- Email: [email protected]
Global competition and global markets: some empirical results
International Business Review 13 (2004) 401–416 www.elsevier.com/locate/ibusrev
Global competition and global markets: some empirical results Louis H. Amato a, Ronald P. Wilder b, a
b
Department of Economics, Belk College of Business Administration, University of North Carolina at Charlotte, Charlotte, NC 28223, USA Department of Economics, Darla Moore School of Business, University of South Carolina, Columbia, SC 29208, USA
Abstract The purpose of this paper is to develop and test a model that relates the profit rates of multinational firms to market share, other firm characteristics, and industry effects where the relevant geographic markets are defined at the global level. We test the proposition that market share and industry fixed effects can explain profitability variation for a sample of multinational firms. The most important finding is that market share has a positive and statistically significant effect on profit rates for a sample of Fortune Global 500 firms. Moreover, the Fortune Global 500 firms exhibit strong industry effects. While there are a number of explanations for a positive market share effect and strong industry effects, a common interpretation is the presence of rivalry among these firms. # 2004 Elsevier Ltd. All rights reserved. Keywords: Global markets; Industrial organization; Market structure; Firm strategy and market performance; Multinational firms; Profitability; Oligopoly
1. Introduction The increasing importance of multinational firms and international trade in manufactured goods leads naturally to the question of whether the relevant geographic market for many manufacturing industries is now global rather than national. With some notable exceptions, much of the previous empirical work building on the structure– conduct–performance paradigm has made use of data that follow national geographic market definitions. This paper is motivated by
Corresponding author. Tel.: +1-803-777-6955; fax: +1-803-777-6876. E-mail address: [email protected] (R.P. Wilder).
0969-5931/$ - see front matter # 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.ibusrev.2003.09.007
402
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
our interest in whether it is meaningful to define manufacturing markets globally. Hout, Porter, and Rudden (1998) suggest that multinational firms compete in two types of industries: multi-domestic industries in which firms pursue independent strategies in each of a number of foreign markets, and global industries in which firms compete with other multinational firms in the world market at large. Our hypothesis in this work is that global industries are sufficiently important so that competitive interaction among firms takes place at this level of industry definition. Richard Caves (1989: p. 1244) summarizes the need for research regarding the consequences of global competition as follows: ‘‘Discussions in the field of business strategy have focused on ‘global competition’ in which international rivals pursue strategies that treat the world’s sub-markets as interdependent. Yet systematic empirical research on international oligopolistic behavior is quite limited.’’ The transaction cost explanation for the existence of multinational firms suggests that such firms will tend to be present in industries with concentrated sellers (Caves, 1996, Chapter 4). Global oligopoly is thus a phenomenon likely to coincide with the growth of multinational enterprise. The potential importance of global competition is also suggested by the growth in international trade in manufactured goods. While merchandise trade relative to GDP has increased for most industrial countries during the past three decades, the trend is dampened by the increasing importance of services as a component of GDP. As pointed out by Feenstra (1998: Table 2), merchandise trade relative to merchandise value-added has increased sharply for most industrial countries during the recent past. This increase is due in part to falling trade barriers and lower transportation costs, but is also related to what Feenstra calls the ‘‘disintegration of production’’, with substantial importing of intermediate inputs. The importance of global vertical production chains also suggests the importance of the study of global competition. The purpose of this paper is to develop and test a model that relates the profit rates of multinational firms to firm characteristics and industry effects, where markets are defined at the global level. The research addresses important public policy issues related to the effectiveness of global competition in restraining oligopolistic pricing behavior of firms that may face little price competition from domestic rivals. Moreover, the findings are relevant to firms concerned about the strategic actions of rivals operating on a global scale. The paper utilizes a methodology similar to that used by Schmalensee (1985), developed further by Rumelt (1991) and Brush, Bromiley, and Hendrickx (1999), which relates profit rates to individual firm characteristics and industry fixed effects. Our data set of international firms was gathered from the Global Fortune 500 for the years 1989–1992. We know of no prior research regarding global competition that employs data similar to those contained in the Global 500. We note, however, that prior studies using US data have consistently found that industry effects are important in explaining cross sectional variation in profit rates. The two major findings reported in this paper are: (1) firm market share in the total global market
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
403
has a positive and statistically significant effect on firm profitability; and (2) global industry fixed effects are also important in explaining firm profitability, a finding that is in keeping with the results of previous studies for national markets. These findings support the hypothesis that multinational firms compete in global industries rather than operating exclusively as multi-domestic firms. 2. Previous works A brief digest of previous works is shown in Table 1. The previous research on firm effects relative to industry effects in explaining profitability began with papers by Ravenscraft (1983) and Schmalensee (1985) making use of national data from the US Federal Trade Commission (FTC). Using FTC Line of Business data for 1975, Schmalensee finds that firm effects do not account for cross-sectional variation in profitability, market share effects account for a negligible proportion, and industry effects account for approximately 20% of profit variation. Schmalensee’s now widely familiar finding that industry effects dominate firm effects has been Table 1 Summary of selected items of previous research on determinants of profitability Effects
Studies
Dataa
Findings
Firm
Scott and Pascoe (1986)
D
Cubbin and Geroski (1987)
D
Wernerfelt and Montgomery (1988) Brush et al. (1999)
D
Firm effects are important; firm and industry effects overlap Persistent profitability differences among firms in the same industry in a dynamic context Firm effects appear as focus effects
D
Firm effects are stronger than industry effects
Business unit
Rumelt (1991)
D
Strong business unit effects in longitudinal panel data; industry effects are important in cross sections
Market share
Ravenscraft (1983)
D
Market share effects dominate industry concentration effects
Industry
Schmalensee (1985)
D
Amato and Wilder (1990)
D
Industry effects dominate firm and market share effects Industry effects dominate firm size effects
Co (2001)
I
Yamawaki et al. (1989)
I
Global competition
a
Imports and foreign direct investment provide competitive discipline European Union seller concentration has a greater effect than domestic market concentration
Data type: D, domestic data from a single country; I, international data.
404
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
offered as evidence that the industry is an appropriate unit of observation for research and policy making in industrial economics. Subsequent studies by Scott and Pascoe (1986), Cubbin and Geroski (1987), Wernerfelt and Montgomery (1988) and Amato and Wilder (1990) provide at least partial support for Schmalensee’s original finding, although neither Scott and Pascoe nor Cubbin and Geroski found industry effects to be as dominant as those revealed in Schmalensee’s original article. More recent empirical work regarding the relative importance of firm and industry effects is found in the strategic management literature, and provides mixed results. Rumelt (1991) employed a variance components model to distinguish between stable and fluctuating firm and industry effects. Rumelt found small or no corporate effects, stable but small industry effects, and a large yet stable business unit effect. While Rumelt’s findings do not support the dominance of industry effects revealed in previous studies, he concludes that industry effects are important in explaining cross-sectional variation in profit rates. Other works on this theme in the strategic management literature include Powell (1996), and Brush et al., (1999). These studies, along with others that they survey, find that line of business effects, firm effects and industry effects are each important, with the relative importance varying across studies. In all cases, industries are defined to include only domestic firms. Another research theme examines the effects of global competition on industries in individual nations. Co (2001) finds a tendency for imports as well as foreign direct investment to provide competitive discipline for domestic industries in the United States. Levinsohn (1993) reports a similar result for imports in Turkey. In a study of paired Japanese and US four digit industries, Yamawaki (1986) finds that the profit margins of Japanese manufacturers increase as entry barriers such as scale economies and product differentiation increase in the US markets that the Japanese exporters serve. Pugel (1980) finds that the price–cost margin on export sales rises with increased domestic concentration, while Yamawaki, Sleuwaegen, and Weiss (1989) conclude that concentration across the entire European Union has greater influence on price–cost margins than concentration within a producer’s own nation. While prior research offers evidence regarding the importance of competition from both domestic rivals and from imports in explaining the profitability of firms and industries at the national or customs union area level, there is virtually no prior research examining the importance of global competition, broadly defined, in determining the profitability of multinational firms. The present work provides an initial examination of competition at the global industry level.
3. Conceptual framework and model specification There are two main traditions for analyzing the impact of industry competition on the profitability of firms. The differential collusion hypothesis is that collusion is likely to be more effective with higher seller concentration and that firm profitabil-
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
405
ity tends to be higher in industries with higher seller concentration. The revisionist tradition, following Demsetz (1973), is that more efficient firms gain higher market shares and earn higher profit rates. This efficiency effect may also increase seller concentration, but the driving force behind higher firm profits under the Demsetz view is efficiency rather than collusion. Although our data set does not allow the testing of the relative importance of these two alternative hypotheses, both suggest the importance of firm market share and industry characteristics in determining firm profitability. The relationship between profitability and market share is developed following Schmalensee (1987), as generalized by Martin (1993: pp. 494–496). This approach begins with the price–cost margin for an oligopoly firm that makes the Cournot assumption about rivals’ quantity adjustment: ð p ci Þ s i ¼ p e
ð1Þ
where p is price, ci is marginal cost, si is market share and e is the priceelasticity of the industry demand function. The firm’s accounting rate of return on capital (PA) can be shown to be: PA ¼ q þ
pqi si pk Ki e
ð2Þ
where q is the economy-wide normal rate of return on capital, qi is the firm output, pk is the price of physical capital and Ki is the quantity of physical capital. This relationship forms the basis for the estimating equation used in hypothesis testing in this paper. We take q to be the global normal rate of return on capital in our sample of multinational firms; it is thus reflected in the intercept term of the estimating equations shown below. Moreover, the Global 500 data used for this study allow for the direct measurement of market share at the global level. While our data do not permit the direct measurement of pk or Ki, the capital intensity variable appearing on the right hand side of our model is intended to capture inter-firm differences in both the price and physical quantity of capital. Finally, we know of no data source that provides demand elasticity measures for firms appearing in the Global 500. However, by including industry fixed effects as regressors, our methodology allows for various inter-industry factors that influence profitability, including differences in the market demand elasticity. A verbally expressed model, building on the relationship developed in (1) and (2) above, relating return on assets to market share, capital intensity, industry fixed effects and other variables is: Return on Assets ¼ f ðMarket Share; Firm Size; Capital Intensity; Industry Effects; Country Effects; Leverage; Trade IntensityÞ ð3Þ
406
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
The empirical specification to be estimated is found in Eq. (4): X ROAijt ¼ B0 þ B1 SHAREijt þ B2 SIZEit þ B3 CAPINit þ B4þj INDj X X X þ B32þt Yeart þ B36þk Countryk þ B51þk Share Countryijt þ B66 Leverageit þ B67 Trade Intensityjt þ l
ð4Þ
where ROAijt is the return on assets for the ith firm, jth industry and tth time period. Return on assets is measured as net income=total assets (Global 500 data). SHAREijt is the market share for the ith firm in the jth industry for the tth year. Market share is measured as firm sales=world output (Global 500 and UNIDO data). SIZEijt is the firm size for the ith firm of the jth industry for the tth year. Firm size is measured as total assets for the firm (Global 500 data). CAPINijt is the capital intensity for the ith firm of the jth industry for the tth year measured as total assets=sales (Global 500 data). INDj are the industry fixed effects—a set of M 1 (M ¼ 29) industry dummy variables measured at the four digit ISIC level. The dummy is equal to one if the observation falls within the ISIC, zero otherwise. Yeart is a set of L 1 (L ¼ 4) year dummy variables included in the model to capture differences in the model over the course of the business cycle. The omitted category is 1991, the year in which the trough of the recession occurred. Countryk is a set of P 1 (P ¼ 15) country dummy variables included in order to capture inter-country differences in the model. The omitted category is the US. Leverageit is the ratio of stockholders equity to total assets for the ith firm for the tth year. Leverage is included in the model to control for inter-firm differences in financing that could bias the return on assets profit measure. Trade Intensityjt is the ratio of international activity for jth industry for time period t. Trade intensity is equal to (imports þ exports) divided by total industry output. This variable is included to capture inter-industry differences in international trade. Share Countryijt is a set of P 1 interaction terms computed as the product of market share and the country dummy variables. The interaction terms are included in the model in order to capture inter-country differences in the relationship between return on assets and market share. The dependent variable, return on assets, has been widely used in previous empirical research and its strengths and weaknesses as a measure of profitability are well known. Profitability measures have received widespread criticism in the works of Fisher and McGowan (1983) and Benston (1985), but we agree with Martin (1993) who argues that the consequences of Fisher and McGowan and Benston’s arguments if taken to their extreme would portend the end of empirical analysis. One could argue that the potential costs of not using data based upon accounting values are far greater than the costs of using data that are somewhat flawed. Moreover, Mueller (1990) observes that the Fisher–McGowan (1983) criticisms of accounting profitability focus on the inability of rate of return measures to adequately proxy the internal rate of return for a single investment project, whereas most empirical research in industrial economics utilizes accounting return measured for the entire firm. After reviewing a number of prior studies, Mueller (1990) concludes that accounting profit rate and the capital market based profit
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
407
measures preferred by critics of accounting return yield comparable measures of economic profit. Country dummy variables were included in the model in order to control for inter-country effects, particularly inter-country differences in the relationship between market share and return on assets. In addition to the country variables, we include interaction terms for market share and the country dummy variables. Interactions between market share and country are particularly likely in light of the rather substantial differences in the size of domestic markets faced by firms included in the Global Fortune 500. We deleted observations for countries with only one firm in order to avoid commingling market share and country effects. The leverage variable was included in the model to capture inter-firm differences in the method of financing and the related bankruptcy risk that might otherwise bias the relationships between return on assets and the right hand side variables. The continuously measured firm variables (market share, firm size and capital intensity) are included based upon the structure–performance model that has it origins in the works of Bain (1956), Shepherd (1972), Weiss (1971) and others. Inclusion of the year dummy variables is intended to capture business cycle effects, which are included following Domowitz, Hubbard, and Petersen (1986) who found significant cyclical variation in the structure–performance relationship. The 1991 sample year, which contains the trough of the recession (in the US), serves as the omitted category. As stated above, the industry dummy variables are included in the model following the fixed effects model first presented by Schmalensee (1985) and adopted by a number of subsequent researchers. We selected the industry whose return on assets was closest to the mean for all industries to serve as the omitted category. For our sample, the mean return on assets for the Malt Liquors and Malt industry (ISIC 3133) most closely approximated the mean for all firms contained in the sample. The mean return for all firms was 4.01%, while the mean for industry 3133 was 4.12%. We deleted the observations for any industry that included only one firm. As stated above, means and standard deviations for all variables included in the model are found in Appendix A. Prior empirical studies focusing on international competition have generally included exports and/or imports as right hand variables. Variables capturing import and export intensity that measure the percentage of total industry sales imported into or exported from a particular country or trade zone (e.g. the European Union) have been included in previous works to capture the international flow of goods and services into and out of the country. Our focus is on world total output; hence there is no comparable measure for flows into and out of the country. However, inter-industry differences in the volume of trade flows are an important indicator of the extent to which various industries are international in scope. We include a single measure, trade intensity, the ratio of imports plus exports to global total industry output, to capture inter-industry differences in the relative volume of international trade. After a description of the data, the remainder of this paper is devoted to the estimation of Eq. (4) and a discussion of the results.
408
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
4. Data A study regarding the importance of firm and industry effects for firms at the international level is constrained by available data. Fortune Global 500 provides a data source that contains profits, revenues, assets, and several other variables for the largest firms in the world. Our sample is based upon Global Fortune 500 data for the years 1989–1992. Fortune 500 data are widely familiar to researchers, as the data for domestic firms have been used for previous research in industrial economics, notably in the work of Shepherd (1972). Although the Fortune Global 500 classifies firms into 27 different industrial sectors, many of the groupings such as food, electronics and electrical equipment, and metal products are too broad to correspond to meaningful industry definitions. The United Nations Industrial Development Organization’s (UNIDO) International Standard Industrial Classification Codes (ISICs) provide a more appropriate set of industries. Moreover, the UNIDO data contain global output totals by ISIC that can be used in the computation of world market share for the multinational firms contained in the study. The level of aggregation is the four-digit ISIC, an industry definition that generally falls somewhere between three- and four-digit US Census SIC codes. The end years of our data set (1989 and 1992) were dictated by data availability. The Fortune Global 500 first appeared in 1990, reporting data for 1989, while 1992 is the last year for which four-digit ISIC data are available.1 We know of no data source that publishes ISIC codes for international corporations. However, The International Directory of Corporate Affiliations publishes Census SIC codes for most of the world’s major corporations. Given the broader definition for ISIC industries as compared to SICs, conversion from the Census SIC industries to ISICs is generally straightforward. Many of the firms listed in the Global 500 operate in several different SIC codes and were thus too diversified to be included in the sample. Additionally, a small number of firms were not listed in The International Directory of Corporate Affiliations. After deleting firms that were too diversified or for which the SIC was unavailable, a final sample of up to 241 firms per year was available for estimation, with a pooled sample of 957 observations.2 Appendix A contains descriptions of the Fortune Global 500 and Unido data sources as well as means and standard deviations for the variables included in Eq. (2). Appendix B contains a list of the industries included in the sample. The data set of 957 observations includes 362 observations for US firms and 227 for Japanese companies. Although US and Japanese companies are very prominent in the Global 500, the data set also contains 318 observations for firms that are based 1 Although UNIDO has published data covering the period after 1992, data for the more recent years are frequently published for combinations of 4-digit ISIC industries. Using the data for years after 1992 would result in excessively broad industry definitions and thus undermine the measurement of market share and industry effects. 2 The number of firms varies somewhat from year to year as firms enter and leave the Global 500.
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
409
in countries other than the US or Japan. All monetary values are expressed in nominal US dollars. 5. Empirical results The profitability relationship expressed in Eq. (4) is estimated using the Global 500 data for the years 1989–1992; the final sample contains 957 observations. As stated above, the level of aggregation for world output is the UNIDO four-digit ISIC industry. Four-digit ISIC industries are generally broader than four digit Census SIC industries, generally falling somewhere between the three and fourdigit SIC levels of aggregation. As a result of the broader industry definitions, we would expect industry effects to be weaker using the ISIC industries as compared to prior studies using Census SIC industries or the even more narrowly defined FTC Lines of Business. A finding that there are important industry effects, even while using ISIC industry classifications, would thus provide strong evidence of competition at the global level. The estimated coefficients for the alternative specifications of the model are contained in Table 2. All standard errors and t-statistics are computed using White’s heteroskedasticity correction factor that yields robust error terms. Greene (2000: p. 579) suggests White’s heteroskedasticity consistent standard errors as an appropriate method for dealing with potential biases in t-statistics that are inherent with the use of panel data. The first column of Table 2 contains the estimated coefficients for the firm effects specification with industry fixed effects excluded, while column 2 Table 2 Estimated regression coefficient return on assets related to firm level and industry effects Variable Intercept SHAREi SIZEi CAPINi Leverage Trade intensity Year1989 Year1990 Year1992 Country effects Share/country interactions Industry fixed effects R2 F-ratio
Firm effects model
Industry effects model
Complete model
0.02 (2.11) 0.03 (2.11) 0:13 106 (0.20) 0.006 (1.94) 0.12 (6.93) 0.0003 (1.76) 0.0002 (0.05) 0.009 (0.52) 0.02 (11.18) Includeda Includeda
0.04 (2.26) Excluded Excluded Excluded 0.11 (7.78) 0.001 (1.21) 0.0004 (0.08) 0.01 (1.70) 0.03 (6.71) Excluded Excluded
0.05 (3.75) 0.06 (2.85) 0:78 107 (0.96) 0.01 (1.02) 0.10 (5.77) 0.001 (1.16) 0.0001 (0.06) 0.01 (1.96) 0.03 (6.71) Included Included
Excluded 0.24 7.09
Includedb 0.29 11.28
Included 0.34 6.69
t-ratios in parentheses, computed using White’s robust standard errors. Significant at the 0.05 level. a Specification in comparison to complete model, F ¼ 5:03, significant 0.01 level. b Specification in comparison to complete model, F ¼ 3:52, significant 0.01 level.
410
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
contains results from the industry fixed effects specification. Column 3 contains the complete model that includes both firm effects and industry fixed effects. The country dummy variables as well as the country/market share interactions are included in both the firm effects and full models. Examination of column 1 reveals that firm effects provide reasonably good explanatory power. The R2 of 0.24 is relatively high as compared to the firm effects found in previous studies. Moreover, the coefficient for market share is positive and statistically significant at the 0.05 level. This finding is consistent with prior research using domestic US data [including Ravenscraft (1983) and Gale and Branch (1982)] that has consistently found a positive relationship between profitability and market share. The coefficient for the leverage variable is positive and significant, a finding that is consistent with the a priori expectation that firms with greater ratios of stockholders equity to total assets have higher accounting profitability, ceteris paribus. Column 2 of Table 2 summarizes the specification that includes the industry fixed effects dummy variables. In order to control for potential biases related to inter-industry differences in international activity, we include the trade intensity variable in the industry fixed effects model. We also include the leverage variable to avoid biases resulting from inter-firm differences in financing and the year dummies to control for business cycle effects. The R2 of 0.29 is in keeping with previous studies that have found substantial explanatory power from the industry fixed effects specification. Our finding of strong industry effects is consistent with previous research by Schmalensee (1985), Powell (1996) and others who found that industry effects dominate firm effects in explaining cross-sectional differences in profitability. A finding that industry effects are important in explaining cross-sectional profit rate variation for a sample of global firms provides tentative evidence regarding the importance of competition at the global level in determining the profitability of multinational firms. Estimated coefficients for the complete model that includes both firm level and industry effects can be found in column 3 of Table 2. The estimated coefficients and explanatory power for the complete model reinforce the conclusions based on the separate firm and industry effects models. The R2 for the complete model is 0.34. When compared with the R2 of 0.24 for the firm effects model in column 1, we see that the industry effects explain 10% of the total variation in rate of return. While explaining 10% of the total variation in profit rate is not comparable to the strongest industry effects found in previous studies (e.g. Schmalensee, 1985), our findings do suggest a meaningful role for industry (with geographic market defined as global) in explaining profit rate variations among Global Fortune 500 firms. Industry effects that explain 10% of total profit rate variation are especially meaningful in light of the very broad industry definitions provided by the UNIDO data. The market share variable, positive and significant in the firm effects model, remain positive and significant in the complete model. Keeping in mind that market share is measured at the global level (i.e. the firm’s percentage of total world industry output), the finding of a positive and significant coefficient for market share provides compelling evidence of competition at the international level. The
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
411
positive and significant coefficient for market share is particularly meaningful given the broad industry definitions. A consistently positive coefficient for market share, together with industry effects that explain 10% of the variation in rate of return, point to the importance of global competition in explaining variation in profitability. While our data do not permit us to directly test for the existence of international competition among large multinational firms, the market share and industry effects revealed in this study both indicate the presence of international competition. The coefficient for the 1992-year dummy variable is negative and significant in every specification (firm effects, industry effects and complete model). Our finding of a negative and significant coefficient for the year following the trough of the recession in the US is not surprising given the diversity of industries and countries represented in our sample. Our findings suggest, therefore, that there is substantial variation across industries and countries in terms of the timing of the business cycle. That the coefficients for the 1992 dummy variable is statistically different from zero in every model tested supports Domowitz et al.’s (1986) findings regarding the importance of business cycle effects in determining profit rate variation. Our results extend the findings of Domowitz et al. for US firms to multinational firms. The trade intensity variable was included as a right hand side variable in our regressions to capture the extent to which various industries are international in scope. The estimated coefficient for trade intensity in our models, however, is statistically insignificant in every specification. This finding is not surprising when analyzed relative to prior results regarding the relationship between profitability and exports. Earlier studies have alternately found no export effect on profitability (Caves & Uekusa, 1976; Jacquemin, de Ghellinck, & Huveneers, 1980; Pagoulatos & Sorenson, 1976a); a positive effect of exports on profitability (Caves, KhalilzadehShirazi, & Porter, 1975; Geroski, 1982; Lyons, 1981; Pugel, 1978); or a negative impact of exports on profitability (Jenny & Weber, 1976; Neumann, Bobel, & Haid, 1979; Pagoulatos & Sorenson, 1976b). Given the conflicting evidence from prior research, the lack of statistical significance for our trade intensity variable is not surprising. The potential for multicollinearity exists in a model that includes market share, firm size, capital intensity and trade intensity as regressors. Gujarati (2003) illustrates the impact of collinearity on coefficient variances for alternative correlations among regressors. Gujarati’s (2003) simulations suggest that correlations among regressors of 0.5 or less have relatively little impact on the estimated coefficient variances and t-statistics. Examining the correlation matrix reveals that the correlations among regressors in our model are all substantially less than 0.5 with the exception of a 0.54 correlation between market share and firm size. A 0.54 correlation among regressors implies an increase in the estimated coefficient variance of slightly more than one-third. Examining the estimated t-statistics for market share and firm size, it appears that the multicollinearity is of no consequence in this case. The coefficient for market share, our main variable of interest, is statistically significant in every model despite the presence of multicollinearity. Moreover, the
412
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
estimated t-statistics of 0.20 and 0.96 for the firm size variable in the firm effects and complete models, respectively, are sufficiently below critical values at the 0.05 level to suggest that slightly more than a one-third increase in the estimated coefficient variances is of no consequence. The overall conclusions from the various models presented in Table 2 include the importance of both firm market share and industry effects in explaining profit rate variation. These findings are very similar to previously published findings using domestic data. We believe that these findings regarding the importance of industry effects at the international level are among the first of their kind. A strong industry effect along with a positive and significant coefficient for global market share for Global Fortune 500 firms provide important evidence that there is rivalry among multinational firms operating in the same global industry. A variety of alternative explanations have been given for the existence of strong industry effects in cross-sectional rate of return models. In addition to the usual interpretation that strong industry effects are evidence of rivalry among firms in the same industry, many critics of the industry effects specification have maintained that strong industry effects may be capturing nothing more meaningful than similarities in accounting conventions among firms operating in the same broad industry. While industry-level similarities in accounting conventions may explain strong industry effects for firms headquartered in the same country, it is not likely that firms from the same industry but based in different nations would be expected to adopt similar accounting standards. It seems reasonable that differences in accounting practices across countries would dominate similarities in accounting practice for firms from the same industry. Our finding of strong industry effects using a sample of global firms provides important evidence that something other than industry level accounting conventions account for the strength of industry effects revealed in this and other studies. While there are alternative explanations for the presence of industry effects at the global level, one cannot discount the likelihood that our findings are evidence that multinational firms compete in global output markets.
6. Conclusion The purpose of this study was to test the proposition that firm market share and industry fixed effects, both measured globally, explain profitability variation for a sample of multinational firms. The most important finding is that a sample of Global Fortune 500 firms exhibits meaningful market share and industry effects, a finding that is consistent with prior studies using domestic data. While there are a number of explanations for industry effects, a common interpretation is that the presence of industry effects signals rivalry among firms. The positive and significant coefficient for market share in both the complete and firm effects models provides additional evidence to support the existence of rivalry among global firms. Our findings provide a critical first piece of evidence regarding rivalry among multinational firms. Moreover, our findings provide some measure of support for the US
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
413
Department of Justice’s most recent merger guidelines that establish international firms as a source of potential competition for the purposes of deciding whether to allow or disapprove mergers. Although globalization is not a new phenomenon, it is true that there has been a sharp increase in the importance of trade in manufactured goods in the last 20 years. This trend, together with the increasing importance of intra-industry trade suggests that the relevant geographic market for many manufactured goods is now global. Our empirical results supply initial evidence that global competition in manufactured goods industries is important. Given that these are the first empirical results of their kind, the usual call for additional evidence before using the findings in support of public policy would appear especially appropriate. Appendix A. Descriptive statistics and characteristics of the data Variable
Mean
Standard deviation
Return on assets Market share Firm size Capital intensity Leverage Trade intensity
4.01% 8.5% 14 326.97 1.07 0.34 9.52
6.0 18.09 22 221.97 0.54 0.18 9.57
The UNIDO has general responsibility for the collection and dissemination of international industrial statistics. The activities regarding industrial statistics include the compilation of manufacturing industry statistics at the four-digit ISIC level. Data are available at the country level and for world totals by industry. Variables included in the UNIDO database are: number of establishments, number of employees, wages and salaries, output, and value added. The UNIDO data are available for a maximum of 81 manufacturing industries, with the industry coverage varying by country. This paper makes use of the UNIDO database for world totals by manufacturing industry, which we use in calculating the global market shares of firms for the years 1989–1992. The Fortune Global 500 was first published in 1990, with data primarily for the previous calendar or fiscal year. The Global 500 listing includes the largest 500 multinational firms ranked by sales, with the requirement that all companies included must have obtained at least 50% of their sales from manufacturing and/ or mining. (In more recent years, the Global 500 has been expanded to include companies in banking and finance.) The Fortune data set includes the following variables: sales, profits after tax, assets, stockholders’ equity, and number of employees. We obtain the following firm level variables from the Fortune Global 500: sales, profits after tax, assets, stockholder’s equity, and the home country of the firm.
414
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
Appendix B. Industries included in sample ISIC code
Industry definition
3113 3116 3119 3122 3131 3133 3134 3140 3220 3411 3420 3511 3521 3522 3523
Canning and preserving fruits and vegetables Grain mill products Manufacture of cocoa, chocolate and sugar confectionery Manufacture of prepared animal feeds Distilling, rectifying and blending spirits Malt liquors and malt Soft drinks and carbonated water industries Tobacco manufactures Manufacture of wearing apparel except footwear Manufacture of pulp, paper and paperboard Manufacture of furniture and fixtures, except primarily metal Manufacture of basic industrial chemicals except fertilizer Manufacture of paints, varnishes and lacquers Manufacture of drugs and medicines Manufacture of soaps and cleaning preparations, perfumes, cosmetics and other toilet preparations Petroleum refineries Tyre and tube industries Manufacture of glass and glass products Iron and basic steel industries Non-ferrous metal basic industries Manufacture of Engines and Turbines Manufacture of agricultural machinery and equipment Manufacture of office, computing and accounting machinery Manufacture of electrical industrial machinery and apparatus Manufacture of radio, television, and communication equipment and apparatus Manufacture of motor vehicles Manufacture of motorcycles and bicycles Manufacture of aircraft Manufacture of professional and scientific and measuring and controlling equipment Manufacture of photographic and optical goods Manufacture of watches and clocks
3530 3551 3620 3710 3720 3821 3822 3825 3831 3832 3843 3844 3845 3851 3852 3853
References Amato, L., & Wilder, R. P. (1990). Firm and industry effects in industrial economics. Southern Economic Journal, 57, 93–105. Bain, J. S. (1956). Barriers to new competition. Cambridge: Harvard University Press.
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
415
Benston, G. J. (1985). The validity of profits-structure with particular reference to the FTC’s line of business. American Economic Review, 75, 37–67. Brush, T., Bromiley, P., & Hendrickx, M. (1999). The relative influence of industry and corporation on business segment performance: an alternative estimate. Strategic Management Journal, 20, 519–547. Caves, R. E. (1989). International differences in industrial organization. In R. Schmalensee, & R. D. Willig (Eds.), Handbook of industrial organization, Vol. II. Amsterdam: Elsevier. Caves, R. E. (1996). Multinational enterprise and economic analysis. (2nd ed.). Cambridge: Cambridge University Press. Caves, R. E., Khalilzadeh-Shirazi, J., & Porter, M. E. (1975). Scale economies in statistical analyses of market power. Review of Economics and Statistics, 57, 133–140. Caves, R. E., & Uekusa, M. (1976). Industrial organization in Japan. Washington: Brookings Institution. Co, C. Y. (2001). Trade, foreign direct investment and industry performance. International Journal of Industrial Organization, 19, 163–183. Cubbin, J., & Geroski, P. (1987). The convergence of profits in the long run: inter-firm and inter-industry comparisons. Journal of Industrial Economics, 35, 427–442. Demsetz, H. (1973). Industry structure, market rivalry, and public policy. Journal of Law and Economics, 16, 1–9. Domowitz, I., Hubbard, R. G., & Petersen, B. C. (1986). Business cycles and the relationship between concentration and price–cost margins. Rand Journal of Economics, 17, 1–17. Feenstra, R. C. (1998). Integration of trade and disintegration of production in the global economy. Journal of Economic Perspectives, 12, 31–50. Fisher, F. M., & McGowan, J. J. (1983). On the misuse of accounting rates of return to infer monopoly profits. American Economic Review, 73, 82–97. Gale, B. T., & Branch, B. S. (1982). Concentration versus market share: what determines performance and why does it matter? Antitrust Bulletin, 27, 83–106. Geroski, P. A. (1982). Simultaneous equations models of the structure–performance paradigm. European Economic Review, September, 145–158. Greene, W. H. (2000). Econometric analysis. (4th ed.). Upper Saddle River, NJ: Prentice Hall. Gujarati, D. N. (2003). Basic econometrics. (4th ed.). Boston, MA: McGraw Hill. Hout, T., Porter, M. E., & Rudden, E. (1998). How global companies win out. In M. E. Porter (Ed.), On competition (pp. 289–308). Boston: Harvard Business School Press. Jacquemin, A., de Ghellinck, E., & Huveneers, C. (1980). Concentration and profitability in a small open economy. Journal of Industrial Economics, 29, 131–144. Jenny, F., & Weber, A.-P. (1976). Profit rates and structural variables in French manufacturing industries. European Economic Review, 7, 187–206. Levinsohn, J. (1993). Testing the imports-as-market discipline hypothesis. Journal of International Economics, 35, 1–22. Lyons, B. (1981). Price–cost margins, market structure, and international trade. In D. Currie, D. Peel, & W. Peters (Eds.), Microeconomic analysis: essays in microeconomics and economic development. London: Croom Helm. Martin, S. (1993). Advanced industrial economics. Oxford: Blackwell. Mueller, D. C. (1990). The dynamics of company profits: An international comparison. Cambridge: Cambridge University Press. Neumann, M., Bobel, I., & Haid, A. (1979). Profitability, risk and market structure in West German industries. Journal of Industrial Economics, 27, 227–242. Pagoulatos, E., & Sorenson, R. (1976a). International trade, international investment, and industrial profitability of US Manufacturing. Southern Economic Journal, January, 424–434. Pagoulatos, E., & Sorenson, R. (1976b). Foreign trade, concentration, and profitability in open economies. European Economic Review, 8, 255–267. Powell, T. C. (1996). How much does industry matter: an alternative empirical test. Strategic Management Journal, 17, 323–334. Pugel, T. A. (1978). International market linkages and US manufacturing: price, profits and patterns. Cambridge, MA: Ballinger.
416
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
Pugel, T. A. (1980). Foreign trade and US market performance. Journal of Industrial Economics, December, 119–129. Ravenscraft, D. J. (1983). Structure–profit relationships at the line of business and industry level. Review of Economics and Statistics, 65, 22–31. Rumelt, R. P. (1991). How much does industry matter? Strategic Management Journal, 12, 167–185. Schmalensee, R. (1985). Do markets differ much? American Economic Review, 75, 341–351. Schmalensee, R. (1987). Collusion versus differential efficiency: testing alternative hypotheses. Journal of Industrial Economics, 35, 399–425. Scott, J. T., & Pascoe, G. (1986). Beyond firm and industry effects of profitability in imperfect markets. Review of Economics and Statistics, 68, 284–292. Shepherd, W. G. (1972). The elements of market structure. Review of Economics and Statistics, February, 25–37. Weiss, L. W. (1971). Quantitative studies of industrial organization. In M. D. Intrilligator (Ed.), Frontiers of quantitative economics. Amsterdam: North Holland. Wernerfelt, B., & Montgomery, C. (1988). Tobin’s Q and the importance of focus in firm performance. American Economic Review, 78, 246–250. Yamawaki, H. (1986). Exporting, foreign market structure and profitability: Japanese and US manufacturing. Review of Economics and Statistics, 68, 618–627. Yamawaki, H., Sleuwaegen, L., & Weiss, L. W. (1989). Industry competition and the formation of the European common market. In L. W. Weiss (Ed.), Concentration and price. Cambridge, MA: MIT Press.
Global competition and global markets: some empirical results Louis H. Amato a, Ronald P. Wilder b, a
b
Department of Economics, Belk College of Business Administration, University of North Carolina at Charlotte, Charlotte, NC 28223, USA Department of Economics, Darla Moore School of Business, University of South Carolina, Columbia, SC 29208, USA
Abstract The purpose of this paper is to develop and test a model that relates the profit rates of multinational firms to market share, other firm characteristics, and industry effects where the relevant geographic markets are defined at the global level. We test the proposition that market share and industry fixed effects can explain profitability variation for a sample of multinational firms. The most important finding is that market share has a positive and statistically significant effect on profit rates for a sample of Fortune Global 500 firms. Moreover, the Fortune Global 500 firms exhibit strong industry effects. While there are a number of explanations for a positive market share effect and strong industry effects, a common interpretation is the presence of rivalry among these firms. # 2004 Elsevier Ltd. All rights reserved. Keywords: Global markets; Industrial organization; Market structure; Firm strategy and market performance; Multinational firms; Profitability; Oligopoly
1. Introduction The increasing importance of multinational firms and international trade in manufactured goods leads naturally to the question of whether the relevant geographic market for many manufacturing industries is now global rather than national. With some notable exceptions, much of the previous empirical work building on the structure– conduct–performance paradigm has made use of data that follow national geographic market definitions. This paper is motivated by
Corresponding author. Tel.: +1-803-777-6955; fax: +1-803-777-6876. E-mail address: [email protected] (R.P. Wilder).
0969-5931/$ - see front matter # 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.ibusrev.2003.09.007
402
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
our interest in whether it is meaningful to define manufacturing markets globally. Hout, Porter, and Rudden (1998) suggest that multinational firms compete in two types of industries: multi-domestic industries in which firms pursue independent strategies in each of a number of foreign markets, and global industries in which firms compete with other multinational firms in the world market at large. Our hypothesis in this work is that global industries are sufficiently important so that competitive interaction among firms takes place at this level of industry definition. Richard Caves (1989: p. 1244) summarizes the need for research regarding the consequences of global competition as follows: ‘‘Discussions in the field of business strategy have focused on ‘global competition’ in which international rivals pursue strategies that treat the world’s sub-markets as interdependent. Yet systematic empirical research on international oligopolistic behavior is quite limited.’’ The transaction cost explanation for the existence of multinational firms suggests that such firms will tend to be present in industries with concentrated sellers (Caves, 1996, Chapter 4). Global oligopoly is thus a phenomenon likely to coincide with the growth of multinational enterprise. The potential importance of global competition is also suggested by the growth in international trade in manufactured goods. While merchandise trade relative to GDP has increased for most industrial countries during the past three decades, the trend is dampened by the increasing importance of services as a component of GDP. As pointed out by Feenstra (1998: Table 2), merchandise trade relative to merchandise value-added has increased sharply for most industrial countries during the recent past. This increase is due in part to falling trade barriers and lower transportation costs, but is also related to what Feenstra calls the ‘‘disintegration of production’’, with substantial importing of intermediate inputs. The importance of global vertical production chains also suggests the importance of the study of global competition. The purpose of this paper is to develop and test a model that relates the profit rates of multinational firms to firm characteristics and industry effects, where markets are defined at the global level. The research addresses important public policy issues related to the effectiveness of global competition in restraining oligopolistic pricing behavior of firms that may face little price competition from domestic rivals. Moreover, the findings are relevant to firms concerned about the strategic actions of rivals operating on a global scale. The paper utilizes a methodology similar to that used by Schmalensee (1985), developed further by Rumelt (1991) and Brush, Bromiley, and Hendrickx (1999), which relates profit rates to individual firm characteristics and industry fixed effects. Our data set of international firms was gathered from the Global Fortune 500 for the years 1989–1992. We know of no prior research regarding global competition that employs data similar to those contained in the Global 500. We note, however, that prior studies using US data have consistently found that industry effects are important in explaining cross sectional variation in profit rates. The two major findings reported in this paper are: (1) firm market share in the total global market
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
403
has a positive and statistically significant effect on firm profitability; and (2) global industry fixed effects are also important in explaining firm profitability, a finding that is in keeping with the results of previous studies for national markets. These findings support the hypothesis that multinational firms compete in global industries rather than operating exclusively as multi-domestic firms. 2. Previous works A brief digest of previous works is shown in Table 1. The previous research on firm effects relative to industry effects in explaining profitability began with papers by Ravenscraft (1983) and Schmalensee (1985) making use of national data from the US Federal Trade Commission (FTC). Using FTC Line of Business data for 1975, Schmalensee finds that firm effects do not account for cross-sectional variation in profitability, market share effects account for a negligible proportion, and industry effects account for approximately 20% of profit variation. Schmalensee’s now widely familiar finding that industry effects dominate firm effects has been Table 1 Summary of selected items of previous research on determinants of profitability Effects
Studies
Dataa
Findings
Firm
Scott and Pascoe (1986)
D
Cubbin and Geroski (1987)
D
Wernerfelt and Montgomery (1988) Brush et al. (1999)
D
Firm effects are important; firm and industry effects overlap Persistent profitability differences among firms in the same industry in a dynamic context Firm effects appear as focus effects
D
Firm effects are stronger than industry effects
Business unit
Rumelt (1991)
D
Strong business unit effects in longitudinal panel data; industry effects are important in cross sections
Market share
Ravenscraft (1983)
D
Market share effects dominate industry concentration effects
Industry
Schmalensee (1985)
D
Amato and Wilder (1990)
D
Industry effects dominate firm and market share effects Industry effects dominate firm size effects
Co (2001)
I
Yamawaki et al. (1989)
I
Global competition
a
Imports and foreign direct investment provide competitive discipline European Union seller concentration has a greater effect than domestic market concentration
Data type: D, domestic data from a single country; I, international data.
404
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
offered as evidence that the industry is an appropriate unit of observation for research and policy making in industrial economics. Subsequent studies by Scott and Pascoe (1986), Cubbin and Geroski (1987), Wernerfelt and Montgomery (1988) and Amato and Wilder (1990) provide at least partial support for Schmalensee’s original finding, although neither Scott and Pascoe nor Cubbin and Geroski found industry effects to be as dominant as those revealed in Schmalensee’s original article. More recent empirical work regarding the relative importance of firm and industry effects is found in the strategic management literature, and provides mixed results. Rumelt (1991) employed a variance components model to distinguish between stable and fluctuating firm and industry effects. Rumelt found small or no corporate effects, stable but small industry effects, and a large yet stable business unit effect. While Rumelt’s findings do not support the dominance of industry effects revealed in previous studies, he concludes that industry effects are important in explaining cross-sectional variation in profit rates. Other works on this theme in the strategic management literature include Powell (1996), and Brush et al., (1999). These studies, along with others that they survey, find that line of business effects, firm effects and industry effects are each important, with the relative importance varying across studies. In all cases, industries are defined to include only domestic firms. Another research theme examines the effects of global competition on industries in individual nations. Co (2001) finds a tendency for imports as well as foreign direct investment to provide competitive discipline for domestic industries in the United States. Levinsohn (1993) reports a similar result for imports in Turkey. In a study of paired Japanese and US four digit industries, Yamawaki (1986) finds that the profit margins of Japanese manufacturers increase as entry barriers such as scale economies and product differentiation increase in the US markets that the Japanese exporters serve. Pugel (1980) finds that the price–cost margin on export sales rises with increased domestic concentration, while Yamawaki, Sleuwaegen, and Weiss (1989) conclude that concentration across the entire European Union has greater influence on price–cost margins than concentration within a producer’s own nation. While prior research offers evidence regarding the importance of competition from both domestic rivals and from imports in explaining the profitability of firms and industries at the national or customs union area level, there is virtually no prior research examining the importance of global competition, broadly defined, in determining the profitability of multinational firms. The present work provides an initial examination of competition at the global industry level.
3. Conceptual framework and model specification There are two main traditions for analyzing the impact of industry competition on the profitability of firms. The differential collusion hypothesis is that collusion is likely to be more effective with higher seller concentration and that firm profitabil-
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
405
ity tends to be higher in industries with higher seller concentration. The revisionist tradition, following Demsetz (1973), is that more efficient firms gain higher market shares and earn higher profit rates. This efficiency effect may also increase seller concentration, but the driving force behind higher firm profits under the Demsetz view is efficiency rather than collusion. Although our data set does not allow the testing of the relative importance of these two alternative hypotheses, both suggest the importance of firm market share and industry characteristics in determining firm profitability. The relationship between profitability and market share is developed following Schmalensee (1987), as generalized by Martin (1993: pp. 494–496). This approach begins with the price–cost margin for an oligopoly firm that makes the Cournot assumption about rivals’ quantity adjustment: ð p ci Þ s i ¼ p e
ð1Þ
where p is price, ci is marginal cost, si is market share and e is the priceelasticity of the industry demand function. The firm’s accounting rate of return on capital (PA) can be shown to be: PA ¼ q þ
pqi si pk Ki e
ð2Þ
where q is the economy-wide normal rate of return on capital, qi is the firm output, pk is the price of physical capital and Ki is the quantity of physical capital. This relationship forms the basis for the estimating equation used in hypothesis testing in this paper. We take q to be the global normal rate of return on capital in our sample of multinational firms; it is thus reflected in the intercept term of the estimating equations shown below. Moreover, the Global 500 data used for this study allow for the direct measurement of market share at the global level. While our data do not permit the direct measurement of pk or Ki, the capital intensity variable appearing on the right hand side of our model is intended to capture inter-firm differences in both the price and physical quantity of capital. Finally, we know of no data source that provides demand elasticity measures for firms appearing in the Global 500. However, by including industry fixed effects as regressors, our methodology allows for various inter-industry factors that influence profitability, including differences in the market demand elasticity. A verbally expressed model, building on the relationship developed in (1) and (2) above, relating return on assets to market share, capital intensity, industry fixed effects and other variables is: Return on Assets ¼ f ðMarket Share; Firm Size; Capital Intensity; Industry Effects; Country Effects; Leverage; Trade IntensityÞ ð3Þ
406
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
The empirical specification to be estimated is found in Eq. (4): X ROAijt ¼ B0 þ B1 SHAREijt þ B2 SIZEit þ B3 CAPINit þ B4þj INDj X X X þ B32þt Yeart þ B36þk Countryk þ B51þk Share Countryijt þ B66 Leverageit þ B67 Trade Intensityjt þ l
ð4Þ
where ROAijt is the return on assets for the ith firm, jth industry and tth time period. Return on assets is measured as net income=total assets (Global 500 data). SHAREijt is the market share for the ith firm in the jth industry for the tth year. Market share is measured as firm sales=world output (Global 500 and UNIDO data). SIZEijt is the firm size for the ith firm of the jth industry for the tth year. Firm size is measured as total assets for the firm (Global 500 data). CAPINijt is the capital intensity for the ith firm of the jth industry for the tth year measured as total assets=sales (Global 500 data). INDj are the industry fixed effects—a set of M 1 (M ¼ 29) industry dummy variables measured at the four digit ISIC level. The dummy is equal to one if the observation falls within the ISIC, zero otherwise. Yeart is a set of L 1 (L ¼ 4) year dummy variables included in the model to capture differences in the model over the course of the business cycle. The omitted category is 1991, the year in which the trough of the recession occurred. Countryk is a set of P 1 (P ¼ 15) country dummy variables included in order to capture inter-country differences in the model. The omitted category is the US. Leverageit is the ratio of stockholders equity to total assets for the ith firm for the tth year. Leverage is included in the model to control for inter-firm differences in financing that could bias the return on assets profit measure. Trade Intensityjt is the ratio of international activity for jth industry for time period t. Trade intensity is equal to (imports þ exports) divided by total industry output. This variable is included to capture inter-industry differences in international trade. Share Countryijt is a set of P 1 interaction terms computed as the product of market share and the country dummy variables. The interaction terms are included in the model in order to capture inter-country differences in the relationship between return on assets and market share. The dependent variable, return on assets, has been widely used in previous empirical research and its strengths and weaknesses as a measure of profitability are well known. Profitability measures have received widespread criticism in the works of Fisher and McGowan (1983) and Benston (1985), but we agree with Martin (1993) who argues that the consequences of Fisher and McGowan and Benston’s arguments if taken to their extreme would portend the end of empirical analysis. One could argue that the potential costs of not using data based upon accounting values are far greater than the costs of using data that are somewhat flawed. Moreover, Mueller (1990) observes that the Fisher–McGowan (1983) criticisms of accounting profitability focus on the inability of rate of return measures to adequately proxy the internal rate of return for a single investment project, whereas most empirical research in industrial economics utilizes accounting return measured for the entire firm. After reviewing a number of prior studies, Mueller (1990) concludes that accounting profit rate and the capital market based profit
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
407
measures preferred by critics of accounting return yield comparable measures of economic profit. Country dummy variables were included in the model in order to control for inter-country effects, particularly inter-country differences in the relationship between market share and return on assets. In addition to the country variables, we include interaction terms for market share and the country dummy variables. Interactions between market share and country are particularly likely in light of the rather substantial differences in the size of domestic markets faced by firms included in the Global Fortune 500. We deleted observations for countries with only one firm in order to avoid commingling market share and country effects. The leverage variable was included in the model to capture inter-firm differences in the method of financing and the related bankruptcy risk that might otherwise bias the relationships between return on assets and the right hand side variables. The continuously measured firm variables (market share, firm size and capital intensity) are included based upon the structure–performance model that has it origins in the works of Bain (1956), Shepherd (1972), Weiss (1971) and others. Inclusion of the year dummy variables is intended to capture business cycle effects, which are included following Domowitz, Hubbard, and Petersen (1986) who found significant cyclical variation in the structure–performance relationship. The 1991 sample year, which contains the trough of the recession (in the US), serves as the omitted category. As stated above, the industry dummy variables are included in the model following the fixed effects model first presented by Schmalensee (1985) and adopted by a number of subsequent researchers. We selected the industry whose return on assets was closest to the mean for all industries to serve as the omitted category. For our sample, the mean return on assets for the Malt Liquors and Malt industry (ISIC 3133) most closely approximated the mean for all firms contained in the sample. The mean return for all firms was 4.01%, while the mean for industry 3133 was 4.12%. We deleted the observations for any industry that included only one firm. As stated above, means and standard deviations for all variables included in the model are found in Appendix A. Prior empirical studies focusing on international competition have generally included exports and/or imports as right hand variables. Variables capturing import and export intensity that measure the percentage of total industry sales imported into or exported from a particular country or trade zone (e.g. the European Union) have been included in previous works to capture the international flow of goods and services into and out of the country. Our focus is on world total output; hence there is no comparable measure for flows into and out of the country. However, inter-industry differences in the volume of trade flows are an important indicator of the extent to which various industries are international in scope. We include a single measure, trade intensity, the ratio of imports plus exports to global total industry output, to capture inter-industry differences in the relative volume of international trade. After a description of the data, the remainder of this paper is devoted to the estimation of Eq. (4) and a discussion of the results.
408
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
4. Data A study regarding the importance of firm and industry effects for firms at the international level is constrained by available data. Fortune Global 500 provides a data source that contains profits, revenues, assets, and several other variables for the largest firms in the world. Our sample is based upon Global Fortune 500 data for the years 1989–1992. Fortune 500 data are widely familiar to researchers, as the data for domestic firms have been used for previous research in industrial economics, notably in the work of Shepherd (1972). Although the Fortune Global 500 classifies firms into 27 different industrial sectors, many of the groupings such as food, electronics and electrical equipment, and metal products are too broad to correspond to meaningful industry definitions. The United Nations Industrial Development Organization’s (UNIDO) International Standard Industrial Classification Codes (ISICs) provide a more appropriate set of industries. Moreover, the UNIDO data contain global output totals by ISIC that can be used in the computation of world market share for the multinational firms contained in the study. The level of aggregation is the four-digit ISIC, an industry definition that generally falls somewhere between three- and four-digit US Census SIC codes. The end years of our data set (1989 and 1992) were dictated by data availability. The Fortune Global 500 first appeared in 1990, reporting data for 1989, while 1992 is the last year for which four-digit ISIC data are available.1 We know of no data source that publishes ISIC codes for international corporations. However, The International Directory of Corporate Affiliations publishes Census SIC codes for most of the world’s major corporations. Given the broader definition for ISIC industries as compared to SICs, conversion from the Census SIC industries to ISICs is generally straightforward. Many of the firms listed in the Global 500 operate in several different SIC codes and were thus too diversified to be included in the sample. Additionally, a small number of firms were not listed in The International Directory of Corporate Affiliations. After deleting firms that were too diversified or for which the SIC was unavailable, a final sample of up to 241 firms per year was available for estimation, with a pooled sample of 957 observations.2 Appendix A contains descriptions of the Fortune Global 500 and Unido data sources as well as means and standard deviations for the variables included in Eq. (2). Appendix B contains a list of the industries included in the sample. The data set of 957 observations includes 362 observations for US firms and 227 for Japanese companies. Although US and Japanese companies are very prominent in the Global 500, the data set also contains 318 observations for firms that are based 1 Although UNIDO has published data covering the period after 1992, data for the more recent years are frequently published for combinations of 4-digit ISIC industries. Using the data for years after 1992 would result in excessively broad industry definitions and thus undermine the measurement of market share and industry effects. 2 The number of firms varies somewhat from year to year as firms enter and leave the Global 500.
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
409
in countries other than the US or Japan. All monetary values are expressed in nominal US dollars. 5. Empirical results The profitability relationship expressed in Eq. (4) is estimated using the Global 500 data for the years 1989–1992; the final sample contains 957 observations. As stated above, the level of aggregation for world output is the UNIDO four-digit ISIC industry. Four-digit ISIC industries are generally broader than four digit Census SIC industries, generally falling somewhere between the three and fourdigit SIC levels of aggregation. As a result of the broader industry definitions, we would expect industry effects to be weaker using the ISIC industries as compared to prior studies using Census SIC industries or the even more narrowly defined FTC Lines of Business. A finding that there are important industry effects, even while using ISIC industry classifications, would thus provide strong evidence of competition at the global level. The estimated coefficients for the alternative specifications of the model are contained in Table 2. All standard errors and t-statistics are computed using White’s heteroskedasticity correction factor that yields robust error terms. Greene (2000: p. 579) suggests White’s heteroskedasticity consistent standard errors as an appropriate method for dealing with potential biases in t-statistics that are inherent with the use of panel data. The first column of Table 2 contains the estimated coefficients for the firm effects specification with industry fixed effects excluded, while column 2 Table 2 Estimated regression coefficient return on assets related to firm level and industry effects Variable Intercept SHAREi SIZEi CAPINi Leverage Trade intensity Year1989 Year1990 Year1992 Country effects Share/country interactions Industry fixed effects R2 F-ratio
Firm effects model
Industry effects model
Complete model
0.02 (2.11) 0.03 (2.11) 0:13 106 (0.20) 0.006 (1.94) 0.12 (6.93) 0.0003 (1.76) 0.0002 (0.05) 0.009 (0.52) 0.02 (11.18) Includeda Includeda
0.04 (2.26) Excluded Excluded Excluded 0.11 (7.78) 0.001 (1.21) 0.0004 (0.08) 0.01 (1.70) 0.03 (6.71) Excluded Excluded
0.05 (3.75) 0.06 (2.85) 0:78 107 (0.96) 0.01 (1.02) 0.10 (5.77) 0.001 (1.16) 0.0001 (0.06) 0.01 (1.96) 0.03 (6.71) Included Included
Excluded 0.24 7.09
Includedb 0.29 11.28
Included 0.34 6.69
t-ratios in parentheses, computed using White’s robust standard errors. Significant at the 0.05 level. a Specification in comparison to complete model, F ¼ 5:03, significant 0.01 level. b Specification in comparison to complete model, F ¼ 3:52, significant 0.01 level.
410
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
contains results from the industry fixed effects specification. Column 3 contains the complete model that includes both firm effects and industry fixed effects. The country dummy variables as well as the country/market share interactions are included in both the firm effects and full models. Examination of column 1 reveals that firm effects provide reasonably good explanatory power. The R2 of 0.24 is relatively high as compared to the firm effects found in previous studies. Moreover, the coefficient for market share is positive and statistically significant at the 0.05 level. This finding is consistent with prior research using domestic US data [including Ravenscraft (1983) and Gale and Branch (1982)] that has consistently found a positive relationship between profitability and market share. The coefficient for the leverage variable is positive and significant, a finding that is consistent with the a priori expectation that firms with greater ratios of stockholders equity to total assets have higher accounting profitability, ceteris paribus. Column 2 of Table 2 summarizes the specification that includes the industry fixed effects dummy variables. In order to control for potential biases related to inter-industry differences in international activity, we include the trade intensity variable in the industry fixed effects model. We also include the leverage variable to avoid biases resulting from inter-firm differences in financing and the year dummies to control for business cycle effects. The R2 of 0.29 is in keeping with previous studies that have found substantial explanatory power from the industry fixed effects specification. Our finding of strong industry effects is consistent with previous research by Schmalensee (1985), Powell (1996) and others who found that industry effects dominate firm effects in explaining cross-sectional differences in profitability. A finding that industry effects are important in explaining cross-sectional profit rate variation for a sample of global firms provides tentative evidence regarding the importance of competition at the global level in determining the profitability of multinational firms. Estimated coefficients for the complete model that includes both firm level and industry effects can be found in column 3 of Table 2. The estimated coefficients and explanatory power for the complete model reinforce the conclusions based on the separate firm and industry effects models. The R2 for the complete model is 0.34. When compared with the R2 of 0.24 for the firm effects model in column 1, we see that the industry effects explain 10% of the total variation in rate of return. While explaining 10% of the total variation in profit rate is not comparable to the strongest industry effects found in previous studies (e.g. Schmalensee, 1985), our findings do suggest a meaningful role for industry (with geographic market defined as global) in explaining profit rate variations among Global Fortune 500 firms. Industry effects that explain 10% of total profit rate variation are especially meaningful in light of the very broad industry definitions provided by the UNIDO data. The market share variable, positive and significant in the firm effects model, remain positive and significant in the complete model. Keeping in mind that market share is measured at the global level (i.e. the firm’s percentage of total world industry output), the finding of a positive and significant coefficient for market share provides compelling evidence of competition at the international level. The
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
411
positive and significant coefficient for market share is particularly meaningful given the broad industry definitions. A consistently positive coefficient for market share, together with industry effects that explain 10% of the variation in rate of return, point to the importance of global competition in explaining variation in profitability. While our data do not permit us to directly test for the existence of international competition among large multinational firms, the market share and industry effects revealed in this study both indicate the presence of international competition. The coefficient for the 1992-year dummy variable is negative and significant in every specification (firm effects, industry effects and complete model). Our finding of a negative and significant coefficient for the year following the trough of the recession in the US is not surprising given the diversity of industries and countries represented in our sample. Our findings suggest, therefore, that there is substantial variation across industries and countries in terms of the timing of the business cycle. That the coefficients for the 1992 dummy variable is statistically different from zero in every model tested supports Domowitz et al.’s (1986) findings regarding the importance of business cycle effects in determining profit rate variation. Our results extend the findings of Domowitz et al. for US firms to multinational firms. The trade intensity variable was included as a right hand side variable in our regressions to capture the extent to which various industries are international in scope. The estimated coefficient for trade intensity in our models, however, is statistically insignificant in every specification. This finding is not surprising when analyzed relative to prior results regarding the relationship between profitability and exports. Earlier studies have alternately found no export effect on profitability (Caves & Uekusa, 1976; Jacquemin, de Ghellinck, & Huveneers, 1980; Pagoulatos & Sorenson, 1976a); a positive effect of exports on profitability (Caves, KhalilzadehShirazi, & Porter, 1975; Geroski, 1982; Lyons, 1981; Pugel, 1978); or a negative impact of exports on profitability (Jenny & Weber, 1976; Neumann, Bobel, & Haid, 1979; Pagoulatos & Sorenson, 1976b). Given the conflicting evidence from prior research, the lack of statistical significance for our trade intensity variable is not surprising. The potential for multicollinearity exists in a model that includes market share, firm size, capital intensity and trade intensity as regressors. Gujarati (2003) illustrates the impact of collinearity on coefficient variances for alternative correlations among regressors. Gujarati’s (2003) simulations suggest that correlations among regressors of 0.5 or less have relatively little impact on the estimated coefficient variances and t-statistics. Examining the correlation matrix reveals that the correlations among regressors in our model are all substantially less than 0.5 with the exception of a 0.54 correlation between market share and firm size. A 0.54 correlation among regressors implies an increase in the estimated coefficient variance of slightly more than one-third. Examining the estimated t-statistics for market share and firm size, it appears that the multicollinearity is of no consequence in this case. The coefficient for market share, our main variable of interest, is statistically significant in every model despite the presence of multicollinearity. Moreover, the
412
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
estimated t-statistics of 0.20 and 0.96 for the firm size variable in the firm effects and complete models, respectively, are sufficiently below critical values at the 0.05 level to suggest that slightly more than a one-third increase in the estimated coefficient variances is of no consequence. The overall conclusions from the various models presented in Table 2 include the importance of both firm market share and industry effects in explaining profit rate variation. These findings are very similar to previously published findings using domestic data. We believe that these findings regarding the importance of industry effects at the international level are among the first of their kind. A strong industry effect along with a positive and significant coefficient for global market share for Global Fortune 500 firms provide important evidence that there is rivalry among multinational firms operating in the same global industry. A variety of alternative explanations have been given for the existence of strong industry effects in cross-sectional rate of return models. In addition to the usual interpretation that strong industry effects are evidence of rivalry among firms in the same industry, many critics of the industry effects specification have maintained that strong industry effects may be capturing nothing more meaningful than similarities in accounting conventions among firms operating in the same broad industry. While industry-level similarities in accounting conventions may explain strong industry effects for firms headquartered in the same country, it is not likely that firms from the same industry but based in different nations would be expected to adopt similar accounting standards. It seems reasonable that differences in accounting practices across countries would dominate similarities in accounting practice for firms from the same industry. Our finding of strong industry effects using a sample of global firms provides important evidence that something other than industry level accounting conventions account for the strength of industry effects revealed in this and other studies. While there are alternative explanations for the presence of industry effects at the global level, one cannot discount the likelihood that our findings are evidence that multinational firms compete in global output markets.
6. Conclusion The purpose of this study was to test the proposition that firm market share and industry fixed effects, both measured globally, explain profitability variation for a sample of multinational firms. The most important finding is that a sample of Global Fortune 500 firms exhibits meaningful market share and industry effects, a finding that is consistent with prior studies using domestic data. While there are a number of explanations for industry effects, a common interpretation is that the presence of industry effects signals rivalry among firms. The positive and significant coefficient for market share in both the complete and firm effects models provides additional evidence to support the existence of rivalry among global firms. Our findings provide a critical first piece of evidence regarding rivalry among multinational firms. Moreover, our findings provide some measure of support for the US
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
413
Department of Justice’s most recent merger guidelines that establish international firms as a source of potential competition for the purposes of deciding whether to allow or disapprove mergers. Although globalization is not a new phenomenon, it is true that there has been a sharp increase in the importance of trade in manufactured goods in the last 20 years. This trend, together with the increasing importance of intra-industry trade suggests that the relevant geographic market for many manufactured goods is now global. Our empirical results supply initial evidence that global competition in manufactured goods industries is important. Given that these are the first empirical results of their kind, the usual call for additional evidence before using the findings in support of public policy would appear especially appropriate. Appendix A. Descriptive statistics and characteristics of the data Variable
Mean
Standard deviation
Return on assets Market share Firm size Capital intensity Leverage Trade intensity
4.01% 8.5% 14 326.97 1.07 0.34 9.52
6.0 18.09 22 221.97 0.54 0.18 9.57
The UNIDO has general responsibility for the collection and dissemination of international industrial statistics. The activities regarding industrial statistics include the compilation of manufacturing industry statistics at the four-digit ISIC level. Data are available at the country level and for world totals by industry. Variables included in the UNIDO database are: number of establishments, number of employees, wages and salaries, output, and value added. The UNIDO data are available for a maximum of 81 manufacturing industries, with the industry coverage varying by country. This paper makes use of the UNIDO database for world totals by manufacturing industry, which we use in calculating the global market shares of firms for the years 1989–1992. The Fortune Global 500 was first published in 1990, with data primarily for the previous calendar or fiscal year. The Global 500 listing includes the largest 500 multinational firms ranked by sales, with the requirement that all companies included must have obtained at least 50% of their sales from manufacturing and/ or mining. (In more recent years, the Global 500 has been expanded to include companies in banking and finance.) The Fortune data set includes the following variables: sales, profits after tax, assets, stockholders’ equity, and number of employees. We obtain the following firm level variables from the Fortune Global 500: sales, profits after tax, assets, stockholder’s equity, and the home country of the firm.
414
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
Appendix B. Industries included in sample ISIC code
Industry definition
3113 3116 3119 3122 3131 3133 3134 3140 3220 3411 3420 3511 3521 3522 3523
Canning and preserving fruits and vegetables Grain mill products Manufacture of cocoa, chocolate and sugar confectionery Manufacture of prepared animal feeds Distilling, rectifying and blending spirits Malt liquors and malt Soft drinks and carbonated water industries Tobacco manufactures Manufacture of wearing apparel except footwear Manufacture of pulp, paper and paperboard Manufacture of furniture and fixtures, except primarily metal Manufacture of basic industrial chemicals except fertilizer Manufacture of paints, varnishes and lacquers Manufacture of drugs and medicines Manufacture of soaps and cleaning preparations, perfumes, cosmetics and other toilet preparations Petroleum refineries Tyre and tube industries Manufacture of glass and glass products Iron and basic steel industries Non-ferrous metal basic industries Manufacture of Engines and Turbines Manufacture of agricultural machinery and equipment Manufacture of office, computing and accounting machinery Manufacture of electrical industrial machinery and apparatus Manufacture of radio, television, and communication equipment and apparatus Manufacture of motor vehicles Manufacture of motorcycles and bicycles Manufacture of aircraft Manufacture of professional and scientific and measuring and controlling equipment Manufacture of photographic and optical goods Manufacture of watches and clocks
3530 3551 3620 3710 3720 3821 3822 3825 3831 3832 3843 3844 3845 3851 3852 3853
References Amato, L., & Wilder, R. P. (1990). Firm and industry effects in industrial economics. Southern Economic Journal, 57, 93–105. Bain, J. S. (1956). Barriers to new competition. Cambridge: Harvard University Press.
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
415
Benston, G. J. (1985). The validity of profits-structure with particular reference to the FTC’s line of business. American Economic Review, 75, 37–67. Brush, T., Bromiley, P., & Hendrickx, M. (1999). The relative influence of industry and corporation on business segment performance: an alternative estimate. Strategic Management Journal, 20, 519–547. Caves, R. E. (1989). International differences in industrial organization. In R. Schmalensee, & R. D. Willig (Eds.), Handbook of industrial organization, Vol. II. Amsterdam: Elsevier. Caves, R. E. (1996). Multinational enterprise and economic analysis. (2nd ed.). Cambridge: Cambridge University Press. Caves, R. E., Khalilzadeh-Shirazi, J., & Porter, M. E. (1975). Scale economies in statistical analyses of market power. Review of Economics and Statistics, 57, 133–140. Caves, R. E., & Uekusa, M. (1976). Industrial organization in Japan. Washington: Brookings Institution. Co, C. Y. (2001). Trade, foreign direct investment and industry performance. International Journal of Industrial Organization, 19, 163–183. Cubbin, J., & Geroski, P. (1987). The convergence of profits in the long run: inter-firm and inter-industry comparisons. Journal of Industrial Economics, 35, 427–442. Demsetz, H. (1973). Industry structure, market rivalry, and public policy. Journal of Law and Economics, 16, 1–9. Domowitz, I., Hubbard, R. G., & Petersen, B. C. (1986). Business cycles and the relationship between concentration and price–cost margins. Rand Journal of Economics, 17, 1–17. Feenstra, R. C. (1998). Integration of trade and disintegration of production in the global economy. Journal of Economic Perspectives, 12, 31–50. Fisher, F. M., & McGowan, J. J. (1983). On the misuse of accounting rates of return to infer monopoly profits. American Economic Review, 73, 82–97. Gale, B. T., & Branch, B. S. (1982). Concentration versus market share: what determines performance and why does it matter? Antitrust Bulletin, 27, 83–106. Geroski, P. A. (1982). Simultaneous equations models of the structure–performance paradigm. European Economic Review, September, 145–158. Greene, W. H. (2000). Econometric analysis. (4th ed.). Upper Saddle River, NJ: Prentice Hall. Gujarati, D. N. (2003). Basic econometrics. (4th ed.). Boston, MA: McGraw Hill. Hout, T., Porter, M. E., & Rudden, E. (1998). How global companies win out. In M. E. Porter (Ed.), On competition (pp. 289–308). Boston: Harvard Business School Press. Jacquemin, A., de Ghellinck, E., & Huveneers, C. (1980). Concentration and profitability in a small open economy. Journal of Industrial Economics, 29, 131–144. Jenny, F., & Weber, A.-P. (1976). Profit rates and structural variables in French manufacturing industries. European Economic Review, 7, 187–206. Levinsohn, J. (1993). Testing the imports-as-market discipline hypothesis. Journal of International Economics, 35, 1–22. Lyons, B. (1981). Price–cost margins, market structure, and international trade. In D. Currie, D. Peel, & W. Peters (Eds.), Microeconomic analysis: essays in microeconomics and economic development. London: Croom Helm. Martin, S. (1993). Advanced industrial economics. Oxford: Blackwell. Mueller, D. C. (1990). The dynamics of company profits: An international comparison. Cambridge: Cambridge University Press. Neumann, M., Bobel, I., & Haid, A. (1979). Profitability, risk and market structure in West German industries. Journal of Industrial Economics, 27, 227–242. Pagoulatos, E., & Sorenson, R. (1976a). International trade, international investment, and industrial profitability of US Manufacturing. Southern Economic Journal, January, 424–434. Pagoulatos, E., & Sorenson, R. (1976b). Foreign trade, concentration, and profitability in open economies. European Economic Review, 8, 255–267. Powell, T. C. (1996). How much does industry matter: an alternative empirical test. Strategic Management Journal, 17, 323–334. Pugel, T. A. (1978). International market linkages and US manufacturing: price, profits and patterns. Cambridge, MA: Ballinger.
416
L.H. Amato, R.P. Wilder / International Business Review 13 (2004) 401–416
Pugel, T. A. (1980). Foreign trade and US market performance. Journal of Industrial Economics, December, 119–129. Ravenscraft, D. J. (1983). Structure–profit relationships at the line of business and industry level. Review of Economics and Statistics, 65, 22–31. Rumelt, R. P. (1991). How much does industry matter? Strategic Management Journal, 12, 167–185. Schmalensee, R. (1985). Do markets differ much? American Economic Review, 75, 341–351. Schmalensee, R. (1987). Collusion versus differential efficiency: testing alternative hypotheses. Journal of Industrial Economics, 35, 399–425. Scott, J. T., & Pascoe, G. (1986). Beyond firm and industry effects of profitability in imperfect markets. Review of Economics and Statistics, 68, 284–292. Shepherd, W. G. (1972). The elements of market structure. Review of Economics and Statistics, February, 25–37. Weiss, L. W. (1971). Quantitative studies of industrial organization. In M. D. Intrilligator (Ed.), Frontiers of quantitative economics. Amsterdam: North Holland. Wernerfelt, B., & Montgomery, C. (1988). Tobin’s Q and the importance of focus in firm performance. American Economic Review, 78, 246–250. Yamawaki, H. (1986). Exporting, foreign market structure and profitability: Japanese and US manufacturing. Review of Economics and Statistics, 68, 618–627. Yamawaki, H., Sleuwaegen, L., & Weiss, L. W. (1989). Industry competition and the formation of the European common market. In L. W. Weiss (Ed.), Concentration and price. Cambridge, MA: MIT Press.