Does Diversification Create Value for the Company? European Evidence.

Transcription

1 GHENT UNIVERSITY FACULTY OF ECONOMICS AND BUSINESS ADMINISTRATION ACADEMIC YEAR Does Diversification Create Value for the Company? European Evidence. Master thesis submitted to obtain the degree of Master in Business Economics Elke Verstraeten Maarten Wybaillie under the guidance of Dr. Olivier De Jonghe

2

3 GHENT UNIVERSITY FACULTY OF ECONOMICS AND BUSINESS ADMINISTRATION ACADEMIC YEAR Does Diversification Create Value for the Company? European Evidence. Master thesis submitted to obtain the degree of Master in Business Economics Elke Verstraeten Maarten Wybaillie under the guidance of Dr. Olivier De Jonghe I

4 CONFIDENTIALITY CLAUSE PERMISSION The undersigned declare that the contents of this masters dissertation can be used and/or consulted and/or reproduced, provided that the sources are quoted. Elke Verstraeten Maarten Wybaillie II

5 ACKNOWLEDGMENTS We would like to thank Olivier De Jonghe, Riet De Baets, Koen Berteele, and our parents and friends for their comments and suggestions. The research in this paper was conducted while the authors were master students at the Ghent University. III

6 TABLE OF CONTENTS Abbreviations Used... V List of Tables... VI Abstract... VII 1. Introduction Literature Review Geographic Diversification Industrial Diversification Combined View European Studies Sample Selection and Methodology Sample Frame and Sample Description Measures Descriptive Statistics Method of Analysis Multivariate Analysis Main Results Sensitivity and Robustness Tests Discussion and Interpretations Conclusion Summary Limitations and Guidelines for Further Investigation References... VIII List of Appendices... XI IV

7 ABBREVIATIONS USED BVL BVTA capex EBIT e.g. EU15 GD ID MTB MVE Obs. OLS R&D SIC U.S. U.K. Book Value of Liabilities Book Value of Total Assets Capital Expenditure Earnings Before Interest and Taxes example given European Union with 15 member states Geographical Diversification Industrial Diversification Market to Book Market Value of Equity Observations Ordinary Least Squares Research & Development Standard Industrial Classification United States of America United Kingdom V

8 LIST OF TABLES Table 1: Concise Summary of Literature Review Table 2: Geographical and Industrial Distribution of the Sample Table 3: Descriptive Statistics of Sample for MTB Value Measures Table 4: Distribution of MTB Value across Diversification Categories Table 5: Multivariate Test for Diversification Value Impacts Table 6: Geographical and Industrial Distribution of the Sample (Sensitivity and Robustness Tests, part 1) Table 7: Multivariate Test for Diversification Value Impacts (Sensitivity and Robustness Tests, part 1) Table 8: Geographical and Industrial Distribution of the Sample (Sensitivity and Robustness Tests, part 2) Table 9: Multivariate Test for Diversification Value Impacts (Sensitivity and Robustness Tests, part 2) APPENDICES: Table 10: Geographical and Industrial Distribution of the Sample... XII Table 11: Regression details for sector, country and year dummies Model IV... XIII Table 12: Regression details for sector, country and year dummies Model V...XIV VI

9 Does Diversification Create Value for the Company? European Evidence. ABSTRACT This paper examines the impact of geographical and industrial diversification on firm value for a sample of European companies. During the period 1996 till 2008, this results in observations. Confirming the predictions of most theories, a geographic (industrial) diversification premium (discount) of 10,2% and 7,6% respectively, is found. Furthermore, the extension of the research model with an interaction coefficient shows that being doubly diversified has a positive impact on firm value. In addition, an interesting comparison between diversification within European firms across European borders and American firms across the U.S. borders, leads to think that diversification is valued higher in European firms. VII

10 1. INTRODUCTION Since the 1980 s, academic and business communities have had substantial interest in the diversification discount. Lang & Stulz (1994) found that multi-activity firms trade at an average discount relative to firms that focus on a single activity. This finding was the starting point of an active debate about the impact of corporate diversification in both its geographical and industrial dimension on the value of a company. Theoretical argumentation leads to value-enhancing as well as value reducing effects, associated with both forms of corporate diversification. The potential benefits of a firm active in multiple lines of business and/or in different countries include lower taxes, economies of scale, the possibility to spread risks, a greater debt capacity, a preference of investors for diversity, and a greater operating efficiency and flexibility. According to Berger & Ofek (1995), the potential costs of diversification include the use of increased discretionary resources to undertake value-decreasing investments, cross-subsidies that allow poor segments to drain resources from better-performing segments, and miss-alignment of incentives between central and divisional managers (Berger & Ofek, 1995, p.40). There is no clear prediction about the overall value effect of diversification, and empirical research in the area of corporate diversification has not led to an univocal opinion yet. However, the impact of a potential advantage or disadvantage caused by diversification, is of growing importance nowadays because of increasing globalization. Industrial diversification is more often subject of study than geographical diversification. Different authors such as Bodnar, Tang & Weintrop (1999), and Barnes & Brown (2006), tried to fill up this gap by making different models that measure the value impact of both geographic and industrial diversification. Literature on the value impact of diversification decisions has focused on U.S. and U.K. firms and has rarely included an interaction coefficient to observe the influence of different diversification combinations. In addition, there are only a few studies with a European sample (Moerman (2008), Joliet & Hubner (2008)), but there is no record of any study which investigates the impact of diversification on the value of European firms. The initial aim of this paper is to exploit this gap by examining the overall impact of being doubly diversified, as well as the independent impact of geographical and industrial diversification on firm value. The research is rooted on methodologies of 1

11 Bodnar, Tang & Weintrop (1999), and Barnes & Brown (2006). By following a long tradition in American and U.K. diversification research, their methodologies provide a model to measure the impact of industrial and geographical diversification on a firm s value, measured by Market To Book (MTB). This paper estimates the value impact of both forms of diversification, for a sample of European companies over the period , that enhances firms which equals observations. The MTB analysis of the European market provides evidence of a significant geographic (industrial) diversification premium (discount) of approximately 10,2% and 7,6% respectively. In addition, the analysis shows an important impact on a firm s value of being doubly diversified: the interaction coefficient provides a value increase that even compensates the industrial diversification discount. Furthermore, the impact and sensitivity of various changes in definitions have been explored. These controls confirm the robustness of the model. In addition, the impact of diversification outside Europe on the value of European firms is studied. The results obtained by this research have important implications for the literature about diversification. First, this study is one of the rare studies about diversification with a fully European sample. Second, it measures the value impact of both geographical and industrial diversification, because of their mutual dependence. Furthermore, this study introduces an interaction coefficient to measure the impact of being doubly diversified on a firm s value. The results of that interaction coefficient prove the usefulness of investigating the impact of being doubly diversified. An interesting comparison of the geographical diversification premium between Europe and the U.S. is made. The results of this paper can help managers in their diversification decision process. The remainder of this paper is organized as follows: Section 2 summarizes the extent literature related to geographic and industrial diversification. Section 3 describes the sample selection and the adopted methodology. Section 4 presents the results of the main analyses and several robustness checks. The final section summarizes, states the limitations and concludes. 2

12 2. LITERATURE REVIEW The literature on both geographic and industrial diversification is extensive. In the combined view literature, Bodnar, Tang & Weintrop (1999), and Barnes & Brown (2006) are some of the most investigated authors. They provide a relativily complete literature overview of both forms of corporate diversification. In case of unclear information and missing references, we will try to complete their literature view Geographic Diversification Empirical research into the value impact of geographic diversification has a rich history. Several synonyms for geographic diversification, such as global diversification or international diversification are used. In the literature about geographic diversification, four general reasons for a company to diversify geographically are studied. These reasons are listed by Bodnar, Tang & Weintrop (1999). First, geographic diversification has its roots in studies concerning foreign direct investment: Because of the imperfection of markets, assets cannot be sold for their internal value (Caves (1971), and Hymner (1976)). Consequently, firms have to invest abroad to exploit firm-specific assets and to obtain rents on these assets. Internationalizing the firms can lead to economies of scale of specific assets, such as marketing and research and development. If so, the incrementing size of the firm s activities using these specific assets, will cause a value increase of the firm. The internalization theory of multinational firms also argues that direct international investment occurs (so that the value of a firm increases) when markets internalize their information-related intangible assets with public good properties. Secondly, geographic diversification can create value through the operational flexibility of a multinational corporate system (Kogut (1983)). An unknown international environment causes a lot of uncertainty e.g. demand shocks are not perfectly correlated. Consequently, a geographically diversified network will give the firm the possibility to exploit market conditions, and this network will add additional value to the firm. On the contrary, Reeb, Kwok & Back (1998) provide evidence of a significant positive relationship between systematic risk and international 3

13 diversification. This means that negative influences, which increase overall firm volatility of returns, dominate the international diversification benefit of reduced cash flow correlation which causes an increase of systematic risk. Their study therefore, suggests a value discount for internationally diversifying firms. Thirdly, because of its ability to arbitrage institutional restrictions, such as tax codes and financial restrictions, geographically diversified firms can be more valuable. In contrast to earlier empirical work, which generally focused on financial performance rather than value, Errunza & Senbet (1981) were the first authors to examine the empirical implications of geographic diversification for firm value. Their research leads to a significant positive relation between excess value and international activity. Multinational firms which operate in multiple geographic locations, show remarkable possibilities to make value-maximizing conditional decisions. As a result, the expected cash flow of diversified firms will increase compared to the expected cash flow of domestic firms. In 1984, they re-examined their research question on a larger database and looked into the effect of firm size, using different measures of international activity. Errunza & Senbet (1984) again find a positive correlation between geographical diversification and excess value. These studies cannot provide an estimate of the geographical diversification discount because they only study multinational firms. Finally, value from corporate geographic diversification can be created by investor preferences. Morck & Yeung (1992) confirm the internalization theory in an event study test. They explain why investors judge diversification to be expensive: to investors, multinational firms represent a portfolio of geographically spread companies as a claim on a collection of profit streams from various areas of the world. Thus, investors should be willing to pay more for shares of global firms for providing this service. This premium is a cause of the increased value of diversified firms. In addition Fauver, Houston & Naranjo (2003) argue that, if there is a lack of shareholder protection in domestic markets and when external financing is difficult, internal capital markets will be particularly valuable for diversified firms. Under these circumstances, there can be a positive impact on the value of a company as the benefits of corporate diversification outweigh the agency-related costs. This premium 4

14 contrasts the diversification discount among high-income countries where capital markets are integrated and well developed, so that external funding is not difficult. This paper will study the impact of diversification on the value of a company. Several studies already showed that geographic diversification can enhance the value of a firm. To reflect the benefits which are only available for geographically diversified firms (for instance economies of scale of specific assets), the value of these firms should increase. More general, the value of geographically diversified firms should be higher than the value of domestic firms. Research on this subject by Kogut & Kulatilaka (1994) concludes that the value of a geographically diversified firm should be increasing with certain characteristics. These characteristics can be both global manufacturing and production shifting as flexibility options. Being operational across different regulatory and consumer markets as well as the volatility of the environment in which it operates, can lead to a higher value for companies. Nevertheless, there are also several studies that show a negative impact of geographic diversification on the firm s value. The dominant logic behind these studies is that the principal-agency problem will increase when the organization becomes more complex. Shareholders seek value maximization in contrast to managers, who act in their own interest. A common solution to solve this incentive problem is giving equity stakes to managers. This results in a higher concern about the firm s specific risk. As a result, managers will favor geographic diversification because it reduces the firm specific risk, even if it results in a lower shareholder value. A recent study of Doukas & Kan (2006) supports a lower shareholder value of geographical diversification in a contingent claims framework. By using a database of 355 U.S. acquisitions during the period 1992 till 1997 with 612 observations, they confirm that more foreign involvement increases bondholder value while it decreases shareholder value. More precisely, they explain it as follows: the univariate analysis results indicate that globally diversified bidders trade at a discount regardless of their industrial structure (Doukas & Kan, 2006, p.358). Their multivariate analysis results also indicate that global diversification harms shareholder value. Furthermore, they provide strong evidence in support of the risk-reduction hypothesis of global diversification, which leads to an increase in bondholders wealth. Both results are in line with the contingent claims theory that predicts that global diversification has a 5

15 positive impact on bondholders' wealth while it has a negative influence on shareholders value, in this case a global diversification discount. They conclude that geographical diversification does not decrease the overall value of a firm. Foregoing arguments lead to zero hypothesis 1: H 01 : Geographic diversification has no impact on the value of European companies. For this zero hypothesis, both positive and negative alternative hypotheses will be tested Industrial Diversification For many years, it was taken for granted that industrial diversification was the key to success because of a greater operating efficiency, the possibility to spread risks, a greater debt capacity, and lower taxes. Since the diversification boom in the 1960s, academics are fascinated by the question of the impact of industrial diversification on the value of a company. As mentioned in the introduction, the impact of industrial diversification on a firm s value has been more thoroughly examined than the impact of geographic diversification. The cornerstone of the literature about industrial diversification is a paper written by Lang & Stulz (1994). They are the first to focus on the value impact of industrial diversification and demonstrate a negative relation between the value, as measured by both Tobin s q and Market To Book, and industrial diversification in the U.S. Berger & Ofek (1995) use the market value of industrially diversified firms to measure the overall value impact of industrial diversification. They compare it to the sum of the imputed value of each industrial segment and register a value loss. The explanation for this are problems of over-investment in industries with low growth opportunities and cross-subsidizing of loss generating activities. Such a value loss will be smaller when the diversified firms stay in the same sector (when they have the same 2-digit SIC code) and when they take profit out of tax benefits of diversification. Servaes (1996) conducts his research about the value impact of being active in different segments during the diversification boom from 1960 till He finds no evidence for a premium. On the contrary, he even finds a waning diversification discount. When the discount was still large during the 1960 s, firms with low insider 6

16 ownership were more inclined to diversify. This effect reversed during the 1970 s when the discount declined to zero. A few years later, Rajan, Servaes & Zingales (2000) modeled the internal power fights by the allocation of resources between divisions of an industrially diversified firm. They conclude that funds will be transferred from divisions with poor opportunities to divisions with good opportunities. Nevertheless, higher levels of diversification might harm these transfers, leading to inefficient investments. This misallocation of funds will destroy value through overinvestment in value-destroying projects (infra, p.10). Furthermore, Brusco & Panunzi (2000) show that this diversification discount will not necessarily be eliminated by ex-post allocations of funds. Moreover, they prove that asymmetries in size and growth prospects increase the diversification discount. Graham, Lemmon & Wolf (2002) find no such a discount in their study about industrial diversification. According to them, the excess value reduction occurs because of acquiring already discounted business units and not because diversifying destroys value. Campa & Kedia (2002) find a strong negative correlation between a firm s choice to diversify and its value. In their own words: firms that choose to diversify have a higher value than existing firms in their industry and lower value than other firms in the industry that remain focused (Campa & Kedia, 2002, p.1759). Villalonga (2004a) uses a unique new database that covers the whole U.S. economy and shows a diversification premium, which is robust to variations in sample, business unit definition and measures of excess value and diversification. In a second study, Villalonga (2004b) points out that firms do not randomly become diversified, but rather endogenously choose to do so. Her study shows that diversified firms trade at a discount prior to becoming diversified. However, when controlling the self-selection bias in diversified firms, the discount disappears. In addition, Bohl & Pal (2006) find a diversification premium of 30%, in contrast with previous findings stressing the agency problem of U.K. conglomerates. Their sample contains all constituent companies of the FTSE all-share index listed on the London Stock Exchange over the period and de-listed companies (both diversified and focused firms) that provide balance-sheet, profit and loss statements for the selected period. A comparison between U.S. studies and their own U.K. study 7

17 indicates a major cause for such a diversification premium. While U.S. studies explain the diversification discount of conglomerates relative to focused firms by firmspecific characteristics, they find significant macroeconomic effects for the U.K. conglomerates. More precisely, less favorable macroeconomic conditions hinder firms growth, decrease their market value and affect positively their decision to operate as a diversified firm (Bohl & Pal, 2006, p22). However, recent findings of Mackey & Barney (2005) tend to support the original conclusion that unrelated acquisitions can reduce firm value. In their study, a comparison is made between the diversification decision versus the decision to pay dividends or repurchase firm stock. They find that diversification destroys value when compared to alternative payout policies. Their result is robust to the use of econometric techniques that control self-selection of the diversification decision. Furthermore Gomes & Livdan (2004), Schoar (2002), and Maksimovic & Philips (2002) investigate the productivity of conglomerates and stand-alone firms as a result of industrial diversification. Because these studies do not handle real value creation, their relevance is limited for the empirical study in this paper. Hence, zero hypothesis 2 states: H 02 : Industrial diversification has no impact on the value of European companies. For this zero hypothesis, both positive and negative alternative hypotheses will be tested Combined View None of the papers on industrial diversification in the literature review above consider geographic diversification, nor do any of the papers on geographic diversification consider industrial diversification. In the literature review above, the authors could not give an univocal conclusion about the existence of a diversification premium or a diversification discount. Their conflicting results and interpretations can be caused by the bias in the estimated effect of diversification on performance across a large variety of industries (Santalo & Becerra (2008)) on the one hand and the estimated value impact of industrial diversification for studies about geographic diversification (and vice versa) if these two phenomena are related on the other hand. According to Bodnar et al. (1999) One must consider both forms of 8

18 diversification simultaneously in order to generate an unbiased estimate of the impact of industrial diversification on firm value. Such an approach is also necessary to obtain an unbiased estimate of the value impact of geographic diversification. (Bodnar, Tang & Weintrop, 1999, p.8). In spite of the large amount of studies without a combined view, there are two leading studies with a combined view on the diversification topic. In a non-published working paper Bodnar et al. (1999) examined the value impact of diversification using a framework that controls both forms of corporate diversification. They used the basic models of Errunza & Senbet (1981) and Lang & Stulz (1994) on a sample of 7000 U.S. firms for the period 1984 to They report evidence of a 2,7% value premium for geographic diversification and a 6% value discount for industrial diversification. In 2006, Barnes & Brown (2006) exploit the Lang & Stulz (1994), Berger & Ofek (1995), and Bodnar et al. (1999) methodologies on a sample of U.K. firms. They control the form of diversification in assessing the value impact on their U.K. sample for the period , and report evidence of a 14% perverse geographic discount and no systematic industrial value impact. The combined view has become a hot topic in today s research about diversification. Denis, Denis & Yost (2002) use financial information of U.S. firms from 1984 till 1997 and their selection results in firm-years associated with firms. They find a rise in the scope of geographical diversification over time. However, they remark that this boost in geographical diversification does not come from a substitution of geographical by industrial diversification. Furthermore, their estimation of Ordinary Least Squares (OLS) regressions of excess value on dummy variables leads to the conclusion that the discounts for both forms of diversification are approximately the same in size. When looking at the effect of changes in diversification, they find that increasing the scope of geographical diversification reduces excess value while a reduction of the scope increases excess value. They conclude that the gains of geographical diversification are more important than the costs. Robustness tests prove that the discount associated with geographical diversification has remained relatively stable over time. By contrast, the value 9

19 discounts associated with industrial diversification decline over time. Similarly, the discount for being both industrially and geographically diversified declines. Fauver, Houston & Naranjo (2004) investigate the value of industrial and international diversification for more than firms in Germany, the U.K., and the U.S. In line with Lang & Stulz (1994), Berger & Ofek (1995), and Rajan et al. (2000), they find an industrial diversification discount in the U.K. and the U.S. Furthermore, they find, just as Denis et al. (2002), that U.S. multinationals trade at a discount relative to firms operating only in the domestic market. This result is robust to different specifications and to different benchmarks used to estimate the value of diversification. On the contrary, they find no such discount for U.K. or German firms. International diversification has no effect on their firm value. There are two possible explanations for this result. Maybe the benefits of diversifying overseas are smaller for U.S. firms and/or the agency and coordination costs of multinational expansion are larger for U.S. firms. In the robustness test, Fauver et al. (2004) control agency costs associated with ownership concentration as suggested by Morck et al. (1988) and Servaes (1990). The effects of ownership concentration are significantly different for focused and diversified firms, and these effects also vary across the three countries. These results suggest that ownership concentration and excess value are linked and that this link varies for focused and diversified firms. In contrast to the majority of studies about diversification, Freund, Trahan, and Vasudevan (2007) use a case study to test the impact of increases in global and industrial diversification on firm value and operating performance directly. The sample they use represents 194 U.S. firms that acquired foreign companies between 1985 and They base their investigation on a recent trend: On the one hand, U.S. firms have greatly expanded overseas operations in the past two decades. But, at the same time, there has been a tendency for firms to divest unrelated assets and to focus on core businesses, in other words, to reduce industrial diversification. (Freund, Trahan, & Vasudevan, 2007, p.159). Their findings lead to several generalizations. First, announcement period returns are significantly positive for the acquirers. The stock-price reaction is greater for firms with fewer growth opportunities and not significant for acquisitions by high-growth firms. Secondly, acquirer firms with fewer growth opportunities, as measured by Tobin's q, create more value than do firms with more growth opportunities. And thirdly, announcement-period returns and changes in operating performance are lower for 10

20 firms that increase their global, industrial, or both forms of diversification. After crosssectional regressions, they conclude that changes in operating performance from pre- to post-merger are lower for the firms that increase their global, industrial, or both forms of diversification. Gao, Ng & Wang (2008) use a database that contains public and private companies worldwide. They make a distinction between single-segment firms and multi-segment firms on the one hand, and domestic and geographically diversified firms on the other hand. Nevertheless, in their database, they don t look for an impact of industrial diversification on the value of the firms. The reason for this distinction is a supposed correlation between being geographically diversified and being industrially diversified. This correlation can cause a bias and that is why their regression models control both industrial and global diversification. In addition, these regression models also control other possible determinants of firm valuation. For example, they include leverage as a proxy for any financing benefits or costs of being geographically diversified. They also take R&D and advertising expenditure as a proxy for a firm's proprietary assets. They find that being geographically diversified affects the value of a firm. Firms with subsidiaries located in different regions of the United States, in other words, geographically diversified firms, experience a valuation discount of 6,2%. This geographic diversification discount increases when firms expand their operations to different regions nationwide. In general they conclude that the geographic location of a company is an essential component of corporate policies and that it has important market valuation implications. Hence, zero hypothesis 3 states: H 03 : The combination of geographic diversification and industrial diversification has no impact on the value of European companies. For this zero hypothesis, both positive and negative alternative hypotheses will be tested. 11

21 2.4. European Studies Table 1 gives an overview of the papers discussed above. Almost all studies about the impact of diversification on the value of a firm, have U.K. or U.S. samples. There is no record of any study which investigates the impact of diversification on the value of European firms. Nevertheless, there are a few studies about diversification with a European database. One of them is a study of Moerman (2008). In his study, Moerman examines the impact of the harmonization of fiscal and economic policies within the European Monetary Union (EMU) on the economies of member countries. He adopts a mean-variance approach and he finds strong evidence that diversification over industries yields more efficient portfolios than diversification over countries. Nevertheless, the study of Moerman has not the same idea of value creation as Lang & Stulz (1994) or Bodnar et al. (1999) So, further elaboration of his research has little importance for this study. Another recent study about diversification with a European sample, is from Joliet & Hubner (2008). Based on a sample of 598 firms, spread over 9 countries, they analyze the impact of corporate international diversification on domestic and world betas through the notion of psychic distance between countries. Because they do not analyze the impact of diversification on the value of companies, further elaboration of their research is less important for this study. 12

22 Table 1: Concise Summary of Literature Review Notes: This table provides a summary of the papers discussed in the Literature Review about geographic and industrial diversification. More precisely, the source, the sample and the impact on the value of a company are given in a concise format. GD stands for Geographical Diversification. A firm is geographically diversified when it is established in more than 1 country or region, depending on the author. ID stands for Industrial Diversification. A firm is industrially diversified when it is operating in more than one sector. The method of stipulating sectors depends on the author. GD ID author source sample premium discount premium discount Caves (1971), Hymner (1976) Economica, MIT Press No empirical study x Errunza & Senbet (1981) Journal of Finance U.S. multinational firms ( ), 236 observations x Kogut (1983) Sloan Managment Review No empirical study x Errunza & Senbet (1984) Journal of Finance U.S. multinational firms ( ), 402 observations x Morck & Yeung (1992) Journal of International Economics U.S. firms (1987) x Kogut & Kulatilaka (1994) Management Science No empirical study x Lang & Stulz (1994) Journal of Finance, Journal of Political Economy U.S. firms ( ), observations (excl. smaller firms: observations) x Berger & Ofek (1995) Journal of Financial Economics U.S. firms ( ), observations x Servaes (1996) Journal of Finance U.S. firms ( ) (x) Reeb, Kwok & Back (1998) Graham, Lemmon & Wolf (2002) Journal of International Business Studies public firms ( ) x Journal of Finance 356 acquisitions ( ) No disc ount Bodnar, Tang & Weintrop (1999) Unpublished working paper U.S. firms ( ) x x Rajan, Servaes & Zingales (2000) Journal of Finance U.S. firms ( ), observations x Brusco & Panunzi (2000) Unpublished working paper No empirical study x Campa & Kedia (2002) Journal of Finance U.S. firms ( ), observations x 13

23 Table 1 (continuation): Concise Summary of Literature Review GD ID author source sample Denis, Denis & Yost (2002) Fauver, Houston & Naranjo (2003) Fauver, Houston & Naranjo (2004) Villalonga (2004a) Villalonga (2004b) Mackey & Barney (2005) Barnes & Brown (2006) Doukas & Kan (2006) Journal of Finance Journal of Financial and Quantitative Analysis Journal of Corporate Finance Journal of Finance Financial Management Unpublished Working Paper Journal of Business Finance and Accounting Journal of International Business Studies U.S. firms ( ), observations more than firms from 35 countries ( ) more than firms, Germany, U.K., U.S. ( ) firms ( ), observations U.S. firms ( ), observations No empirical study 495 U.K. firms ( ), observations 355 U.S. acquisitions ( ), 612 observations premium x discount x No impact Depends on value metric No disc ount premium x x discount x No disc ount x No impact Bohl & Pal (2006) Unpublished Working Paper 796 U.K. firms ( ), observations x Freund, Trahan, & Vasudevan (2007) Financial Management 194 U.S. acquiring industrial firms ( ) x x Gao, Ng & Wang (2008) Journal of Corporate Finance U.S.-based firms ( ), observations x 14

24 3. SAMPLE SELECTION AND METHODOLOGY 3.1. Sample Frame and Sample Description The population of this study is defined as all European companies. The sample is constructed by applying the criteria of different authors to the longest possible period for which business segment data are available and comparable. Consequently, in the first place, all listed, non-financial European (EU15) companies as recorded in Amadeus are taken. In line with Campa & Kedia (2002), and Graham, Lemmon & Wolf (2002) (referencing Ofek & Berger (1995)) non-financial firms are defined as having no primary or secondary SIC codes in the range of to Firms with segments in the financial sector are excluded because the valuation methods used in this paper have proven to be problematic for these firms. Specifically, earnings before interest and taxes (EBIT) are not meaningful for financial companies. The EU15 is defined as the member countries in the European Union prior to the accession of ten candidate countries on May 1, 2004 and enhances the following 15 countries: Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, Netherlands, Portugal, Spain, Sweden, United Kingdom (OECD, Main Economic Indicators, Paris). This leads to firms during the period In the second place, non-small companies as recorded in Datastream are taken. According to Campa & Kedia (2002), Graham, Lemmon & Wolf (2002), Villalonga (2004b), and Denis et al. (2002) (referencing Ofek & Berger (1995)) nonsmall firms are defined as having a total sales amount of more than $20 million per year. This is more or less equal to 20 million per year, following the exchange rate of To avoid distortions by ratio calculations from firms with sales or assets close to zero, firms will be required to have minimum sales of 20 million. From the firms of Amadeus, 435 firms had no full data in Datastream. After the restriction of total sales, another firms where lost, making for a sample of companies. During the period of 1996 till 2008, this results in observations. 1 The average exchange rate in 2002 equals $1 = 1,

25 Outliers in the data are modified using the winsorizing technique: all reported variables are 1% winsorized, which means that all numbers outside the first and the 99 th percentile are confined to the 1-respectively 99-percentiel number. Table 2 reports the number of firms in the sample classified by country and by industry (at level one SIC code). This information can be used to take an in-depth look at the geographical and industrial distribution of the sample. When looking at the total number of observations of the different countries, one can state that the U.K. (2 631 observations), Germany (2 291 observations) and France (1 874 observations) have most observations. This can be explained by the size and the higher level of economic activity of these countries. There are ten different sectors, reported at level one SIC code. It has to be pointed out that the financial sector is excluded in the sample based on the SIC codes reported in Amadeus. Furthermore, looking at the details of the firms 2, only Italy has observations in the sector public administration and other (PAO). These two observations come from the company A2A, which is a listed Italian based conglomerate that provides energy, district heating, waste management, networks and other services to several countries. The primary and only sic-code is 9 600, which raised some suspicion and might be due to differences in reporting standards. Omitting these observations has no impact on the results. Besides, during the past decennia, agriculture gave way to the tertiary industry in West Europe. This is well reflected in the sample where the sector agricultural, forestry, and fishery products (AFF) counts 14 firms, which is less than 1% of the total observations. Considering the number of observations, this research has a sufficiently large dataset to perform cross-country as well as cross-sector analyses. Furthermore, with this sample, an average of approximately 6,5 observations per company is reached. The sample implies an attrition rate of 50,2% for individual firms. This attrition rate is larger than the rate of Barnes & Brown (2006), which is 35%. (Barnes & Brown, 2006, p.1 514) A possible explanation would be the difference in period: Barnes & Brown (2006) have a five-year period of observation; this study covers a thirteen-year period. 2 More details about the geographic and industrial distribution of the firms of this sample, can be found in Appendix I, Table

26 Table 2: Geographical and Industrial Distribution of the Sample Notes: This table reports the number of observations in the sample classified by country and by broad industry. The last row and column provides the number of firms. Country codes (two letters) are: AT = Austria; BE = Belgium; DK = Denmark; FI = Finland; FR = France; DE = Germany; GR = Greece; IE = Ireland; IT = Italy; LU = Luxembourg; NL = the Netherlands; PT = Portugal; ES = Spain, SE = Sweden; GB = United Kingdom. SIC codes (three letters) are: AFF = agricultural, forestry, and fishery products; MCP = mining and construction products; LMP = light manufactured products; HMP = heavy manufactured products; TCE = transportation, communications, electric, gas, and sanitary service; WTR = wholesale trade; FIR = finance, insurance, and real estate; SER = services; HSE = health services; PAO = public administration and other. In the table, the level one SIC-code are indicated between brackets. Source: The sector data are taken from Amadeus, the country data are taken from Datastream. SECTOR COUNTRY AT BE DK FI FR DE GR IE IT LU NL PT ES SE GB (0) AFF (1) MCP (2) LMP (3) HMP (4) TCE (5) WTR (6) FIR (7) SER (8) HSE (9) PAO # Obs # Firms # Obs. # Firms

27 3.2. Measures In this paper, the impact of diversification on the value of a company is examined. The value of a firm is expressed by the variable Market To Book (MTB). Consistent with Lang & Slutz (1994), MTB is defined as follows: MTB i,t = (MVE i,t + BVL i,t ) / BVTA i,t (1) In other words, the Market To Book ratio for firm i in year t is equal to the Market Value of Equity for firm i in year t plus the Book Value of Liabilities for firm i in year t divided by the Book Value of the Total Assets for firm i in year t. To observe the impact of diversification, a distinction is made between the two most common diversification forms. Each company is examined on whether it is geographically diversified, industrially diversified, none of them or both of them. A firm is geographically diversified (multinational) when it has one or more subsidiaries established in a different country than its (registered) European country of origin. A firm is industrially diversified (multi-activity) when it is operating in more than 1 sector. Consistent with Bodnar et al. (1999), Denis et al. (2002), Doukas & Kan (2006), and Gao et al. (2008) different sectors are indicated by their four level SIC code. A firm is as well industrially and geographically diversified (doubly diversified, fully diversified) when it satisfies both descriptions above. In the analysis part of this paper, GEOG is the proxy for geographic diversification for each firm, INDUST is the proxy for industrial diversification for each firm and GEOGxINDUST is the proxy for fully diversification for each firm. The control variables measure size, leverage, profitability, investments, volatility, sector and country characteristics. Profitability is measured by EBIT and investments by capex. 18

28 3.3. Descriptive Statistics The description of the variables is based on the method elaborated by Barnes & Brown (2006). To reproduce diversification details, they made a compilation of the various methods used by, among other, Berger & Ofek (1995) and Doukas & Kan (2006). In performing this review, medians will be empasized rather than means, in line with Berger & Ofek (1995), and Barnes & Brown (2006). Table 3 displays the descriptive statistics for the sample used for the MTB value measure. There are four different combinations of diversification: two forms of industrial diversification namely single- and multi-activity, and two forms of geographical diversification; domestic and multinational companies. There appears to be an abundance of industrially diversified companies: multi-activity firms represent 71% of the total observations. More specific, multinational, multi-activity firms dominate the observations with 55,6% of the total observations. Furthermore, European industrially diversified firms on average are active in four different sectors. European geographically diversified firms on average are active in ten different countries. When zooming in on the impact of diversification, the information about the variables in Table 3 can be discussed. Keeping in mind companies with sales less than 20 million per year have already been excluded from the sample, diversification seems to have a positive correlation with the size of a company, either measured by total assets or total sales. Especially geographic diversification has a great influence. In other words, single-activity, multinational companies (YN), (respectively multiactivity, multinational companies (YY)) have a sales volume of 192 million ( 202 million), and total assets medians of 230 million ( 208 million). On the contrary, companies that are not globally diversified (NN, NY) have total assets and sales volumes that are below 100 million. In addition, the test of differences indicates that there is no statistically significant difference in size of geographic diversified firms, whether industrially diversified or not (YN-YY is not statistically significant). Next, following the co-insurance theory, diversified firms would be more leveraged. By examining the results, one can conclude that it is the case for geographically diversified firms: domestic firms are not significantly different (NN-NY is not statistically significant), and the median value of leverage for multinational firms is clearly higher even if this effect is not as outspoken as with size. Thirdly, there is a positive relation between EBIT to sales and geographic diversification. However, 19

29 industrial diversification does not influence EBIT to sales: the profitability medians of domestic firms are not significantly different, the same can be said about multinational firms. Fourthly, capex to sales has a different relationship with diversification: being industrially diversified (NY, YY) is not statistically different from having a pure focus (NN). Only single-activity, multinational companies have a distinct higher level of investments (capex to sales). Furthermore, diversification has no impact on R&D to sales: the median is zero for all combinations of diversification. And last but not least, for volatility there is no statistical difference between the combinations of diversification (NN-YN, NN-NY, YN-NY, and YN-YY are not statistically significant). Overall, geographical diversification seems to have a positive impact on the value of a company. Geographically diversified firms (YN, YY) have a higher sales volume, higher total assets, a slightly higher leverage, and a better profitability. Table 4 provides the distribution of the MTB ratio across the four different combinations of diversification. The results suggest that being geographically diversified has a positive value impact on the median values of MTB: the MTB value of multinational firms (YN, YY) is higher than the MTB value of domestic firms (NN, NY). These results are statistically significant on a 1% level for single- as well as multi-activity firms. Furthermore, focusing on a single-activity has a positive impact on the MTB values, although this effect is not significant for multinational companies. The medians from domestic, multi-activity firms (NY) are below those of focused firms (NN). As a conclusion, the row and column tests indicate that industrial focus and geographical diversification are both value enhancing effects. The diagonal tests, displayed in Table 4, can provide more information on the relative strengths of these value enhancing effects. First, the diagonal test statistics on the left indicate that domestic, multi-activity firms are statistically different from multinational, single-activity firms (NY-YN is statistically significant). Thus, one can conclude that domestic multi-activity firms (NY) have a lower value than multinational single-activity firms (YN). This result could have been expected given that domestic industrial diversified firms combine both value enhancing effects. Second, the diagonal test statistics on the right indicate that focused firms are statistically different from doubly diversified firms (NN-YY is statistically significant). Comparing domestic, single-activity firms with doubly diversified firms results in a significant value surplus 20

30 for firms that are both industrially and geographically diversified. This result indicates that, not taking other variables into account, the positive impact on MTB of geographical diversification outweighs the negative impact of industrial diversification. Summarized, the description of the variables shows a larger positive impact of geographical diversification and a smaller positive impact of industrial focus on a firm s value. 21

31 Table 3: Descriptive Statistics of Sample for MTB Value Measures Notes: The sample comprises firms which equates to observations. Industrial Segments are the different industries, indicated on level 4 SIC code. Geographical Segments are the different countries a company is active in. The first row of each variable represents the mean, the second row represents the median. Between group significance of the means are tested using the two sample t-test, the medians using the non-parametric Mann-Witney test. The t-statistic (MW p-value) is shown in the first (second) row for each variable. The overall significance level is 5%. Source: The diversification and segment data are taken from Amadeus, the other data are taken from Datastream. Total Single-Activity Multi-Activity Test of differences Domestic (NN) MNC (YN) Domestic (NY) MNC (YY) (NN)- (YN) (NN)- (NY) (NN)- (YY) (YN)- NY) (YN)- (YY) (NY)- (YY) Total Observations # Industrial Segments 1 1 4,26 4, # Geographic Segments 1 9, , Total Sales (million ) ,000,000,000,000,020, ,000,000,000,000,092,000 Total Assets (million ) ,000,185,000,000,003, ,000,001,000,000,693,000 Leverage,163,151,165,175,161,047,002,105,070,348,007,105,072,117,092,105,000,118,000,000,011,000 EBIT/Sales,065,051,070,051,069,006,950,007,000,663,000,066,057,070,052,070,000,349,000,000,625,000 Capex/Sales,076,088,082,082,071,277,296,001,929,000,003,038,037,042,032,039,001,416,104,000,000,000 R&d/Sales,023,005,029,006,028,000,732,000,000,539,000,000,000,000,000,000,000,040,000,000,000,000 Volatility 10,24 9,83 10,25 10,13 10,31,044,182,013,456,624, ,060,292,008,379,186,040