Ownership concentration, investment, and firm value in the shipping industry

Transcription

1 Ownership concentration, investment, and firm value in the shipping industry Wolfgang Drobetz Malte Janzen Ignacio Requejo s HFRC Working Paper Series No.30 February 2018 Hamburg Financial Research Center e.v. c/o Universität Hamburg Moorweidenstr Hamburg Tel (0) Fax: 0049 (0) info@hhfrc.de

2 Ownership concentration, investment, and firm value in the shipping industry a Wolfgang Drobetz b, Malte Janzen c, and Ignacio Requejo d Abstract We study the efficiency of capital allocation in the shipping industry and test whether it is affected by ownership concentration. For a sample of 126 globally listed shipping firms over the period 1997 to 2016, we find that investment in the commercial shipping sector follows freight rates, a measure of the potential income stream from owning a vessel. We show that ownership concentration, measured as the ownership stake of the largest shareholder, strengthens the positive effect of freight rates on investment, indicating a higher efficiency of capital allocation. Our results are not driven by overinvestment because firms with concentrated ownership structures also show stronger reduction of investment when freight rates are low. Ownership concentration positively affects firm value. An analysis of investor types reveals that our results are explained by the group of firms where the largest owner is a financial investor. Keywords: Capital allocation, Corporate investment, Firm value, Ownership concentration, Shipping firms. JEL Classification Codes: G01, G31, G32, L62, L91 a We thank Andreas Andrikopoulos, Nikos Nomikos, Anna Merika, Mike Tsionas, and participants at the 2017 Hamburg Maritime Management Symposium for valuable comments on previous versions of the study. b University of Hamburg, Hamburg Business School, Moorweidenstrasse 18, Hamburg, Germany. wolfgang.drobetz@uni-hamburg.de. c University of Hamburg, Hamburg Business School, Moorweidenstrasse 18, Hamburg, Germany. malte.janzen@uni-hamburg.de. d University of Salamanca, IME and Department of Business Administration, Campus Miguel de Unamuno, Edificio FES, E37007 Salamanca, Spain. irequejo@usal.es

3 1. Introduction Efficient investment decisions ensure operational profitability, enable sustainable firm growth, and are a key determinant of corporate success. The efficient allocation of capital is crucial in capital-intensive industries, where the majority of revenue is generated through ownership and operation of large and indivisible assets. Commercial shipping firms are a paradigmatic case study of these asset-heavy industries. Investments in the shipping industry are large in size and aim to provide additional capacity for a highly standardized service, the transportation of cargo around the world. In this study, we investigate to what extent investment decisions are linked to investment opportunities in the shipping industry. We follow Wurgler (2000) and Mortal and Reisel (2013) and interpret the reaction of investment to changes in investment opportunities as an indicator of the efficiency of capital allocation. Firms should invest more when the expected income from these investments is high, and less when the expected income is low. We take advantage of a distinctive feature of the shipping industry to circumvent the traditional problem of measuring investment opportunities in empirical studies of corporate investment behavior. As shipping firms provide a highly standardized service, they face a competitive, global market that determines the price of transportation, the freight rate. We use a freight rate index to proxy for the investment opportunities a shipping firm faces and expect firms to invest more when freight rates are high, and less when freight rates are low. In addition to its capital intensity, the shipping industry is a distinctive case because of its ownership structure. Shipping firms have traditionally been closely-held private companies, often family-founded and family-owned (Syriopoulos and Tsatsaronis, 2011). Although a substantial portion of large shipping companies have gone public in recent years (Merikas et al., 2009), ownership is often concentrated compared to other industries (Tsionas et al., 2012). 2

4 Therefore, we investigate whether ownership concentration affects the efficiency of capital allocation and firm value in the shipping sector. Berle and Means (1932) hypothesize that the transformation of firms, from closely held entities to public corporations with diffuse ownership, will reduce monitoring of managers through owners, eventually resulting in value-decreasing activities by managers. Demsetz (1983) challenges the validity of this proposition by arguing that ownership does not affect firm value but is endogenous in that it is the outcome of profit maximization of investors and not random. This critique is supported in Demsetz and Lehn (1985), who find no relation between ownership structure and firm performance, when ownership is treated as endogenous. Ignoring potential endogeneity concerns, Morck et al. (1988) document a piece-wise linear relation between inside ownership concentration and firm value. 1 The majority of studies that explore the effect of firm ownership on corporate investment decisions analyze the dependence of investment on the availability of internal funds, i.e., the so-called investment-cash flow sensitivity. In one of the first empirical studies on this topic, Oliner and Rudebusch (1992) do not find any effect of ownership structure on the link between investment and cash flow among NYSE firms. Hadlock (1998) shows that investmentcash flow sensitivities decrease when inside ownership exceeds a certain threshold. Goergen and Renneboog (2001) investigate the effect of different types of large shareholders on investment-cash flow sensitivities and find that large corporate owners increase the dependence of investment on the availability of cash flow, while large institutional investors are associated with lower sensitivities. Focusing on the Euro zone, Pindado et al. (2011) show that family firms exhibit lower investment-cash flow sensitivities than non-family ones. 1 Several other studies (McConnell and Servaes, 1990; Hermalin and Weisbach, 1991; Loderer and Martin, 1997; Himmelberg et al., 1999; Cho, 1998; Holderness et al., 1999; and Demsetz and Villalonga, 2001; among others) examine the relation between ownership concentration and firm value. Results vary substantially and depend to a large extent on methodological choices. Demsetz and Villalonga (2001) provide a comprehensive discussion of these studies. 3

5 Three studies that are closely related to ours analyze whether a different dimension of ownership, the dichotomy private versus public ownership, affects the sensitivity of investment to changes in investment opportunities. Mortal and Reisel (2013) find that Western European public firms adjust investments more than private firms when investment opportunities are high, while Asker et al. (2015) document the opposite result for the U.S. Mortal and Reisel (2013) argue that public firms are better positioned to make use of growth opportunities due to their access to capital markets. In contrast, Asker et al. (2015) attribute their findings to managerial short-termism due to stock market pressure among public firms, which leads to underinvestment. Gilje and Taillard (2016) compare investments of private and public U.S. firms in the natural gas industry. They find that public companies increase drilling activities more when the price of natural gas is high, indicating a relatively higher efficiency of capital allocation compared to private firms. The unique features of the shipping industry have led to empirical studies that investigate investment and financing decisions of shipping firms. Drobetz et al. (2013) examine capital structures of shipping firms and document high financial and operating leverage, even compared to similar manufacturing firms. Alexandrou et al. (2014) report positive abnormal returns for targets and acquirers in shipping M&A transactions, which are most pronounced for cross-border transactions and deals that aim to diversify the lines of business. Drobetz et al. (2016) find that financially weak shipping firms, even during normal times, face restricted access to external capital market. Although investments of both financially weak and healthy firms depend on the availability of internal funds during the recent crisis, the investment-cash flow sensitivity is lower in healthy shipping firms because they are still able to raise longterm debt. Ahrends et al. (2017) document that shipping firms hold more cash than comparable firms from the manufacturing sector. Their higher levels of cash holdings are not due to higher tangibility, but are instead driven by a stronger tendency of precautionary savings. 4

6 The effect of ownership in the shipping industry has been mainly studied in the context of firm valuation. Lambertides and Louca (2008) study the link between ownership structure and firm performance for a sample of firms from the maritime sector in Europe between 2002 and They document that the stake of foreign investors and the presence of investment companies increase firm performance. In a cross-sectional study of 107 globally listed shipping firms in the year 2009, Tsionas et al. (2012) find that ownership concentration and firm performance, measured by return on assets and return on equity, have a positive and bidirectional effect on each other. They conclude that large shareholders play an important role in the governance structure of shipping firms. In the current work, we show that investment decisions in the shipping industry react positively to changes in freight rates using a sample of 126 globally listed shipping firms over the period 1997 to We include corporate ownership structure in our analyses to evaluate the effect of ownership concentration. Our results show that ownership concentration increases the efficiency of capital allocation, i.e., investment sensitivity to changes in freight rates is intensified with increasing ownership concentration. By using a dynamic difference generalized method of moments (GMM) estimation method, we are able to control for (i) the dynamics of investment decisions, (ii) the potential endogenous nature of firms ownership structure, and (iii) time-constant firm-specific factors that affect investment behavior. Our model setup allows us to make sure the results are not driven by overinvestment. To this aim, we differentiate two freight rate states (i.e., high versus low freight rates), and for each state we test separately investment reactions to freight rates and the moderating role of ownership concentration. On the one hand, we confirm that shipping firms invest more when freight rates are high, compared to normal times, and less when freight rates are low. On the other hand, these reactions are symmetrically amplified by ownership concentration. Furthermore, consistent with these findings, we document that ownership concentration increases 5

7 firm value of shipping firms, a pattern that is robust to the consideration of other determinants of firm value. As our dataset also includes ownership characteristics, we are able to test whether our results are homogeneous across different types of investors. In particular, we distinguish between firms with a financial investor as the largest shareholder and strategic investor firms. We expect financial investors to be more active in monitoring in order to increase efficiency within companies to maximize firm value and realize capital gains (Thomsen and Pedersen, 2000; Attig et al., 2012). Our analyses reveal that the increase in the efficiency of capital allocation is indeed driven by the subsample of financial investor firms. Moreover, when we compare high and low freight rate states, we find that ownership concentration in strategic investor firms reduces investment during high freight rate states. Consistent with the results on the efficiency of capital allocation, our empirical evidence shows that ownership concentration affects firm value positively only in the subsample of financial investor firms. Our findings are robust to numerous concerns. First, we replace our shipping industryspecific measure of investment opportunities (i.e., the freight rate index) and use a conventional variable that is broadly used in empirical research on corporate investment, a firm s market-to-book ratio. All patterns derived from our main specifications persist when we use this alternative measure of investment opportunities. Second, we use a more narrow definition of investor subsamples and create a sample of activist investors; namely, those financial entities that are more shareholder-value oriented and are more likely to actively engage in shareholder activism, such as hedge funds and venture capital investors. We find that our results are more pronounced in the activist investor firm sample, pointing to an important role of activism and shareholder-value orientation in explaining the association between investment efficiency and ownership concentration. Third, our main results remain unchanged when we control for the presence of a second large owner. Fourth, the findings from the investment 6

8 regressions continue to hold when we use a broader measure of investment that accounts for increases in fixed assets through any kind of acquisition. Lastly, we test whether the positive impact of ownership concentration on firm value can be observed for an accounting performance measure and find that ownership concentration increases return on assets in the full sample. Confirming the empirical evidence from the firm value regressions, this effect is explained by the subsample of financial investor firms. Our paper contributes to different strands of literature in corporate finance and maritime economics. First, we add to the literature that examines investment decisions of shipping companies, which are at the center of value-creation in this asset-heavy industry. Second, we show that earlier evidence that links ownership concentration with increases in firm performance is also applicable to firm value. Following the suggestion by Tsionas et al. (2012), we show that the effect of ownership on firm value and performance can also be observed in a panel data setting. This approach represents a noteworthy methodological improvement because it allows us to control for time-varying aspects of investment behavior as well as timeconstant firm-specific determinants of investment. Third and most importantly, we document an important mechanism through which ownership concentration increases firm value and performance: high ownership concentration increases the relative efficiency of capital allocation. In a more general sense, we provide insights into the effects of ownership concentration by focusing on an industry that is historically different from other sectors not only due to the concentrated ownership structures of companies, but also due to these firms distinctive risk characteristics (Albertijn et al., 2011). The remainder of the study is organized as follows. Section 2 describes our sample and the methodological approach. Section 3 presents the main results. Section 4 checks the robustness of the results. Section 5 concludes. 7

9 2. Research design To test whether ownership concentration has an effect on the sensitivity of investment to changes in investment opportunities in the shipping industry, we combine data from multiple sources. In this section, we describe the datasets we use and how we combine them for our analysis. We also present the variables and the setup of our empirical specification, and discuss how we control for the potential endogenous nature of a firm s ownership structure Sample We use financial and accounting data from Compustat North America and Compustat Global annual files. We follow Drobetz et al. (2013) and restrict our sample to firms that own and operate commercial ships. 2 The identification of these firms is based on Thomson Reuters Datastream business descriptions and publicly available sources, e.g., their websites and annual reports. The initial sample of shipping firms consists of 154 public shipping firms, 114 active and 40 inactive, from 36 countries. We use ownership information from Thomson Reuters Eikon Ownership & Profiles module and match the shipping firms identified in Compustat to the Thomson Reuters data by ISIN. 3 Eikon provides ownership information for 146 of the initial 154 shipping firms. We primarily use annual information about the percentage of shares held by the largest shareholder in our analysis. However, Eikon provides more details about the owners of a company. We also consider the type of owner, as categorized by Thomson Reuters, and the percentage of shares held by the second largest shareholder. In particular, we categorize owners into financial investors and strategic investors according to the types of ownership assigned to them 2 We exclude firms from other areas of the maritime sector, e.g., ship yards, passenger ships, and supply vessels. Although we cannot rule out that their business cycles are affected by freight rates, we do not expect them to be an important determinant of investment decisions in these companies. Our sampling procedure ensures that the sample only comprises freight shipping companies. 3 We identify firms without information on the ISIN in Compustat manually by company name in Eikon. 8

10 by Eikon. By doing so, we aim to exploit the heterogeneity between different types of ownership to provide further insights into the effect of ownership concentration. We add freight rate information from the Clarksons Shipping Intelligence Network (SIN) to our dataset. In particular, we use the monthly Clarksons ClarkSea index, a weighted average index of earnings in the main sectors of commercial vessels, weighted by fleet size of the respective vessel type. Lastly, we require at least four consecutive years of non-missing ownership and financial data for a firm to be included in the sample. This criterion is necessary due to the use of lagged values in dynamic difference GMM estimations. It reduces our final sample to 126 firms from 33 countries with 1,370 observations over the period from 1997 to Table I reports the number of firms by country of incorporation. Table II shows the number of firm-year observations by type of owner and illustrates the categorization into financial and strategic investors. Insert Table I here Insert Table II here Table III reports the descriptive statistics of the variables we use in the subsequent analyses. We winsorize all variables at the 1 st and 99 th percentiles to avoid that our results are affected by extreme values. In our sample, the mean percentage of shares held by the largest shareholder (OC) amounts to 28.2%, with a median percentage of 23.2%. In over 75% of all firm-year observations, the largest owner holds more than 10% of shares. This concentration of ownership is far from the dispersed ownership structure in public corporations anticipated by Berle and Means (1932) and points to a distinctive feature of the shipping industry. These levels of ownership concentration are in line with the cross-sectional evidence provided by Tsionas et al. (2012) for the year 2009 and show that, with respect to ownership concentration, their results are representative of a longer time-series. Insert Table III here 9

11 On average, the second largest owner (SECO) holds 8.3% of the shares in our sample, with a median of 7%. The investment ratio (INV), the firm-level variable of interest in this study, has a mean of 8.6%, but it exhibits a high variation over time and across firms. The same tendency holds true for cash flow, with a mean of 6.3%, but a standard deviation that is almost double. The variation of investment and cash flow can be explained to a great extent by the cyclicality and volatility in the shipping industry (Albertijn et al, 2011). Focusing on the distribution of the ClarkSea index (CSI), we find that our sample period captures sufficient heterogeneity and includes both boom and bust phases of the shipping cycle. The low market-to-book ratios (MB), the high leverage ratios (LEV), and the high cash holdings (CASH) confirm well documented unique characteristics of shipping firms (Albertijn et al., 2011). The median market-to-book ratio is below 1, indicating a valuation below book value. However, comparing the first and third quartiles of the distribution reveals a substantial variation within our sample. What seems even more important in our setup is that this volatility is more pronounced when we consider the (unreported) time-series variation. For example, the mean market-to-book ratio is above 1.5 in 2007, when freight rates reached their all-time high, and only 0.9 in The mean leverage ratio, the sum of short and long term debt divided by total assets, stands at 43.1%, well above the levels in other industries. The high use of debt financing in the shipping industry is primarily driven by the high capital intensity of shipping firms and a historically restricted access to organized equity capital market that has only recently started to relax (Drobetz et al., 2013). The high mean cash holdings ratio of 12.2% is consistent with evidence in Ahrends et al. (2017). The mean dividend ratio (DIV) is 2%, while on average shipping firms hold intangible assets (IA) amounting to 1.8% of total assets. Both variables have much lower medians. Lastly, we define a firm s age (AGE) as years since the firm s IPO. On average, the firms in our sample have been listed for 14 years. 10

12 2.2. Empirical design In this study, we test whether ownership concentration has an effect on the efficiency of capital allocation by comparing the sensitivity of investment to changes in investment opportunities. We build on the approach presented by Wurgler (2000), who shows that countries with developed financial markets have stronger investment increases in industries with high investment opportunities, and stronger investment decreases in industries with low investment opportunities. As explained above, this country-level comparison of relative capital allocation efficiency has been transferred to the firm-level to investigate the effect of a specific dimension of ownership, private versus public ownership, in Mortal and Reisel (2013), Asker et al. (2015), and Gilje and Taillard (2016). The q theory of investment posits that investment decisions of a firm should only be dependent on Tobin s (1969) marginal q, the investment opportunities the firm faces. However, an empirical test of investment efficiency based on marginal q, the ratio of expected discounted cash flows a new investment project generates to its replacement costs, is challenging as neither the nominator nor the denominator of marginal q is directly observable. Conventionally, empirical studies thus use average q, 4 the ratio of the market value of assets of the firm to their replacement costs, approximated by the firm s lagged market-to-book ratio (MB), as the measure of a firm s investment opportunities. To study the efficiency of capital allocation, we use a standard investment regression that models a firm s investment as a function of cash flow and a measure of investment opportunities. While numerous empirical studies on firms investment behavior are based on the inclusion of MB, the choice is subject to severe criticism. Erickson and Whited (2000) argue that a well-established aspect of research on firm s investment decisions, the positive link 4 Hayashi (1982) shows that the two measures (i.e., marginal q and average q) are essentially (p. 218) the same under conceivable assumptions. 11

13 between investment and cash flow, 5 is in fact the result of measurement error in q. Using evidence from simulations and data on manufacturing firms, they show that, once accounting for a potential measurement error, the dependence of investment on cash flow disappears. Similar evidence is provided by Alti (2003), who shows how investment regressions are biased because Tobin s q is not able to control for investment opportunities that arise throughout the year. He thus concludes that q is a noisy measure of short-term investment opportunities. We exploit a distinctive feature of the shipping industry to mitigate concerns about Tobin s q as a measure of investment opportunities in our study of capital allocation efficiency. The shipping firms in our sample own and/or operate commercial ships, i.e., they offer a homogeneous good, seaborne transportation of goods, and generate income by carrying cargo (freight contracts) or chartering out ships (time charter). Investments of these firms should focus on sustaining or improving their ability to fulfill these services. Because shipping services are highly standardized by type of cargo and per route, the price of transportation (i.e., the freight rate) is also determined in the global freight market. As pointed out by Stopford (2009), the link between freight rates and investment in the shipping industry is unambiguous: when freight rates are high, potential earnings from additional ships new buildings or second-hand vessels increase and owners would want to increase fleet size. Therefore, we use the Clarksons ClarkSea index as a measure of the potential income from owning a ship and include it in our analysis as the proxy for investment opportunities. This methodological approach enables us to circumvent the problems inherent in the use of MB as a measure of investment opportunities. The idea of using the price of a homogeneous good within one industry to study the efficiency of capital allocation across different types of ownership is borrowed from Gilje and 5 Dating back to Fazzari et al. (1988), studies document a positive link between a firm s investment and its cash flow. Although the interpretation of these findings is challenged by some authors (Kaplan and Zingales, 1997; Chen and Chen, 2012), recent studies still document positive investment-cash flow sensitivities (Lewellen and Lewellen, 2016; Moshirian et al., 2017). 12

14 Taillard (2016), who study whether private and public firms in the natural gas industry adjust investment differently to changes in gas prices. Our basic investment model, not considering for the moment the effect of ownership concentration, can thus be formalized as follows: INV it = β 1 INV i,t 1 + β 2 CF it + β 4 log(csi t ) + ε it (1) where the i sub-index refers to firms, and the t sub-index is related to fiscal years. 6 INV is the firm s investment ratio, defined as capital expenditures net of asset sales divided by total assets. CF is the firm s cash flow, defined as income before extraordinary items plus depreciations divided by total assets. Finally, log(csi) is the natural logarithm of the 12-month average of the Clarkson ClarkSea index. 7 Figure I compares the Clarkson ClarkSea index to the annual mean investment ratio of shipping firms in our sample. This figure highlights that periods of high freight rates coincide with periods of high investment rates (and vice versa), supporting our choice of freight rates as a proxy for investment opportunities. Insert Figure I here Since we are interested in examining whether capital allocation efficiency is moderated by ownership concentration, our main empirical specification can be formulated as follows: INV it = β 1 INV i,t 1 + β 2 OC it + β 3 CF it + β 4 log(csi t ) + β 5 log(csi t ) OC it + ε it (2) where OC is the percentage of shares held by the firm s largest investor. The coefficients of interest to evaluate the efficiency of capital allocation are β 4 and β 5. We expect investment to 6 For the sake of readability, we omit these sub-indices in the description of variables or the discussion of results. 7 We cannot include time dummies in the model because the freight rate index takes the same value for all firms each time period, and thus they would be perfectly correlated with the Clarkson ClarkSea index. In the robustness tests, we use a different measure of investment opportunities and include time dummies in the model. 13

15 increase when freight rates are high and investments can generate higher cash flows. Following this rationale, we expect β 4 to be positive. Ex ante, it is unclear whether ownership concentration increases or decreases investment efficiency, and thus there in no clear prediction for the sign of β 5. Prior literature on the effects of ownership structure on different firm dimensions, dating back to Demsetz (1983) and Demsetz and Lehn (1985), emphasizes that ownership structure is endogenous. For the shipping industry, Tsionas et al. (2012) show a positive and bilateral association between ownership concentration and performance after accounting for the endogenous nature of ownership, which is consistent with the notion that ownership concentration serves as a corporate governance mechanism. Endogeneity of ownership concentration in our setting may arise from two sources. First, owners may choose to hold a larger share in firms that invest more efficiently, leading to reverse causality. Second, both investment and ownership concentration may be affected by a third confounding but unobserved factor; e.g., management quality. We address these concerns by applying the dynamic difference GMM (Arellano and Bond, 1991) in our model estimations. This method allows us to control for the effect of unobserved heterogeneity on the investment behavior of shipping firms by differencing out within-firm means, while simultaneously addressing the endogeneity problem of ownership concentration. Specifically, we include lags from t 1 to t 3 of all right-hand side variables as instruments in our models, except for the lagged value of investment, whose instruments are lags from t 2 to t 4. 8 Including the first lag of the dependent variable as an explanatory variable enables us to control for the dynamic nature of investment decisions. 8 Freight rates and time dummies, which are beyond the influence of firms, are treated as strictly exogenous and instrumented by themselves. 14

16 3. Results In this section, we present the main results from the estimation of our investment model. We then analyze whether the impact of ownership concentration on the efficiency of capital allocation is uniform across firms and over market phases, i.e., during high and low freight rate environments. Finally, we test whether the positive effect of ownership concentration on investment efficiency translates into higher firm value Investment, freight rates, and ownership concentration Table IV reports the results from dynamic difference GMM estimations of equations (1) and (2). Column 1 reports estimates for the basic investment model. We find that prior year s investment drives current investment, as shown by the positive estimate of the lagged investment ratio, supporting an accelerating mechanism of investment (Aivazian et al., 2005). In addition, investment is positively linked to cash flow. From each additional dollar of cash flow, shipping firms in our sample will allocate 3.6 cents to finance investments. The estimate of interest in this column is the coefficient on log(csi), our measure of investment opportunities in the shipping industry. We find that shipping firms, on average, invest according to our expectations. The significantly positive coefficient indicates that investment of shipping firms increases when freight rates increase. A one standard deviation change in log(csi), 0.442, leads to an increase of percentage points in investment, equivalent to 16.3% of the mean investment ratio in our sample. 9 This finding supports the association already depicted in Figure I and documents that shipping firms adjust their investment decisions based on changes in freight rates. In column 2, we estimate our main model, including ownership concentration and the interaction term between ownership concentration and the freight rate index, log(csi) OC, 9 The calculation is: = The mean investment ratio in our sample is (see Table III). 15

17 and using the same dynamic difference GMM approach. The effect of lagged investment on current investment remains positive and significant, but it decreases in magnitude. The same holds true for cash flow. The coefficient on CF is highly significant, but decreases to only The level of investment decreases with ownership concentration, as indicated by the significantly negative coefficient on OC. A possible explanation for this finding could be that investors with a high ownership stake in the firm become more conservative and prefer to avoid new risky projects. The positive effect of freight rates that we observe in column 1 persists after the inclusion of ownership concentration. Insert Table IV here More importantly, we are interested in the moderating effect of ownership concentration in the relation between investment and freight rates. The estimate of the interaction term log(csi) OC is positive and statistically significant. To illustrate the economic magnitude of the effect, we consider a one standard deviation change in log(csi) at ownership concentration (OC) levels equivalent to the first (Q1: 0.105) and third quartiles (Q3: 0.419) of its distribution. A one standard deviation change in log(csi), holding everything else constant, translates into an increase in the investment level from percentage points at Q1 to percentage points at Q3. This difference (0.004) implies an increase in the investment-freight rate sensitivity of about one fourth when moving from the first to the third quartile of the ownership concentration distribution. 10 The results presented thus far point to a positive effect of ownership concentration on investment efficiency in the shipping industry. While an increase in ownership concentration leads to less investment per se, firms become more sensitive to changes in freight rates, indicating an improvement in the efficiency of capital allocation. Next, we analyze whether this 10 A one standard deviation change in log(csi) is The first and third quartiles of the distribution of OC in our sample are and 0.419, respectively. The respective computations are then as follows: for Q1: = 0.014; for Q3: =

18 efficiency improvement is uniform across types of owners. We distinguish between financial and strategic investor firms, based on the classification provided in Table II. We categorize a firm as a financial investor firm if a financial investor is the largest owner in at least one year of the sample period. 11 The remaining firms are categorized as strategic investor firms. Our analysis is motivated by Lambertides and Louca (2008), who document that performance of maritime companies increases with the ownership share of financial investors. We then estimate equation (2) separately for both types of firms. Thomsen and Pedersen (2000) posit that institutional investor ownership, a classification that is comparable to our financial investor definition, focuses on increasing shareholder value. They show that stakes by institutional investors increase firm value and profitability. Attig et al. (2012) find that the presence of institutional investors with a long investment horizon reduces investment-cash flow sensitivities. They argue that, thanks to an informational advantage over individual investors, institutional investors have a higher incentive to monitor, and their monitoring activities are more efficient. We thus expect the increase in the efficiency of capital allocation to be more pronounced for financial investor firms. These investors aim to increase the efficiency of the firm and maximize firm value often through activism in order to realize capital gains on their investments. Conversely, strategic investors may pursue goals that are not primarily driven by the maximization of firm value but, for instance, by the transfer of knowledge through vertical ties (Thomsen and Pedersen, 2000). Columns 3 and 4 in Table IV report the estimation results. We again find that the coefficient on ownership concentration is negative. 12 For financial investor firms the main effect 11 The category to which a firm is assigned remains constant over time to ensure a series of consecutive observations for each firm. This requirement is necessary to use the GMM when estimating the investment model separately for each firm category. 12 While this coefficient appears to be large, it is almost entirely offset by the coefficient on the interaction term log(csi) OC, which has the opposite sign. When we set log(csi) to its mean (9.667), the change in the investment ratio in response to a one standard deviation change in ownership concentration (0.221) is ( =)

19 of an increase in freight rates, indicated by the estimate for log(csi), is positive but smaller in magnitude than for the average shipping firm in the pooled sample. However, this sensitivity is substantially amplified with increasing ownership concentration. The coefficient on the interaction term is more than four times larger compared to the result obtained in the pooled sample. A one standard deviation change in freight rates leads to an increase of in investment when we set OC to its third quartile, and only at its first quartile. In stark contrast, the results for strategic investor firms do not support a moderating effect of ownership concentration on the investment-freight rate sensitivity, since the coefficient on the interaction term log(csi) OC is insignificant in column 4. Although the main effect of log(csi) is larger for strategic investor firms (coefficient of vs ), the combined effect of log(csi) and log(csi) OC in financial investor firms exceeds the total impact in strategic investor firms at a low ownership concentration level of only about 3.1%. 13 Overall, we conclude that the increase in the efficiency of capital allocation is driven by financial investor firms, i.e., firms with an investor that aims to maximize firm value and is more likely to become active rather than a strategic investor High vs. low freight rate environments The results of our main analysis reveal three stylized facts. First, we find that, in all our specifications, investment activity of shipping companies is positively linked to changes in freight rates. Second, ownership concentration increase the efficiency of capital allocation as it has a positive moderating effect on the investment-freight rate sensitivity. Third, this amplifying effect is only observed for financial investor firms. Next, to examine whether this positive effect of ownership concentration is symmetric and to separate it from potential overinvestment, we compare firms investment behavior in environments with high and low freight 13 The calculation is as follows: , which is equal to the coefficient on log(csi) in column 4. About 95% of the observations in our financial investor firm sample exceed this threshold. 18

20 rates. Investment efficiency implies that shipping firms invest more when freight rates are high, and less when freight rates are low (compared to normal times). An argument that would contradict our line of reasoning is that ownership concentration increases investment in all states of the shipping cycle, thus pointing to an overinvestment problem rather than higher efficiency of capital allocation. A first indicator that this alternative explanation is unlikely is the negative coefficient on OC in all specifications in Table IV, which indicates that in general OC leads to lower investment. To further mitigate this concern and examine the effect of ownership concentration in different states, we follow Gilje and Taillard (2016) and compare periods of high and low freight rates to normal times. We classify observations as high (low) freight rate observations if the 12-month average ClarkSea index is in the top (bottom) 20% of its distribution over our sample period. 14 We adjust equations (1) and (2) and add two dummies, CSI_HIGH and CSI_LOW, instead of the continuous freight rate measure in our baseline models. CSI_HIGH (CSI_LOW) takes a value of 1 if an observation belongs to a high (low) freight rate period, and 0 otherwise. Our main specification, including ownership concentration, is now formulated as: INV it = β 1 INV i,t 1 + β 2 OC it + β 3 CF it + β 4 CSI HIGHt + β 5 CSI HIGH t OC it (3) + β 6 CSI_LOW t + β 7 CSI_LOW t OC it + ε it Considering the evidence presented in Table IV, we expect β 4 to be positive and β 6 to be negative, suggesting that shipping firms increase investment, compared to normal times, when freight rates are high, and decrease investment when freight rates are low. If the effect of ownership concentration we find is the result of overinvestment tendencies in firms with high ownership concentration, both β 5 and β 7 should be positive. In this case, ownership concentration would increase investment in all states, even during phases of low freight rates. By 14 In unreported tests, we use broader thresholds (33 rd and 66 th percentiles) as in Gilje and Taillard (2016). The results become slightly less pronounced but remain qualitatively similar. 19

21 contrast, if our results are driven by an increase in the efficiency of capital allocation, we would expect β 5 to be positive and β 7 to be negative. Table V reports the results from dynamic difference GMM estimation of equation (3). In column 1, we exclude OC and its interaction terms with the freight rate environment dummies. The estimates for CSI_HIGH and CSI_LOW reveal that the results we obtain for the continuous measure in Table IV are in fact symmetric. Shipping firms invest more than in normal times when freight rates are high, and less when freight rates are low, as indicated by the positive and negative coefficient on CSI_HIGH and CSI_LOW, respectively. Insert Table V here In column 2, we include OC together with the interaction terms CSI_HIGH OC and CSI_LOW OC. The estimates do not support the alternative overinvestment explanation. OC reduces investment-freight rate sensitivities during low freight rate environments and increases them during high freight rate environments, thereby steering shipping firms towards a more efficient allocation of capital. For a firm with OC set to the first quartile, a change from normal to high freight rates results in an increase of investment by 0.018, which is equivalent to 20.8% of the mean investment ratio. The same change in freight rates leads to an increase of investment by 0.025, or 28.9% of the mean investment ratio, for firms with OC set to the third quartile. 15 Similarly, a comparison for the change from normal to low freight rates results in a decrease of 0.02, or 23.2% of the mean investment ratio, for the first quartile of OC, and a decrease of 0.023, or 26.7% of the mean investment ratio, for the third quartile of OC. The estimates for the full sample of shipping firms in column 1 indicate that shipping firms invest more when freight rates are high, and less when freight rates are low. The results in column 2 show that the magnitude of the investment response is amplified by the degree of 15 At Q1 of OC (0.105), the investment reaction to a change from normal to high freight rates is computed as follows: = At Q3 (0.419), the computation is: =

22 ownership concentration. This evidence rules out an overinvestment explanation and rather points to an increase in the efficiency of capital allocation through ownership concentration. Next, in columns 3 and 4, we again split the sample into financial and strategic investor firms to assess whether our results hold for all investor types. In column 3, we restrict the sample to financial investor firms. Financial investor firms invest more when freight rates are high, as indicated by the positive coefficient on CSI_HIGH, and less when freight rates are low, as shown by the negative coefficient on CSI_LOW, thus confirming the results from the full sample. When we consider the moderating effect of ownership concentration on the sensitivity of investment to freight rates, the results are different from those obtained in the full sample. The increase in investment during high freight rate periods becomes larger with increasing ownership concentration. Albeit the coefficient on the dummy variable CSI_HIGH is smaller than in the full sample (0.011 vs ), the coefficient on the interaction term is much larger in magnitude. Even at very low levels of ownership concentration, 16 increases in investment will be more pronounced in the financial investor sample than in the full sample during high freight rate periods. Conversely, the effect of ownership concentration on investment is not affected by ownership concentration when freight rates are low. However, we note that the main effect, indicated by the coefficient on CSI_LOW ( 0.031) in column 3, is slightly larger than the decrease in investment during low freight rates for the full sample in column 2 for a hypothetical firm with only one owner (OC = 1: ( ) = 0.030). In column 4, we restrict the analysis to the sample of strategic investor firms. The pattern we find for these companies is different from the findings for both the full sample in column 2 and the group of financial investor firms in columns 3. First, the increase in investment during high freight rate periods becomes smaller with increasing ownership concentration and 16 The exact cutoff point is at an ownership concentration level of 5.1%. Financial investor firms increase investment during high freight rates by ( =) , while the effect in the full sample is only ( =)

23 even becomes negative if OC exceeds 31.4%. Second, the coefficient on CSI_LOW is statistically equal to zero in column 4, implying that, if OC is set to zero, these firms will not adjust investment when freight rates are low. However, when we take the interaction term CSI_LOW OC into consideration and set OC to its mean in the strategic investor sample (0.343), the decrease in investment during low freight rate states is percentage points, or -22% of the mean investment ratio. Taken together, the results presented in this section show that the amplifying effect of ownership concentration we find in the main analysis is a symmetric one that holds both during low and high freight rate periods. Our findings rule out an alternative explanation of overinvestment caused by ownership concentration, since ownership does not increase investment during all states but only when freight rates are high. In the following section, we test whether this increase in the efficiency of capital allocation ultimately translates into higher firm value Ownership concentration and firm value We show that ownership concentration increases the efficiency of capital allocation in the shipping industry. Wurgler (2000) argues that the allocation of capital between a country s industry sectors (i.e., increasing investment in profitable industries and decrease investment in unprofitable industries) is crucial to the overall economic success. This rationale should also hold at the firm level, especially in a capital-intensive industry, such as the shipping sector. Given the effect of ownership concentration on the efficiency of capital allocation, we next analyze whether ownership concentration also affects firm value in the shipping industry. For a cross-section of shipping firms in the year 2009, Tsionas et al. (2012) document that ownership concentration increases (accounting) performance. We add to their results by analyzing whether this positive effect on performance also exists for firm value, and by extending their cross-sectional test to a panel data context that allows us to control for within-firm varia- 22

24 tion. We first estimate the marginal effect of ownership concentration on firm value by modelling a firm s current market-to-book ratio (MB) as a function of lagged MB (MB t-1 ) and ownership concentration (OC). In a second step, we add other firm-level determinants of firm value to control for value creation through other firm-level dimensions. In particular, we estimate the following equation: MB it = β 1 MB i,t 1 + β 2 OC it + β 3 LEV it + β 4 INV it + β 5 DIV it + β 6 SIZE it (4) +β 7 IA it + β 8 CF it + β 9 AGE it + Year t + ε it where MB is the firm s market-to-book ratio, defined as the market value of equity divided by the book value of equity. OC, INV, and CF are already defined above. LEV is the firm s leverage ratio, defined as total debt divided by total assets. DIV is the ratio of dividends to total assets. SIZE is the natural logarithm of total assets in USD. IA is intangible assets divided by total assets. AGE is the number of years since the firm s IPO. Year denotes (fiscal) year dummies. Table VI presents the results of dynamic difference GMM estimations of equation (4). In columns 1 to 3, we estimate the parsimonious form of this equation and model MB as a function of lagged MB and OC together with (unreported) year dummies. In column 1, we find persistence in firm value (i.e., lagged MB positively affects current MB), which highlights the importance of considering a dynamic model when examining firm value. More importantly, OC exerts a positive effect on firm value. Interestingly, when we analyze the financial and strategic investor firm samples separately, we find that the positive effect of OC is only confirmed for the financial investor firm sample. This finding is consistent with the results from our investment models. Insert Table VI here 23

25 In columns 4 to 6, we include other firm-level factors as determinants of firm value and estimate the extended equation (4). The estimates for the full sample in column 4 confirm the results from the estimation in column 1. Firm value is positively related to prior year s MB and OC. The magnitude of the positive effect of OC on firm value increases substantially after controlling for other determinants of firm value. In particular, a one standard deviation change in OC translates into an increase in MB equal to 0.026, equivalent to 2.5% of the average MB in our sample. A comparison of the financial investor firm sample with the strategic investor firm sample also shows that the results reported in columns 2 and 3 are robust to the inclusion of additional control variables. Ownership concentration affects firm value only in the sample of financial investor firms, where a one standard deviation change in OC translates into an increase in MB by 0.039, equivalent to 3.2% of the average MB in the sample of financial investor firms. Taken together, the evidence presented in this section confirms and extends the results of Tsionas et al. s (2012) study by highlighting that OC affects not only accounting performance but also firm value. Moreover, our findings continue to hold even when controlling for other determinants of firm value and for time-constant firm-specific factors (thanks to the use of a panel data estimation method). 4. Robustness checks In this section, we perform several tests to examine the sensitivity of our results to different adjustments in the methodology. First, we modify the specification of our main analysis. In particular, we use an alternative measure of investment opportunities, adopt a more stringent definition of investor types, and include a dummy variable to control for the presence of a second large owner. Second, we use a broader measure of investment as dependent variable. Lastly, we consider accounting performance, instead of MB, in the firm value model. 24

26 4.1. Alternative specifications In the previous analyses, we adopt the research design proposed by Gilje and Taillard (2016), who use the price of a homogeneous good to identify investment opportunities within the natural gas industry. We apply this approach to commercial shipping by using a freight rate index to capture expected profits from investments. One could argue that this identification strategy ignores firm-specific factors that relate to individual investment opportunities. Albeit our approach controls for time-constant, firm-specific characteristics, we address this concern by adjusting the measure of investment opportunities in our baseline model. We follow prior studies on the determinants of corporate investment decisions and adjust equation (2) to include MB t-1, the firm s previous year s market-to-book ratio, instead of log(csi) as a proxy measure for investment opportunities. This specification is equivalent to the traditional Fazzari et al. (1988) investment setup that models investment as a function of cash flow and Tobin s q, approximated by lagged MB. 17 We extend this model by including OC and the interaction term MB t-1 OC as explanatory variables. Columns 1 to 3 in Table VII present the results from this adjusted investment specification. As in Table IV, column 1 shows that investment is positively related to lagged investment and negatively to ownership concentration. More importantly, however, our main results on the association between investment and changes in investment opportunities from Table IV persist. Investment reacts positively to changes in investment opportunities, and this positive link increases with ownership concentration, as indicated by the positive coefficients on MB t-1 and the interaction MB t-1 OC. We conclude that the increase in the efficiency of capital allocation we observe is not exclusive to the use of freight rates as a measure of investment opportunities. 17 This model is widely used in empirical studies, including the works by Kaplan and Zingales (1997), Chen and Chen (2012), Lewellen and Lewellen (2016), and Moshirian et al. (2017). 25

27 Insert Table VII here The results in columns 2 and 3 also confirm the differences between financial and strategic investor firms. Similar to our previous results, reported in Table IV, the main effect of MB t-1 is larger (and significant only) in the sample of strategic investor firms. However, ownership concentration has a positive effect on the investment-q sensitivity only for the sample of financial investor firms, offsetting the larger coefficient on MB t-1 at an ownership concentration level of 17.4%. 18 Although this breakpoint is larger than in the main analysis, more than half of the observations in our sample (56.5%) exhibit a higher OC level. Ownership concentration increases the efficiency of capital allocation for the sample of financial investor firms. This result can be attributed to the efficiency-oriented nature of these investors and their tendency to become active, i.e., to interfere with corporate decision making if they are discontent with management actions. We now consider only investors that are most likely to become active shareholders and that, by definition, are primarily interested in capital gains through increases in efficiency and firm value. We call this group of investors active investors and include in this category investors that are labeled as hedge funds, investment advisors/hedge funds, private equity, and venture capital. We label the remaining investors passive. However, it should be noted that this investor group needs not necessarily be passive, but at least it is expected to be more passive than the active group. A firm is categorized as an active investor firm if an active investor is the largest owner in at least one year during the sample period, and passive otherwise. If the increase in the efficiency of capital allocation is driven by the efficiency-orientation and active nature of financial investors, the coefficient on the interaction term log(csi) OC should be larger in the active investor firm sample. 18 We obtain this breakpoint by dividing the main effect (the coefficient on MB t-1 ) in column 4 by the coefficient on the interaction term MB t-1, thus 0.078/0.448 = This percentage gives the level of OC at which the investment reaction to changes in investment opportunities will be of equal magnitude across subsamples. 26

28 Columns 4 and 5 of Table VII present the results for the active and passive investor subsamples. We find a negative (and marginally significant) coefficient on log(csi) in the sample of active investor firms in column 4, together with a positive and large coefficient on the interaction term log(csi) OC that offsets the negative coefficient at an ownership concentration level as low as 4.9%, 19 resulting in the expected positive effect beyond this point. Moreover, the coefficient is substantially larger than the corresponding coefficient in the financial investor firm subsample in Table IV, indicating that active ownership and efficiency orientation can explain the increase in efficient capital allocation. To illustrate the economic impact, we again consider a one standard deviation change in log(csi) (0.422) at the first (Q1: 0.071) and third quartiles (Q3: 0.204) of the distribution of OC in the active investor sample; investment increases by at Q1 and by at Q3. The difference we report in the same analysis using the full sample in Section 3.1 above is substantially smaller (0.004). 20 For passive investor firms, both coefficients are positive, but the coefficient on the interaction term is much smaller, leading to a lower response of investment to changes in freight rates at an OC breakpoint of 12.2%. Previous studies points to an important role of the second largest owner. La Porta et al. (1999) show that non-u.s. firms are often owned by multiple large shareholders. Claessens et al. (2000) find that for a third of East Asian firms in their sample, the second owner holds at least 10% of shares. For Western Europe, Faccio and Lang (2001) estimate that between 45% and 55% of firms have more than one large owner. In our sample of shipping firms, over three thirds of all observations have a second owner that holds over 10% of shares. Prior empirical studies report that multiple large owners reduce investment-cash flow sensitivities (Pindado et 19 Almost 90% of the observations in the active investor firm sample exceed this threshold. 20 This difference remains sizable even after adjusting for the larger mean investment ratio in the active investor sample (0.099) compared to the full sample (0.086). 27

29 al., 2011), 21 decrease expropriation of minority shareholders through dividend policy (Faccio et al., 2001), increase firm value (Maury and Pajuste, 2005; Laeven and Levine, 2007; Attig et al., 2009), and lower the cost of equity (Attig et al., 2008). Taken together, the evidence suggests a positive effect of a second large owner that is attributable to additional monitoring of managers and monitoring of the largest owner. As we do not account for the presence of second large shareholders in our main analyses, the positive effect we attribute to the largest owner could be in part explained by the presence of a second large owner. We test this alternative hypothesis by adjusting equation (2) and including a dummy (SECO) that takes the value of 1 if the second largest shareholder holds more than 10% of shares, and 0 otherwise. In particular, we add SECO and log(csi) SECO to the equation. Columns 6 to 8 in Table VII present the results. All coefficients that include SECO are significant in column 6, where we use the full sample of firms, pointing to an important role of the second largest owner in investment decisions of shipping firms. For the full sample in column 6, the presence of a second large owner decreases the efficiency of capital allocation, as suggested by the negative coefficient on the interaction term log(csi) SECO. However, the main result from Table IV persists and the efficiency of capital allocation increases with the stake of the largest owner. When we split the sample into financial and strategic investor firms, the results confirm our expectations. The efficiency of capital allocation increases with ownership concentration for financial investor firms. The presence of a second large owner also has a positive effect on the efficiency of capital allocation in this sample. For strategic investor firms, however, the inclusion of the SECO dummy reveals that both ownership concentration and the presence of a second large owner adversely affect the efficiency of capital allocation. 21 Pindado et al. (2011) find that a second large non-family owner reduces investment-cash flow sensitivities in European family firms. 28

30 4.2. Investment measure In our main analysis, we use a conventional measure of investment by defining the investment ratio as capital expenditures minus assets sales, divided by total assets. Lewellen and Lewellen (2016) question this standard definition and use a broader measure of investment, the change in fixed assets. The advantage of this measure is that it includes acquisitions, independent of the means of payment. Given consolidation tendencies in the shipping industry in the early 2000s and in recent years (Alexandrou et al., 2014; Alexandridis and Singh, 2016), it is conceivable that a sizable fraction of investment in the fleet is done indirectly through the acquisition of smaller competitors. 22 We follow Lewellen and Lewellen (2016) and replace INV, the dependent variable in equation (2), with the change in fixed assets net of depreciation divided by total assets ( FA). Table VIII presents the results from a dynamic difference GMM estimation of this modified version of equation (2). In the first column, to assess whether the relation we find between investment and freight rates holds when we use FA as the investment ratio, we do not consider OC and the interaction term log(csi) OC in the model. The coefficient on the variable of interest, log(csi), is positive and indicates that investment, measured by FA, increases with freight rates. A one standard deviation change in log(csi) results in an increase of 0.04 in FA, equivalent to 66% of the mean of FA in our sample. In column 2, we introduce OC and log(csi) OC to test whether OC affects the efficiency of capital allocation. As in our main analysis, we find that the coefficient on log(csi) OC is positive, indicating that ownership concentration leads to larger adjustments of investment in response to increases in investment opportunities. Insert Table VIII here 22 Brooks and Ritchie (2006) show that a large portion of acquisitions in their sample of shipping M&A transactions was undertaken to acquire assets. 29

31 For the separate analysis of financial and strategic investor firms in columns 3 and 4, we find that log(csi) and log(csi) OC are positive in both subsamples. Although the coefficient on the interaction term is larger in the sample of strategic investor firms, the main effect, captured by the coefficient on log(csi), is substantially larger in the sample of financial investor firms. It follows that, although OC increases the efficiency of capital allocation in both subsamples, the increase in investment in response to changes in freight rates is larger in financial investor firms at every possible level of ownership concentration, as by definition this variable can only take values between 0 and 1. Figure II illustrates this notion by comparing the change in investment, measured by FA, in response to changes in freight rates at different OC levels between the financial and strategic investor firm subsamples; the implied reaction is always stronger in the financial investor sample. Insert Figure II here 4.3. Ownership concentration and accounting performance In our main analysis, we focus on the effect of ownership concentration on firm value in the shipping industry. We now repeat this analysis and replace MB, our measure of firm value, with return on assets (ROA), an accounting-based measure of performance. A similar approach is taken by Tsionas et al. (2012), who show that accounting performance is enhanced by ownership concentration in a cross-section of shipping firms in the year Therefore, this adjustment not only serves as a robustness test of our own results, but also extends the insights from Tsionas et al. (2012) to a panel data setting, which allows us to study the relation over time and to control for unobserved heterogeneity. Table IX presents the results from a dynamic difference GMM estimation of the ROA version of equation (4). We define ROA as income before extraordinary items divided by total assets. The evidence obtained with the accounting-based performance measure corroborates 30

32 our results from the firm value models in Table VI. Ownership concentration affects ROA positively, as indicated by the positive coefficient on OC in column 1. Similar to the results reported with firm value as dependent variable, this effect is driven by the subsample of financial investor firms and cannot be observed for the strategic investor firm sample, as indicated by the insignificant coefficient on OC in column 3. Insert Table IX here 5. Conclusions We study the effect of ownership concentration on the efficiency of capital allocation of shipping firms. The shipping industry provides an ideal case study for at least two reasons. First, due to the capital intensive nature of commercial shipping, efficient investment decisions are a key factor that determines success in this sector. Second, firms in the shipping industry generate most of their revenues from a standardized, homogeneous good; namely, providing capacity for the transportation of cargo. Therefore, we adopt an innovative methodological approach and use a freight rate index to proxy for the investment opportunities of shipping firms. Our empirical strategy is especially suitable because freight rates are an observable and standardized measure of the potential income stream from owning a commercial vessel and providing transportation capacity. We document a positive link between investment and freight rates in the shipping industry. Firms increase investments in response to increases in freight rates, and reduce investments when freight rates decline. More importantly, this response is strengthened by the level of ownership concentration, especially when the largest owner is a financial investor. We conclude that a large owner increases the efficiency of capital allocation. Our empirical approach and the use of panel data enable us to control for the potentially endogenous nature of a firm s ownership structure as well as for time-constant firm-specific factors that deter- 31

33 mine capital allocation efficiency. Moreover, we show that our results are not driven by overinvestment of firms with high ownership concentration. Ruling out the overinvestment argument, we find that firms with high ownership concentration not only invest more when freight rates are high, but also reduce investments to a larger extent when freight rates are low. As expected, the increase in the efficiency of capital allocation of shipping firms attributable to ownership concentration is also reflected in firm value. In line with the findings from the investment models, the better firm performance is driven by financial investor firms, i.e., firms with an owner that aims to increase efficiency to maximize firm value and realize capital gains. Our results are robust to numerous robustness tests, including changes in the measure of investment opportunities, a more stringent classification of firms in subsamples based on owner types, and the consideration of multiple large owners in the analysis. Finally, the results from our investment and performance models continue to hold when we use alternative measures for the dependent variables. 32

34 References Ahrends, M., W. Drobetz, and N. Nomikos, 2017, Corporate cash holdings in the shipping industry, Transportation Research Part E: Logistics and Transportation Review, forthcoming. Aivazian, V., Y. Ge, and J. Qiu, 2005, Debt maturity structure and firm investment, Financial Management 34 (4), Albertijn, S., W. Bessler, and W. Drobetz, 2011, Financing shipping companies and shipping operations: A risk-management perspective, Journal of Applied Corporate Finance 23 (4), Alexandridis, G., and M. Singh, 2016, Mergers and acquisitions in shipping, in: M. Kavussanos, I. Visvikis (eds.), The International Handbook of Shipping Finance. Theory and Practice, Springer. Alexandrou, G., D. Gounopoulos, and H. Thomas, 2014, Mergers and acquisitions in shipping, Transportation Research Part E: Logistics and Transportation Review 61, Alti, A., 2003, How sensitive is investment to cash flow when financing is frictionless?, The Journal of Finance 58 (2), Arellano, M., and S. Bond, 1991, Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations, The Review of Economic Studies 58 (2), Asker, J., J. Farre-Mensa, and A. Ljungqvist, 2014, Corporate investment and stock market listing: A puzzle?, The Review of Financial Studies 28 (2), Attig, N., S. Cleary, S. El Ghoul, and O. Guedhami, 2012, Institutional investment horizon and investment-cash flow sensitivity, Journal of Banking & Finance 36(4), Attig, N., S. El Ghoul, and O. Guedhami, 2009, Do multiple large shareholders play a corporate governance role? Evidence from East Asia, Journal of Financial Research 32 (4), Attig, N., O. Guedhami, and D. Mishra, 2008, Multiple large shareholders, control contests, and implied cost of equity, Journal of Corporate Finance 14 (5), Berle, A., and G. Means, 1932, The modern corporation and private property. Transaction Publishers. Brooks, M, and P. Ritchie, 2006, Mergers and acquisitions in the maritime transport industry , Transportation Journal 45 (2), Chen, H., and S. Chen, 2012, Investment-cash flow sensitivity cannot be a good measure of financial constraints: Evidence from the time series, Journal of Financial Economics 103 (2), Cho, M.-H., 1998, Ownership structure, investment, and the corporate value: An empirical analysis, Journal of Financial Economics 47 (1),

35 Claessens, S., S. Djankov, and L. Lang, 2000, The separation of ownership and control in East Asian corporations, Journal of Financial Economics 58 (1), Demsetz, H., 1983, The structure of ownership and the theory of the firm, The Journal of Law and Economics 26 (2), Demsetz, H., and K. Lehn, 1985, The structure of corporate ownership: Causes and consequences, Journal of Political Economy 93 (6), Demsetz, H., and B. Villalonga, 2001, Ownership structure and corporate performance, Journal of Corporate Finance 7 (3), Drobetz, W., D. Gounopoulos, A. Merikas, and H. Schröder, 2013, Capital structure decisions of globally-listed shipping companies, Transportation Research Part E: Logistics and Transportation Review 52, Drobetz, W., R. Haller, and I. Meier, 2016, Cash flow sensitivities during normal and crisis times: Evidence from shipping, Transportation Research Part A: Policy and Practice 90, Erickson, T., and T. Whited, 2000, Measurement error and the relationship between investment and q, Journal of Political Economy 108 (5), Faccio, M., L. Lang, and L. Young, 2001, Dividends and expropriation, American Economic Review 91 (1), Faccio, M., and L. Lang, 2002, The ultimate ownership of Western European corporations, Journal of Financial Economics 65 (3), Fazzari, S., R. Hubbard, and B. Petersen, 1988, Financing constraints and corporate investment, Brookings Papers on Economic Activity 1, Gilje, E., and J. Taillard, 2016, Do private firms invest differently than public firms? Taking cues from the natural gas industry, The Journal of Finance 71 (4), Goergen, M., and L. Renneboog, 2001, Investment policy, internal financing and ownership concentration in the UK, Journal of Corporate Finance 7 (3), Hadlock, C., 1998, Ownership, liquidity, and investment, The RAND Journal of Economics 29 (3), Hayashi, F., 1982, Tobin s marginal q and average q: A neoclassical interpretation, Econometrica 50 (1), Hermalin, B., and M. Weisbach, 1991, The effects of board composition and direct incentives on firm performance, Financial Management 20 (4), Himmelberg, C., R. Hubbard, and D. Palia, 1999, Understanding the determinants of managerial ownership and the link between ownership and performance, Journal of Financial Economics 53 (3), Holderness, C., R. Kroszner, and D. Sheehan, 1999, Were the good old days that good? Changes in managerial stock ownership since the great depression, The Journal of Finance 54 (2),

36 Kaplan, S., and L. Zingales, 1997, Do investment-cash flow sensitivities provide useful measures of financing constraints?, The Quarterly Journal of Economics 112 (1), La Porta, R., F. Lopez-de-Silanes, and A. Shleifer, 1999, Corporate ownership around the world, The Journal of Finance 54 (2), Laeven, L., and R. Levine, 2007, Complex ownership structures and corporate valuations, The Review of Financial Studies 21 (2), Lambertides, N., and C. Louca, 2008, Ownership structure and operating performance: Evidence from the European maritime industry, Maritime Policy & Management 35 (4), Lewellen, J., and K. Lewellen, 2016, Investment and cash flow: New evidence, Journal of Financial and Quantitative Analysis 51 (4), Loderer, C., and K. Martin, 1997, Executive stock ownership and performance tracking faint traces, Journal of Financial Economics 45 (2), Maury, B., and A. Pajuste, 2005, Multiple large shareholders and firm value, Journal of Banking & Finance 29 (7), McConnell, J., and H. Servaes, 1990, Additional evidence on equity ownership and corporate value, Journal of Financial Economics 27 (2), Merikas, A., D. Gounopoulos, and C. Nounis, 2009, Global shipping IPOs performance, Maritime Policy & Management 36 (6), Morck, R., A. Shleifer, and R. Vishny, 1988, Management ownership and market valuation: An empirical analysis, Journal of Financial Economics 20, Mortal, S., and N. Reisel, 2013, Capital allocation by public and private firms, Journal of Financial and Quantitative Analysis 48 (1), Moshirian, F., V. Nanda, A. Vadilyev, and B. Zhang, 2017, What drives investment-cash flow sensitivity around the World? An asset tangibility Perspective, Journal of Banking & Finance 77, Oliner, S., and G. Rudebusch, 1992, Sources of the financing hierarchy for business investment, The Review of Economics and Statistics 74 (4), Pindado, J., I. Requejo, and C. de la Torre, 2011, Family control and investment-cash flow sensitivity: Empirical evidence from the Euro zone, Journal of Corporate Finance 17 (5), Stopford, M., 2009, Maritime economics (3 rd edition), Taylor & Francis. Syriopoulos, T., and M. Tsatsaronis, 2011, The corporate governance model of the shipping firms: Financial performance implications, Maritime Policy & Management 38 (6), Thomsen, S., and T. Pedersen, 2000, Ownership structure and economic performance in the largest European companies, Strategic Management Journal 21 (6),

37 Tobin, J., 1969, General equilibrium approach to monetary theory, Journal of Money, Credit and Banking 1 (1), Tsionas, M., A. Merikas, and A. Merika, 2012, Concentrated ownership and corporate performance revisited: The case of shipping, Transportation Research Part E: Logistics and Transportation Review 48 (4), Wurgler, J., 2000, Financial markets and the allocation of capital, Journal of Financial Economics 58 (1),

38 Tables Table I Shipping firms by country This table presents the distribution of shipping firms in our sample by country of incorporation. The sample consists of 126 listed shipping firms from 33 countries. Annual financial data is obtained from Compustat North America and Compustat Global. Ownership data is from Thomson Reuters Eikon Ownership & Profiles module. We restrict our sample to firms with at least four consecutive years of non-missing data for the variables in our empirical specifications. The sample covers the period from 1997 to Country Firms Country Firms United Arab Emirates 1 Japan 8 Belgium 2 South Korea 5 Bahamas 1 Liberia 3 Bermuda 11 Luxembourg 1 Canada 1 Latvia 1 Chile 1 Marshall Islands 25 China 6 Malaysia 4 Cayman Islands 1 Norway 9 Germany 1 Pakistan 1 Denmark 4 Philippines 1 Hong Kong 1 Saudi Arabia 1 Indonesia 2 Singapore 2 India 5 Sweden 5 Ireland 1 Thailand 3 Italy 1 Taiwan 8 Jersey 1 U.S. 8 Jordan 1 Total

39 Table II Largest owner This table presents the firm-year observations for the largest owner of the 126 shipping firms in our sample by type of ownership. Annual financial data is obtained from Compustat North America and Compustat Global. Ownership data and type of ownership is from Thomson Reuters Eikon Ownership & Profiles module. We restrict our sample to firms with at least four consecutive years of non-missing data for the variables in our empirical specifications. The sample covers the period from 1997 to Type of owner Obs. Category Bank and Trust 15 Financial Corporation 723 Strategic Government Agency 48 Strategic Hedge Fund 24 Financial Holding Company 82 Strategic Individual Investor 181 Strategic Insurance Company 12 Financial Investment Advisor 137 Financial Investment Advisor/Hedge Fund 103 Financial Other Insider Investor 14 Strategic Pension Fund 5 Financial Private Equity 10 Financial Research Firm 1 Strategic Sovereign Wealth Fund 14 Financial Venture Capital 1 Financial 38

40 Table III Descriptive statistics This table reports the number of observations (Obs.), the mean, the standard deviation (S.D.), the 25 th (P25), the 50 th (P50), and the 75 th (P75) percentiles of the variables in our empirical specification. The sample consists of 126 listed shipping firms from 33 countries. Annual financial data is obtained from Compustat North America and Compustat Global. Annual ownership data is from Thomson Reuters Eikon Ownership & Profiles module. We restrict our sample to firms with at least four consecutive years of non-missing data for the variables in our empirical specifications. The sample covers the period from 1997 to OC is ownership concentration defined as the percentage share of a firm s largest shareholder. SECO is the percentage share of the second largest owner. INV is the firm s investment ratio, defined as capital expenditures minus asset sales, divided by total assets. CF is the firm s cash flow, defined as income before extraordinary items plus depreciation, divided by total assets. CSI is the 12-month average of the monthly Clarksons ClarkSea Index. log(csi) is the natural logarithm of CSI. MB is the firm s market-to-book ratio, defined as the market value of assets divided by the book value of assets. LEV is the firm s leverage ratio, defined as total debt divided by total assets. CASH is the firm s cash and cash equivalents divided by total assets. DIV is the ratio of dividends to total assets. IA is the ratio of intangible assets to total assets. AGE is the number of years since the firm s IPO. TA is total assets in million USD. All financial variables are winsorized at the 1 st and 99 th percentiles. Variable Obs. Mean S.D. P25 P50 P75 OC 1, SECO 1, INV 1, CF 1, CSI 1,370 17, , , , , log(csi) 1, MB 1, LEV 1, CASH 1, DIV 1, IA 1, AGE 1, TA 1,370 2, , ,

41 Table IV Investment, freight rates, and ownership concentration This table reports estimates from a dynamic difference GMM (Arellano and Bond, 1991) regression of a firm s investment ratio (INV) on ownership concentration (OC), the natural logarithm of the 12-month average of the Clarkson ClarkSea index (log(csi)), and an interaction term of the two variables (log(csi) OC). Our sample consists of 126 listed shipping firms from 33 countries. Annual financial data is obtained from Compustat North America and Compustat Global. Annual ownership data is from Thomson Reuters Eikon Ownership & Profiles module. We restrict our sample to firms with at least four consecutive years of non-missing data for the variables in our empirical specifications. The sample covers the period from 1997 to We use three lags of all righthand side variables as instruments, except for log(csi), which is considered as strictly exogenous. In columns 3 and 4, we split the sample into financial and strategic investor firms. A firm is classified as a financial investor firm if a financial investor is the largest owner in at least one year and as strategic investor firm otherwise. All financial variables are winsorized at the 1 st and 99 th percentiles. *, **, and *** denote statistical significance at the 1%, 5%, and 10% level, respectively. Financial Strategic (1) (2) (3) (4) INV t *** *** *** (58.02) (97.63) (1.11) (54.95) OC *** *** * (-5.28) (-50.64) (-1.77) CF 0.036*** *** *** 0.021*** (3.88) (9.19) (6.36) (6.91) log(csi) 0.032*** *** *** 0.023*** (21.86) (14.38) (11.82) (5.53) log(csi) OC *** *** (5.64) (50.44) (0.67) Observations 1,118 1, Firms AR(1) AR(2) Hansen (p-value)

42 Table V High vs. low freight rate environments This table reports estimates from a dynamic difference investment GMM (Arellano and Bond, 1991) regression. The dependent variable is INV, the firm s investment ratio. The independent variables are the lagged investment ratio (INV t-1 ), ownership concentration (OC), the firm s cash flow (CF), and two dummies (CSI_HIGH and CSI_LOW) together with their respective interaction term with OC. CSI_HIGH (CSI_LOW) is a dummy that takes the value of 1 during periods when the 12-month average Clarkson ClarkSea index (CSI) is in the top (bottom) quintile of its distribution over our sample period, and 0 otherwise. Our sample consists of 126 listed shipping firms from 33 countries. Annual financial data is obtained from Compustat North America and Compustat Global. Annual ownership data is from Thomson Reuters Eikon Ownership & Profiles module. We restrict our sample to firms with at least four consecutive years of non-missing data for the variables in our empirical specifications. The sample covers the period from 1997 to We use three lags of all right-hand side variables as instruments, except for log(csi), which is considered as strictly exogenous. In columns 3 and 4, we split the sample into financial and strategic investor firms. A firm is classified as a financial investor firm if a financial investor is the largest owner in at least one year and as strategic investor firm otherwise. All financial variables are winsorized at the 1 st and 99 th percentiles. *, **, and *** denote statistical significance at the 1%, 5%, and 10% level, respectively. Financial Strategic (1) (2) (3) (4) INV t *** 0.137*** 0.024* 0.218*** (71.18) (49.68) (1.78) (37.67) OC 0.043*** ** *** (7.25) (-2.24) (-3.06) CF 0.075*** 0.051*** 0.075*** 0.065*** (9.13) (13.99) (4.28) (15.82) CSI_HIGH 0.015*** 0.016*** 0.011*** 0.016* (16.05) (9.64) (4.25) (1.98) CSI_HIGH OC 0.021*** 0.120*** *** (5.63) (12.68) (-3.01) CSI_LOW *** *** *** (-11.65) (-15.03) (-6.85) (-0.29) CSI_LOW OC *** *** (-4.11) (-0.34) (-6.75) Observations 1,118 1, Firms AR(1) AR(2) Hansen (p-value)

43 Table VI Ownership concentration and firm value This table reports estimates from a dynamic difference GMM (Arellano and Bond, 1991) regression of the firm s market-to-book ratio (MB) on ownership concentration (OC) and additional controls. In columns 1 to 3 we include the firm s lagged market-to-book ratio (MB t-1 ). In columns 4 to 6 we add the firm s leverage ratio (LEV), the firm s investment ratio (INV), the firm s dividend ratio (DIV), the natural logarithm of total assets in USD (SIZE), the ratio of intangibles to total assets (IA), the firm s cash flow (CF), and the firm s age, defined as years since its IPO (AGE). All regression equations include (unreported) year dummies. Our sample consists of 126 listed shipping firms from 33 countries. Annual financial data is obtained from Compustat North America and Compustat Global. Annual ownership data is from Thomson Reuters Eikon Ownership & Profiles module. We restrict our sample to firms with at least four consecutive years of non-missing data for the variables in our empirical specifications. The sample covers the period from 1997 to We use three lags of all right-hand side variables as instruments, except for the year dummies, which are considered as strictly exogenous. In columns 3 and 4 and columns 5 and 6, we split the sample into financial and strategic investor firms. A firm is classified as a financial investor firm if a financial investor is the largest owner in at least one year and as strategic investor firm otherwise. All financial variables are winsorized at the 1 st and 99 th percentiles. *, **, and *** denote statistical significance at the 1%, 5%, and 10% level, respectively. Financial Strategic Financial Strategic (1) (2) (3) (4) (5) (6) MB t *** *** *** *** *** 0.153*** (45.07) (12.98) (10.19) (13.25) (5.90) (4.32) OC 0.058*** ** ** * (2.86) (2.27) (-1.32) (2.59) (1.84) (-0.81) LEV *** (4.63) (-0.85) (1.33) INV ** (0.59) (-0.50) (2.45) DIV ** (1.55) (-1.59) (2.17) SIZE *** *** *** (-17.89) (-4.52) (-4.55) IA *** *** (3.58) (2.67) (-0.50) CF *** (5.14) (0.20) (0.67) AGE *** *** (7.19) (1.58) (3.38) Observations 1, , Firms AR(1) AR(2) Hansen (p-value)

44 Table VII Alternative specifications This table reports estimates from different dynamic investment difference GMM (Arellano and Bond, 1991) regressions. The dependent variable is the firm s investment ratio (INV) in all specifications. We also include the firm s lagged investment ratio (INV t-1 ), ownership concentration (OC), and firm s cash flow (CF) as independent variables in all models. In columns 1 to 3, we add the firm s market-to-book ratio (MB) as a measure of investment opportunities instead of the freight rate index. In columns 4 and 5, we split the sample into firms with a largest investor that is more likely to become active vs. passive investor firms. Detailed definitions are provided in the text. In columns 6 to 8, we add a dummy (SECO) that takes the value of 1 if the second largest owner holds more than 10% of all shares, and 0 otherwise. We also include the interaction between this dummy and the variable of interest, log(csi). Our sample consists of 126 listed shipping firms from 33 countries. Annual financial data is obtained from Compustat North America and Compustat Global. Annual ownership data is from Thomson Reuters Eikon Ownership & Profiles module. We restrict our sample to firms with at least four consecutive years of non-missing data for the variables in our empirical specifications. The sample covers the period from 1997 to We use three lags of all right-hand side variables as instruments, except for the year dummies and log(csi), which are considered as strictly exogenous. In columns 2 and 3, and columns 7 and 8, we split the sample into financial and strategic investor firms. A firm is classified as a financial investor firm if a financial investor is the largest owner in at least one year and as strategic investor firm otherwise. All financial variables are winsorized at the 1 st and 99 th percentiles. *, **, and *** denote statistical significance at the 1%, 5%, and 10% level, respectively. Alternative investment opportunity measure Active vs. passive Second largest owner Financial Strategic Active Passive Financial Strategic (1) (2) (3) (4) (5) (6) (7) (8) INV t *** ** *** 0.119*** *** (9.63) (-0.23) (2.26) (0.46) (22.13) (23.20) (-0.86) (15.09) OC *** *** *** *** *** *** 0.265* (-3.21) (-4.44) (0.52) (7.78) (-3.93) (-16.21) (-9.37) (1.82) SECO 0.174*** *** 0.429*** (37.46) (-3.13) (31.54) CF *** ** * (-1.28) (0.61) (-1.65) (4.41) (-0.52) (-0.93) (2.23) (-1.91) log(csi) * *** 0.035*** 0.031*** 0.065*** (1.93) (13.69) (23.58) (5.30) (8.47) log(csi) OC *** *** 0.043*** 0.130*** ** (7.99) (3.17) (15.07) (9.64) (-2.16) log(csi) SECO *** 0.009** *** (-38.51) (2.57) (-30.47) MB t *** *** (4.27) (-1.22) (2.87) MB t-1 OC *** 0.448*** (3.82) (6.70) (-0.14) Observations 1, , Firms AR(1) AR(2) Hansen (p-value)

45 Table VIII Total investment, freight rates, and ownership concentration This table reports estimates from a dynamic difference GMM (Arellano and Bond, 1991) regression of a firm s total investment, defined as the change in fixed assets ( FA), on ownership concentration (OC), the natural logarithm of the 12-month average of the Clarkson ClarkSea index (log(csi)), and an interaction term of the two variables (log(csi) OC). Our sample consists of 126 listed shipping firms from 33 countries. Annual financial data is obtained from Compustat North America and Compustat Global. Annual ownership data is from Thomson Reuters Eikon Ownership & Profiles module. We restrict our sample to firms with at least four consecutive years of non-missing data for the variables in our empirical specifications. The sample covers the period from 1997 to We use three lags of all right-hand side variables as instruments, except for log(csi), which is considered as strictly exogenous. In columns 3 and 4, we split the sample into financial and strategic investor firms. A firm is classified as a financial investor firm if a financial investor is the largest owner in at least one year and as strategic investor firm otherwise. All financial variables are winsorized at the 1 st and 99 th percentiles. *, **, and *** denote statistical significance at the 1%, 5%, and 10% level, respectively. FA FA FA FA Financial Strategic (1) (2) (3) (4) FA t *** *** *** *** (32.06) ( ) (-9.53) (-24.34) OC *** *** *** (-14.56) (-8.59) (-20.56) CF 0.187*** 0.160*** 0.315*** 0.295*** (12.04) (49.72) (7.57) (30.39) log(csi) 0.091*** 0.073*** 0.090*** 0.016*** (39.70) (82.94) (15.95) (3.87) log(csi) OC 0.079*** 0.129*** 0.145*** (26.51) (8.63) (20.30) Observations 1,016 1, Firms AR(1) AR(2) Hansen (p-value)

46 Table IX Ownership concentration and accounting performance This table reports estimates from a dynamic difference GMM (Arellano and Bond, 1991) regression of the firm s return on assets (ROA) on ownership concentration (OC) and additional controls. We include the firm s lagged return on assets (ROA t-1 ), the firm s leverage ratio (LEV), the firm s investment ratio (INV), the firm s dividend ratio (DIV), the ratio of intangibles to total assets (IA), the firm s cash flow (CF), and the firm s age (AGE). All regression equations include (unreported) year dummies. Our sample consists of 126 listed shipping firms from 33 countries. Annual financial data is obtained from Compustat North America and Compustat Global. Annual ownership data is from Thomson Reuters Eikon Ownership & Profiles module. We restrict our sample to firms with at least four consecutive years of non-missing data for the variables in our empirical specifications. The sample covers the period from 1997 to We use three lags of all right-hand side variables as instruments, except for the year dummies, which are considered as strictly exogenous. In columns 2 and 3, we split the sample into financial and strategic investor firms. A firm is classified as a financial investor firm if a financial investor is the largest owner in at least one year and as strategic investor firm otherwise. All financial variables are winsorized at the 1 st and 99 th percentiles. *, **, and *** denote statistical significance at the 1%, 5%, and 10% level, respectively. ROA ROA ROA Financial Strategic (1) (2) (3) ROA t ** * (-2.11) (-0.73) (-1.79) OC *** ** (3.84) (2.13) (0.53) LEV *** (4.61) (0.98) (-1.32) INV *** *** 0.037*** (15.60) (6.22) (5.17) DIV *** (-4.91) (-0.60) (-1.54) SIZE *** *** (14.28) (3.45) (1.03) IA *** *** (15.13) (3.13) (0.36) CF *** *** 1.006*** (259.73) (47.96) (155.23) AGE *** *** (-7.53) (-4.02) (-1.44) Observations 1, Firms AR(1) AR(2) Hansen (p-value)

47 Figures Figure I. Investment and freight rates. This graph plots the annual mean investment ratio of the shipping firms in our sample and the Clarkson ClarkSea index. The solid line is the annual Clarkson ClarkSea index, the bars represent the mean investment ratio of all firms included in our sample in the respective year. 46

48 Figure II. Subsample comparison. This graph plots the sensitivity of investment, using the broader investment measure ( FA), to changes in the freight rate index at different levels of ownership concentration (OC) for the financial and strategic investor firm subsamples. The graphic representation is based on the derivation of the estimation equation used in Table VIII with respect to log(csi). Using the estimates in columns 3 and 4 of Table VIII yields the following functions for the financial and strategic investor firm subsamples: δ FA Financial investor firms: it = β δlog(csi t ) 4 + β 5 OC it = OC it δ FA Strategic investor firms: it = β δlog(csi t ) 4 + β 5 OC it = OC it. The solid black line represents the financial investor firm sample, whereas the dashed grey line represents the strategic investor firm sample. 47