics indicate the shape of distribution. Results show a deviation from a normal distribution to a leptokurtic distribution, which suggests a sharp peak around the mean. A leptokurtic distribution reveals that corporations group their performance around the mean value, while at the same time there are more deviations than in any normal distribution. Moreover, the positive values for Skewness Statistics indicate skewed distribution where the majority of corporations have higher values than the mean.
4.1.2 Independent Variables
The independent variables represent country, industry, and firm determinants of firm performance. Determinants for country effects are based on macro-economic conditions. First, economic growth is measured in terms of an increase in the size of a nation’ economy by the percentage change of Gross Domestic Product (GDP). Second, inflation measured by the percentage change of average consumption prices. Data concerning macro-economic conditions are collected from the database of Worldbank and are an average of the period 2005-2009. Table 1 presents statistics concerning country, industry, and firm determinants.
As a final source of variation, firm determinants that are included in our empirical analysis are firm size and capital intensity. First, firm size is measured by the log value of total assets (Lee, 2009). Second, capital intensity is measured by the ratio of total assets over sales
(Oustapassidis, 1997; Lee, 2009). As for firm performance, data concerning these firm-specific characteristics are collected from Orbis. Remarkably, the minimum value and mean value of capital intensity in developed countries is negative because of extreme outliers for developing countries. A possible explanation of negative capital intensity ratios are negative sales revenues found for corporations in developing countries. Especially in the beginning, corporations may have relatively low sales. The actual sales revenues of a corporation can be less than the amount of refunds that are paid to customers. Negative sales revenues are a result of actual sales that are less than for example the amount of refunds. As a result of the relative low mean value for developed countries, the mean value for developing countries is higher than for developed countries. This outcome suggests that industries in developing countries require relatively more amounts of assets to operate than industries in developed countries. This finding is however not consistent with the results of firm size, which suggests that corporations in developed countries have more assets than in developing countries. All descriptive statistics are an average of the period 2005-2009.
4.1.3 Sample and Sampling Design
Since this research includes predictors at three different levels (country, industry and firm), selecting a sample will be a delicate issue as these three different levels have different degrees of freedom, which indicates that a certain number of firm observations does not provide this research with the degrees of freedom at the industry or country level.
system. The IMF World Economic Outlook 2010 presents economic developments at global level in individual countries and classifies 33 countries as developed and 150 countries as developing.
Data for this research is collected from the database of Orbis, which contains 82,458,793 corporations from 203 countries. The sample for this research contains data of every corporation from the list of developed and developing countries stated by IMF World Outlook 2010 available from Orbis in the period 2005-2009. All records are screened based on previous research to generate a database for cross-sectional analysis (e.g. Rumelt, 1991; Spanos and Lioukas, 2001; Galbreath and Galvin, 2008; Goldszmidt et al., 2011). First, this research includes only corporations for which data are available on all the relevant variables. Second, to moderate against the potential bias of single year performance results and a minimum operating structure, this research will follow Spanos and Lioukas (2001) and Galbreath and Galvin (2008) in the selection process that only corporations from three years and older and consisting of 20 or more employees are included. By following previous research, (Rumelt, 1991; McGahan and Porter, 1997, 2005) relative small corporations are excluded. These exclusions guarantee that our analyses are not distorted by relative small corporations that do not represent the mainstream of economic activity (McGahan and Victer, 2010). Moreover, banking, insurance, and financial corporations and institutions are excluded because those corporations will differ in accounting definitions from financial corporations and would make it difficult for comparison with non-financial corporations. Further, countries that have only one or two industries are excluded from this research.Furthermore, from a perspective on construct validity, all observations with a ROA below -100% or above 100% have been eliminated as well. The final sample consists of 47,111 corporations from 670 specific industries in 41 different countries of which 11,407 corporations are from 521 industries in 14 developing countries and 35,704 corporations from 634 industries in 27 developed countries. More specific descriptive statistics concerning our sample can be found in the Appendix A.1. These sample criteria generate a database of sufficient number of corporations from developed and developing countries to analyse the percentage influence of country-level, industry-level and firm-level heterogeneity on variation in firm performance.
research in terms of scope in our analysis of decomposing firm performance (i.e. broad coverage of developed and developing countries as well as industries, range of years and corporations included).
4.2 Method
Previous research applies variance components analysis to estimate the relative importance of different factors in accounting for variation in firm performance. Following previous research (Rumelt, 1991; McGahan and Porter, 1997; Makino et al., 2004), the first objective of our research is to categorize variation of firm performance by country-level, industry-level and firm-level heterogeneity. Since the research question contains both country, industry and firm effects as sources of variation in firm performance as well as more specific country, industry and firm determinants, our research consists of two parts. The first part of our research focuses on VCA to estimate the relative importance of country-level, industry-level and firm-level heterogeneity for firm performance, in both developed and developing countries. VCA attempts to decompose variation of ROA into components that represents the contribution of each random effect causing the total variance in firm performance. In our research corporations (level 1) are nested in industries (level 2), which are nested in countries (level 3), each level contributing to the total variance in ROA. The following model is used for estimating this relationship (Equation 1):
Rijk = α + βi+ βj + βij + εijk (1)
given country. Random industry effects reflect the influence of industry structure on all corporations in a certain industry. The random firm effect εijk is the residual not explained by country, industry, and country-industry interaction effects. Regarding the relative importance of country, industry, and firm effects in this specified model, proponents of IO would assume relatively high βj, whilst in line with RBV; random firm effect εijk should take over. This specification helps us understand how variation in firm performance arises because of variation in one of the independent variables, while the other independent variables are held fixed. Our statistical research will test hypotheses in a quantitative manner and generalization of findings will be presented based on the representativeness of the research sample and the validity of the research design. The statistical research will consist of variance components analysis with Maximum Likelihood Method (MLE), which provides estimates for parameters specified in Equation 1.
Rather than estimating the usefulness of parameters, this research estimates the variance explained by independent variables. Given that our independent influences are random, estimating regression parameters would contain less information than estimating variances in firm performance. Therefore, a variance components equation estimates the input of independent variables on the variation in firm performance as follows (Equation 2):
δ ROA = α + δi + δj + δij + δε (2)
performance into components that capture country (δi), industry (δj), country-industry interaction (δij) and firm effects (δε). Random modelling provides our research the advantage of hypothesizing the relative importance of country, industry, country-industry interaction, and firm heterogeneity in
general without being interested in specific countries, industries, or corporations. Difference
between this research and previous ones is that this variance decomposition model embraces country, industry, country-industry interaction as well as firm effects.
After establishing that such differences exist, our research continues to the second part of the question. This second part examines specific country, industry, and firm determinants as explanation of firm performance. To capture the extent to which country, industry and firm determinants explain firm performance; covariates are added to Equations 1 and 2. All determinants are added separately to existing models as covariates. Results of VCA without determinants as covariates and results of VCA with determinants as covariates provide us with results of variation, which are only explained by the specific determinant. The R2 presents the percentage of variation in firm performance differences that are explained by the specific country, industry, and firm determinant. Based on the R2 values our research determines whether firm performance can be explained by country, industry, and firm factors.
5. EMPIRICAL RESULTS
In this Chapter, results are provided in three steps. Foremost, variance component analysis is conducted both for the relative importance of country-level, industry-level and firm-level heterogeneity as for specific country, industry and firm determinants to explain variation in firm performance. As a result, the outcomes of variance component analysis are presented in separate subsections. Further, robustness checks are added to ensure that the presented results of our research are not sensible for chosen estimation method and sample size.
5.1 Variance Components of Firm Performance
To decompose performance differences between corporations, we analyse variance components of firm performance, in both developed and developing countries. Our VCA provides the percentage relevance of country-level, industry-level, country-industry interaction, and firm-level heterogeneity. The results of Equations 1 and 2 are presented in Table 2 and indicate the variance of firm performance.
Table 2. Variance Components of Mean Firm Performance
Developed Countries Developing Countries
Component Estimates % Estimates %
Between-Country Variation (δi) 5.808 6.240 9.366 5.066
Between-Industry Variation (δj) 4.963 5.332 12.58 6.802
Country-Industry Interaction (δij) 14.42 15.49 13.80 7.467
Firm Heterogeneity (δk) 67.89 72.94 149.1 80.67
Total 93.07 100 184.9 100
Notes: Dependent variable Mean ROA for 2005-2009.
firm effects attribute 72.94% to the total variation in firm performance in developed countries and 80.67% in developing countries. These results indicate that firm effects account for the largest share in performance differences across corporations, while other effects are smaller. This finding supports the RBV theory that argues that firm’ resources and capabilities are the most important determinants of firm performance. In comparison, corresponding figures for industry-level heterogeneity, country-industry interaction effect and for country-industry-level heterogeneity are as followed. Between industry variation accounts for 5.332% in developed and 6.802% in developing countries. Those results of industry heterogeneity are lower than the 18.68% found by McGahan and Porter (1997), 8.32% by Rumelt and 19.59% by Schmalensee (1985). Further, between-country variation attributes for 6.240% of firm performance differences in developed countries and for 5.332% in developing countries. The country-industry interaction effect of 15.49% of total variation in firm performance in developed countries and 7.467% in developing countries is relatively stronger than the country and industry effects alone.
Therefore, when we decompose and describe firm performance differentials among corporations, firm-level heterogeneity accounts for more variation in firm performance than country-level heterogeneity, industry-level heterogeneity, and industry-country interaction effects. Due to high estimates for between firms variance both in developed and in developing countries, country-level and industry-level heterogeneity account for less variation. Comparing the percentages of country-level, industry-level, country-industry interaction effects and firm-level heterogeneity, we notice that country factors account for a larger share of total variation in firm performance in developed countries compared to developing countries whereas industry factors account for a larger share of total variation in firm performance in developing countries.
results are consistent with Hypothesis 1a and in conflict with Hypothesis 1b. Because of these outcomes, our results for industry effects question the findings of Makino et al. (2004) who suggest that external factors are a relative more important source of variation in firm performance for developing countries than for developed countries.
To summarize, our results suggest that internal assets and competencies are central to competitive advantage regardless of country and specific industry characteristics. Although country and industry influence the environment in which a corporation performs, those factors do not seem to provide an explanation of corporations’ competitive advantage. Any corporation in the same environment would benefit or disadvantage from those external factors. Despite those external factors, internal competencies may provide a corporation with different opportunities than its competitors in the same country and/or industry (Hawawini et al., 2003). Theoretically, corporations should focus on effectively leveraging their resources instead of controlling and manipulating their external environment. This outcome is consistent with previous research by Roquebert et al. (1996), McGahan and Porter (2002), Ruefli and Wiggins (2003) and Hawawini et al. (2003) and supports the RBV of firms, which argues that idiosyncratic, immobile resources and capabilities are the source of competitive advantage and determine firm performance (Wernerfelt, 1984; Barney, 1991; Nelson, 1991).
5.2 Determinants of Variation in Firm Performance
5.2.1 Country Determinants
Country factors that as discussed might explain variance of firm performance between countries are economic growth and inflation. Table 3 presents results of VCA and the influence of economic growth and inflation as possible explanation for between country variation and total variation in performance. The table presents results of VCA without determinants as covariates, with determinants as covariates and results of variation, which are only explained by the determinant. The R2 presents the percentage of variation in firm performance differences that are explained by the specific determinants. Based on the R2 values we can determine whether firm performance can be explained by country factors. The results from Table 3 show that country-level heterogeneity and total variance in developed and developing countries have been reduced by adding economic growth to our model. A reduction of country heterogeneity of 14.20% in developing countries and 2.29% in developed countries suggests that economic growth explains part of country heterogeneity and more so for developing than developed countries. Further, as expected, economic growth provides not an explanation of between corporations variation, solely for between country variations. Moreover, Pearson Correlation confirms that there is a positive correlation between economic growth and firm performance for both developed and developing countries, 0.194 (n=27) and 0.221 (n=14) respectively. The correlation is stronger for developing countries, which suggest that economic growth is a more important determinant of firm performance in developing countries than in developed countries.
Table 3. Country Determinants
Developed Countries Developing Countries
Empty Model
Product Difference explained by Determinant
Empty Model
Product Difference explained by Determinant Industry Determinants Estimates1 Estimates2 Estimates3 R2 Estimates1 Estimates2 Estimates3 R2
Economic Growth
Between Country Variance (δi) 7.904 7.723 0.181 2.290 9.737 8.354 1.383 14.20
Between Firm Variance (δk) 78.58 78.58 0.000 0.000 177.0 177.0 0.000 0.000
Total 86.48 86.30 0.181 0.209 186.7 185.3 1.383 0.741
Inflation
Between Country Variance (δi) 7.904 7.912 0.008 0.101 9.737 6.995 2.742 28.16
Between Firm Variance (δk) 78.58 78.58 0.000 0.000 177.0 177.0 0.000 0.000
Total 86.48 86.49 0.008 0.009 186.7 184.0 2.742 1.468
Notes: 1 Results of Variance Component Analysis without Determinants 2 Results of Variance Component Analysis with Determinants as Covariates
3 Estimates and R2 explained by specific country determinants Dependent Variable: Mean ROA
Looking back at our Hypotheses, Hypotheses 2a and 2b were focused on the relative importance of country determinants in developed compared to developing countries. Hypotheses 2a assumed that economic growth is a more important determinant of firm performance in developing countries than in developed countries. In addition, Hypothesis 2b states that inflation is a more important determinant of firm performance in developing countries than in developed countries. Our results of country determinants economic growth and inflation show that economic growth and inflation are more important for explaining firm performance in developing countries than developed countries. Those results are consistent with our hypotheses 2a and 2b.
5.2.2 Industry Determinants
Industry determinants that, as discussed before, might explain between-industry variation in firm performance are industry sales growth and industry profits. Table 4 presents results of VCA without determinants as covariates, with determinants as covariates and results of variation, which are only explained by the determinant. The R2 presents the percentage of variation in firm performance differences that are explained by the specific industry determinants. In this way, our research can find out whether firm performance can be explained by industry factors.
Table 4 shows that the estimates of total variance in firm performance are reduced by 0.66 in developed countries and only by 0.08 in developing countries because of lower industry-level heterogeneity. More importantly, our results show that industry growth explains 5.061% of between industry variation and 0.715% of the total variation in firm performance in developed countries. In comparison, industry growths provide an explanation of 0.313% for industry effects and 0.042% for the total variation of firm performance in developing countries. Furthermore, Pearson Correlation between Mean ROA and industry growth are 0.058 (n=634) in developed countries and -0.170 (n=521) in developing countries. Those findings indicate that industry sales growth is a more important determinant for explaining firm performance in developed countries than in developing countries.
that R2 for between-industry variation is 97.53%, 1.695% for firm heterogeneity, and 14.55% for the total variation in firm performance in developing countries. In addition, the results of R2 for developed countries are 0.000%, which indicates that industry profits do not provide an explanation of variance in developed countries. Moreover, Pearson Correlation for industry profits with Mean ROA are 0.000 (n=634) for developed countries and 0.376 (n=521) for developing countries, which shows that there is a stronger positive correlation between industry profits and firm performance for developing countries. This confirms our findings that industry profits are a more important determinant for firm performance in developing countries than in developed countries.
Our Hypotheses 3a and 3b were focused on the importance of industry determinants of firm performance in developed compared to developing countries. Hypotheses 3a said that
industry growth is a more important determinant of firm performance in developing countries
Table 4. Industry Determinants
Developed Countries Developing Countries
Empty Model
Product Difference explained by Determinant
Empty Model
Product Difference explained by Determinant Industry Determinants Estimates1 Estimates2 Estimates3 R2 Estimates1 Estimates2 Estimates3 R2
Industry Sales growth
Between Industry Variance (δi) 13.04 12.38 0.660 5.061 25.59 25.51 0.080 0.313
Between Firm Variance (δk) 79.23 79.23 0.000 0.000 165.2 165.2 0.000 0.000
Total 92.27 91.61 0.660 0.715 190.8 190.7 0.080 0.042
Industry Profits
Between Industry Variance (δi) 13.04 13.04 0.000 0.000 25.59 0.626 24.96 97.53
Between Firm Variance (δk) 79.23 79.23 0.000 0.000 165.2 162.4 2.800 1.695
Total 92.27 92.27 0.000 0.000 190.8 163.0 27.76 14.55
Notes: 1 Results of Variance Components Analysis without Determinants 2 Results of Variance Components Analysis with Determinants as Covariates
5.2.3 Firm Determinants
Besides country and industry determinants of firm performance, performance differences between corporations can be explained by corporations’ idiosyncratic resources and capabilities. Two firm determinants that might explain variance of firm performance between corporations are firm size and capital intensity. Table 5 presents results of VCA without determinants as covariates, with determinants as covariates and results of variation, which are only explained by the determinant. The R2 presents the percentage of firm performance differences that are explained by firm determinants. Thereby, our research can determine whether firm performance can be explained by firm factors.
Table 5 presents the results of firm size as possible explanation for firm-level heterogeneity. The results from Table 5 reveal that firm size does not explain firm-level heterogeneity in developing countries. Furthermore, by including country effects, industry effects, country-industry interaction effects, our results do not only show how much firm size explains firm-level heterogeneity. Results display as well how much firm size explains or disturbs country effects, country-industry interaction effects, and industry effects because of systematic variation at higher levels. Moreover, Pearson Correlation between firm size and Mean ROA are 0.020 (n=35,704) for developed and -0.020 (n=11,407) for developing countries. Those correlation results illustrate that there is a small positive correlation between firm size and firm performance for developed countries and a small negative correlation for developing countries. As a result, although firm size only explains a marginal part of variation, our results point out that firm size is a more important determinant of firm performance in developed countries than in developing countries.
heterogeneity in developed countries. Because of systematic variation on higher levels, capital intensity is more important for explaining country and industry components.
However, Pearson Correlation shows that there is a positive correlation between capital intensity and firm performance for developed countries of 0.100 (n=35,704) and a negative correlation for developing countries of -0.110 (n=11,407). A negative correlation is in line with previous research of Scott (1982), Ravenscraft (1983), Kwoka and Ravenscraft (1986) and Oustapassidis (1997), who found a negative effect of capital-to-sales ratio on firm profitability. Porter (1976) argues that the negative relationship between capital intensity and firm profitability can be explained by huge investments. He suggests that huge capital investments hinder corporations to close down in case of low profits. Since closing down the corporation might only result in higher losses. In contrast, a positive correlation for developed countries is more in line with the majority of previous research on capital intensity (Hay and Morris, 1991), that suggest that higher levels of capital intensity leads to greater profit margins of the firm. Since capital intensity contributes to a quick production process and reduced costs because of reduction in waste and improved quality. Kwoka and Ravenscraft (1986) confirm that capital intensity influences firms’ profitability through the ability to pursue strategic advantages. Nonetheless, our results show that capital intensity is a more important determinant for explaining variation in firm performance in developed countries than in developing.
Table 5. Firm Determinants
Developed Countries Developing Countries
Empty Model
Product Difference explained by Determinant
Empty Model
Product Difference explained by Determinant Firm Determinants Estimates1 Estimates2 Estimates3 R2 Estimates1 Estimates2 Estimates3 R2
Firm Size
Between-Country Variation (δi) 5.808 5.907 0.099 1.705 9.366 9.342 0.024 0.256 Between-Industry Variation (δj) 4.963 5.007 0.044 0.887 12.58 12.61 0.030 0.238 Country-Industry Interaction (δij) 14.42 14.42 0.000 0.000 13.80 13.77 0.030 0.217 Between Firm Variance (δk) 67.89 67.79 0.100 0.147 149.1 149.1 0.000 0.000
Total 93.08 93.12 0.043 0.046 184.9 184.8 0.024 0.013
Capital Intensity
Between-Country Variation (δi) 5.808 5.800 0.008 0.137 9.366 9.343 0.023 0.246 Between-Industry Variation (δj) 4.963 4.947 0.016 0.322 12.58 12.57 0.010 0.079 Country-Industry Interaction (δij) 14.42 14.41 0.010 0.069 13.80 13.81 0.010 0.072 Between Firm Variance (δk) 67.89 67.88 0.010 0.015 149.1 149.1 0.000 0.000
Total 93.08 93.04 0.000 0.047 184.9 184.8 0.023 0.012
Notes: 1 Results of Variance Components Analysis without Determinants 2 Results of Variance Components Analysis with Determinants as Covariates
3 Estimates and R2 explained by specific firm determinants
5.3 Robustness
In this subchapter, we report two robustness checks to validate our results. First, we examine whether the difference in estimation method for variance decomposition affects our results. Second, we assess the possible impact of size differences between corporations, which have Mean ROA of between -100% and 100% with a subsample of corporations, which have Mean ROA values between -50% and 50%.
5.3.1 Results for ANOVA
An important question is whether the above results are robust. To analyse the robustness of our results, robustness check is executed with an alternative estimation method. One alternative estimation method for variance components analysis is ANOVA, which has been used by previous research (Rumelt, 1991; McGahan and Porter, 1997). Previous variance components analyses are conducted with Maximum Likelihood Estimation, therefore the robustness check is executed with ANOVA. Table 6 presents the results of component variance conducted by ANOVA for developed and developing countries.
Table 6. Robustness for Variance Components
Developed Countries Developing Countries
Component Estimates % Estimates %
Between-Country Variation (δi) 14.15 15.21 19.77 10.01
Between-Industry Variation (δj) 0.120 0.001 14.84 7.513
Country-Industry Interaction (δij) 11.97 12.86 13.68 6.925
Firm Heterogeneity (δk) 66.93 71.92 149.3 75.55
Total 93.05 100 197.6 100
Notes: Dependent Variable: Mean ROA
for 71.92% of performance differences in developed countries and 75.55% in developing countries compared to 72.94% and 80.67% respectively found in Table 2. Further, corresponding statistics for industry heterogeneity are 0.001% in developed countries and 7.513% in developing countries compared to 5.332% in developed and 6.802% in developing countries discovered in Table 2. Remarkably, because of another estimation method, the relative importance of industry is almost reduced to zero for between-industry variation. Instead of industry heterogeneity, variation is attributed to industry interaction effects. Results show that the country-industry interaction effect attributes 6.925% to variation in firm performance in developed countries compared to 15.49% found in Table 2. For developing countries, the country-industry interaction effect accounts for 12.86% in developing countries compared to 7.467% discovered in Table 2. Additional, country heterogeneity attributes for 15.21% of firm performance differences in developed countries and 10.01% for in developing countries. The results from Table 6 concerning the relative importance of country-level heterogeneity, is twice as high as the results from Table 2 of MLE, which presented 6.240% in developed and 5.332% in developing countries
Same results are found for country, industry, and firm determinants as explanation for variation in firm performance. The results for country, industry, and firm determinants can be found in Appendix B. Our results on country determinants suggest that economic growth and inflation are more important determinants of firm performance in developing countries than in developed countries. Further, industry growth is more important for developed countries whereas industry profits are for developing countries. However, our robustness results are not consistent with our earlier finding of firm determinants that suggest that firm size and capital intensity are more important determinants for improving firm performance in developed countries. Table B.3 in Appendix B shows that firm size and capital intensity account for more variation in developing countries than developed countries. However, percentages of firm size and capital intensity are relatively small and thereby provide relatively little explanation of firm performance.
country-level, industry-level and firm-level heterogeneity are untouched, we assume that our results are not affected by small deviations and techniques.
5.3.2 Results for Different Sample
Besides testing our variance decomposition with ANOVA, robustness is checked for sample size as well. Mainly for CVA, treatment of outliers is an important concern. Research by Hawawini et al. (2003) show that excluding outliers from VCA leads to an increase of relative importance of industry effects over firm effects. This suggests that the impact of outliers is present in variance decomposition of firm performance. To account for the influence of outliers, our data is tested for abnormal values of firm performance. Therefore, instead of including all corporations with values for Mean ROA between -100% and 100%, our robustness check analyses variance decomposition for a subsample with corporations that have values for Mean ROA between -50% and 50%.
Table 7 presents our results for variance components analysis where variation in firm performance is attributed to country-level, industry-level, country-industry interaction effects, and firm-level heterogeneity.
Table 7. Robustness for Outliers in Variance Components
Developed Countries Developing Countries
Component Estimates % Estimates %
Between-Country Variation (δi) 4.769 5.924 7.731 5.058
Between-Industry Variation (δj) 4.348 5.401 9.970 6.523
Country-Industry Interaction (δij) 11.27 13.99 11.75 7.687
Firm Heterogeneity (δk) 60.12 74.68 123.4 80.73
Total 80.51 100 152.9 100
firm performance in developed countries. In addition, industry-level heterogeneity and firm-level heterogeneity account for higher percentages in developing countries.
Tables 8, 9, and 10 report results of our variance decomposition with specific country, industry, and firm determinants. Table 8 presents robustness of country determinants (economic growth and inflation). Our results show that economic growth explains 13.13% of between country variation and 0.623% in developing countries and only 1.903% and 0.146% in developed countries. Further, inflation explains 29.48% of between industry variation and 1.407% of the total variation in firm performance for developed countries. Those results are consistent with results of our main analysis that suggests that economic growth and inflation are more important for explaining firm performance in developing countries than in developed countries.
Results for industry sales growth shows that this determinant explains 5.084% of between industry variation and 0.708% of the total variation in firm performance for developed countries whereas it only explains 0.333% and 0.044% respectively for developing countries. Furthermore, industry profits explain 96.43% of industry-level heterogeneity and 1.758% of total variation for developing countries whereas it does not explain developed countries. Again, results are consistent with our earlier findings that industry growth is more important for explaining variation in firm performance in developed countries whilst industry profits are more important for explaining variation in firm performance in developing countries.
Looking at the results for firm size measured by total assets, Table 10 shows that firm size explains 0.15% of firm-level heterogeneity in developed countries whereas firm size has no contribution in explaining variance in developing countries. Further, capital intensity does not explain firm heterogeneity in developed and developing countries. In contrast with our earlier findings, our robustness check shows that firm size and capital intensity are more important determinants of firm performance in developing countries than in developed countries. The relative importance of firm size is not explained through firm heterogeneity for both factors but through their influence on country and industry heterogeneity. This is a result of systematic variation at higher levels such as country-level and industry-level.
