On the relationship of stewardship and valuation – Empirical evidence from German firms

Abstract This empirical study examines the association between the stewardship and valuation roles of financial reporting. While analytical literature based on agency theory has hinted at differences between the two objectives, recent empirical studies based on US data suggest a positive relationship. This paper contributes to the literature by providing evidence from a context that differs from the US with regard to the corporate governance setting. We present empirical findings on the relationship using data from German firms between 2006 and 2013, all preparing their financial statements according to International Financial Reporting Standards (IFRS). We find a positive empirical relation between our proxies of valuation and stewardship in univariate and multivariate settings, the latter including firm and corporate governance factors. Our study might be of interest for the International Accounting Standards Board (IASB), which currently is reconsidering the importance of stewardship in its conceptual framework, as we provide support for the standard-setter’s assessment of a generally positive relationship of the two roles of financial reporting information. Keywords: Decision usefulness, IFRS, stewardship, valuation

1

1

Introduction

Financial reporting serves multiple purposes. In particular, the role of financial reporting information in informing capital market participants’ investment decisions (valuation usefulness) and its role in providing incentives for employed managers (stewardship usefulness) have repeatedly been analysed in the accounting literature. Bushman and Smith (2001, p. 270) maintain that “it is widely accepted that reported earnings play a fundamental role in both determining stock prices, and in evaluating and compensating executives. However, the relation between the governance and value relevance of earnings is not well understood.” While some theoretical literature has analysed the relationship of stewardship and valuation (e.g. Gjesdal, 1981; Paul, 1992), “there is very little evidence on whether there is any association between these two roles” (Banker et al., 2009, p. 648; also see O’Connell, 2007). We concur with Bushman, Engel, and Smith (2006) that analysing the relations among different roles of accounting information is essential to gain a better understanding of the forces that shape accounting and reporting practices. In this vein, the present paper aims at investigating empirically the research question of how stewardship and valuation are related. To address this research question we use data from German firms listed in the regulated market of the German stock exchange. We use hand-collected data on CEO and executive board compensation and accounting and market data from Compustat and Datastream for the period between 2006 and 2013. As German listed firms are required to apply IFRS since 2005, our study analyses the relation of the two purposes based on IFRS financial statement information. While recent studies have investigated the relationship of stewardship and valuation using US data (Bushman et al., 2006; Banker et al., 2009), we are not aware of any study analysing the relationship for a setting where IFRS are applied. As the importance of stewardship in US-GAAP has been less pronounced than in IFRS (e.g. Whittington, 2008; Zeff, 2013), an analysis of empirical data based on IFRS provides additional insights into the relationship. Moreover, and more importantly, conceptual and analytical literature has emphasized that the relationship of the two objectives might be affected by the local corporate governance and capital market setting (Walker, 2010; Kuhner and Pelger, 2015). Therefore, findings for the US context that reveal a positive relation between proxies for the two objectives (Bushman et al., 2006; Banker et al., 2009), might not hold in different environments. Given the context-specific roles of financial accounting, it is at least not immediately obvious that the relation is similar for settings outside the US. In contrast to the US, in Germany board and ownership structures differ substantially in that there exists a separate supervisory board that is, among other things, responsible for executive pay contracts and managerial ownership is rather low, while families and strategic investors 2

play a more important role (Engelen, 2015). Furthermore, in Germany management pay has used to be relatively low compared to other countries (Conyon and Schwalbach, 2000) and the relation of pay to performance has been found to be rather weak (Rapp and Wolff, 2010). Thus, the specific features of corporate governance make the German context an appropriate setting to analyse the relationship between the stewardship and valuation roles of financial reporting information. Our study follows the approach by Bushman et al. (2006) in analysing the relationship. Therefore, we use three regression models. In the first model, we measure the valuation coefficient of earnings and adopt the approach from prior literature (e.g. Barth et al., 1999; Engel, Hayes, and Wang, 2003; Bushman et al., 2006; Banker et al., 2009). The second model examines the stewardship coefficient of earnings and the third model combines both coefficients to analyse the association between valuation and stewardship. Similar to prior studies, we apply cash compensation and use total compensation for robustness tests. In addition to extant literature, we do not solely focus on the CEO compensation but also consider the compensation received by the executive board as a whole. Our empirical analyses indicate that, in our sample of German firms, stewardship and valuation are positively related. This finding pertains to the cash and total compensation of CEOs, and is visible in the average compensation of executive board members. We use several robustness checks, which in general provide support for the positive relationship. In a further step, we explore whether the relationship is affected when we control for firm factors, such as size, leverage and market-to-book ratio, and corporate governance factors, in particular firm ownership. Our results show that the positive association between valuation and stewardship remains both for CEO and average board compensation. Thus, firm or corporate governance factors do not seem to alter the relationship for our sample. Apart from the academic literature, accounting objectives, and in particular the role of stewardship, have recently been discussed by accounting standard-setters in revisions of their conceptual frameworks. In 2010, the IASB and the FASB decided to drop stewardship as a separate objective in their framework because standard-setters see it as a part (or a sub-objective) of the valuation objective (Pelger, 2016). In its second revision of the framework, following criticism by constituents, the IASB now intends to put more emphasis on stewardship as a part of the objective of financial reporting, but still maintains that there is no need to state stewardship separately as it can be encompassed in the valuation objective (IASB, 2015, 1.3, BC1.6-10). As this statement by the IASB is not necessarily in line with academic research (Cascino et al., 2014), 3

the present paper intends to contribute to the current debate by presenting empirical findings on the relationship between stewardship and valuation drawing on a sample of firms preparing their financial statements according to IFRS. As our results suggest that there is a positive relationship between the two roles of financial reporting information and that this relationship is not affected by controls for firm and governance factors, we provide support for the standard-setters’ statement that valuation useful information is in general also useful for stewardship purposes. However, our findings do not preclude that these objectives might demand different accounting information in specific circumstances. In the following section, we provide some background on the objectives of financial reporting and an overview of the theoretical and empirical literature. Section 3 introduces our research design and our sample data. Section 4 presents our empirical results, while section 5 offers some conclusions.

2 2.1

Literature review Theoretical background

Starting with Gjesdal (1981), several analytical studies have analysed the relationship of the two objectives. Gjesdal (1981) formally distinguishes between the value of accounting information for decision-making by investors and for solving the incentive problem between agent and principal. Starting comparison with the use of the Blackwell theorem combining this approach with an agency model, leads Gjesdal (1981) to conclude that the rankings of different accounting systems might differ for the two purposes. Paul (1992) compares stock-based and accounting-based compensation contracts in a linear agency model. He shows that different accounting signals are aggregated by the capital market in a way that is not necessarily optimal for contracting. He analytically distinguishes the two objectives in that from a stewardship perspective, accounting should be informative about managerial effort, while from a valuation perspective accounting should rather provide information on the stochastic (and thus uncertain) part of firm value. This difference between the two objectives was also delineated by Lambert (2001), while Heinle and Hofmann (2011) extend the agency setting to the availability of soft (non-contractible) information. They show that the disclosure of such information is positive from a valuation perspective but entails negative effects for stewardship purposes. While all these studies regard their findings as corroborating the difference (or trade-off) between the two objectives, it is to stress that the information settings in these models not necessarily capture real-world complexities in information provision and use. Nonetheless, the general take-away from the analytical literature

4

is that there are constellations in which the two objectives do require different information. A recent analytical paper by Kuhner and Pelger (2015) analyses the relationship in an agency setting and reveals that, in their model, the two objectives react similarly to changes in accounting quality parameters (relevance, reliability) but differ in their reaction with respect to opportunities for earnings management offered by the accounting system. Important for our study, the authors stress that the relation of the two objectives hinges on the context, that is, parameters of the manager, the firm and characteristics of the economy or the corporate governance system as a whole. First, we take up these findings as a motivation why a study of the association of stewardship and valuation in a different geographical setting than the US might be warranted. Second, we control for firm-specific factors and corporate governance factors in our multivariate analysis of the relationship of stewardship and valuation. 2.2

Empirical background

Bushman et al. (2006) investigate how accounting information are used for valuation and incentive contracting. They refine the analytical model by Paul (1992) to specify their null hypothesis that there is no relation between the two objectives. The authors find that firm and industry specific valuation and compensation earnings coefficients are significant positively related with respect to changes between two periods of 15 years (before 1986 and between 1986 and 2000). Their cash compensation data was taken from the annual Forbes study and contains 141 to 3,134 firm-years per industry for a 30-year period from 1970 to 2000. They study 28 different industries defined on the basis of two-digit Standard Industry Classification (SIC) codes. As the CEO compensation measure they apply cash compensation, following Core, Guay, and Verrecchia (2003) defined as the salary plus bonus payments. The approach of Bushman et al. (2006) combines a univariate and a multivariate analysis. Their univariate analysis estimates valuation and compensation earnings coefficients firm and industry specific. Firm specific means that for each firm there are at least 20 annual observations during the sample period. The firm specific analysis estimates one valuation coefficient for each of these 379 firms with an average of 26 observations per firm. Industry specific signifies the creation of an industry specific valuation coefficient (for each of the 28 industries) controlling for year fixed effects. The univariate analysis first regresses cumulative market adjusted stock returns on earnings to determine the valuation coefficient of earnings and second quantifies the compensation coefficient of earnings by regressing CEO cash compensation on earnings controlling for public performance information with the proxy of stock market returns. Bushman et al. (2006) employ a Pearson and Spearman rank correlation table with the estimated firm 5

and industry specific valuation and compensation earnings coefficients indicating the association between valuation and stewardship weights placed on earnings. Additionally, a multivariate analysis estimates the relation between the valuation and compensation earnings coefficients controlling for growth opportunities, regulation, noise in earnings and other public performance information. These analyses reveal a significant positive firm and industry specific relationship for the whole period and the two sub-periods. Banker et al. (2009) conduct an empirical study that concentrates on the relationship between the value relevance and incentive contracting relevance of earnings and cash flows for US data between 1993 and 2003. Their sample contains 7,076 CEO-year observations and uses CEO cash compensation in line with Bushman et al. (2006). Their first regression estimates pay-sensitivity and value relevance of earnings and cash flows. In a second regression, Banker et al. (2009) use cross-sectional firm and yearly regressions to analyse the relationship between pay-sensitivity and value relevance coefficients of earnings. They find significant positive coefficients supporting that the compensation weight on earnings is increasing in the value relevance of earnings. The same holds for cash flows. Thus, the main result of their paper is that value relevance of performance measures plays a major role in the use of accounting performance measures for CEOs’ evaluation. In a related study, Gassen (2008) analyses the relation of valuation and stewardship by using a different approach for a dataset of US firms between 1990 and 2005. First, instead of using the concept of value relevance, that is employed by Bushman et al. (2006) and Banker et al. (2009), he uses an event study that focuses on the effect that the publication of financial statement information has one share price in a certain (short) time frame. Second, Gassen (2008) does not use compensation data to approach the stewardship perspective, but instead models a market for stewardship information, where the supply of stewardship information is proxied by conditional conservatism (asymmetric timeliness of earnings) and the demand for stewardship information by the importance of non-equity stakeholders. Gassen (2008) finds a negative relation between the valuation relevance of earnings and his proxies for stewardship. However, it is to note that his definition of stewardship is not directly following the principal-agent setting between owners and employed managers, usually considered by the academic literature (and the standard-setters) when referring to stewardship. Instead, his definition of stewardship rather reflects a more general contracting perspective (Watts, 2003), considering other stakeholders than owners. Thus, he provides evidence that valuation useful information might not necessarily be useful for such a broader notion of contracting.

6

The effect of mandatory IFRS adoption on value relevance in relation to equity markets shows different evidence for European countries. Devalle, Onali, and Magarini (2010) analysed IFRS data with respect to their valuation relevance and find an increase in value relevance of earnings and a decrease in value relevance of book value of equity with the adoption of IFRS in Germany and France. In contrast, the adoption of IFRS in the UK results in an increase in value relevance of earnings and book value of equity, while the IFRS adoption in Italy and Spain results in a decrease in both measures. Their sample contains 3,721 firms and observes a period between 2002 and 2007. Choi, Peasnell, and Toniato (2013) investigate the accuracy of analysts’ forecasts for IFRS adoption between 2003 and 2007 in the UK and state that value relevance is significantly higher under IFRS.1 A few studies have analysed IFRS data with respect to their stewardship relevance. In particular, these studies focus on the effect of IFRS adoption on the relation of accounting measures and CEO/executive compensation. For instance, Voulgaris et al. (2014) examine the effect of IFRS adoption in the UK on the type of performance measures firms use to evaluate and compensate their managers. Their data is hand-collected and includes 3,000 UK firms over eight years (2002-2009), four years before and after IFRS adoption each. Their findings indicate that IFRS adoption results in a decrease in the use of accounting numbers for managerial performance measurement purposes. These findings support the arguments by Watts (2006) saying that IFRS are not appropriate for the use in managerial incentive contracts because of the discretion offered by its standards, and in particular by fair value measurement. Another study with European data, by Ozkan, Singer, and You (2012) analyses the mandatory adoption of IFRS in Europe and the contractual usefulness of accounting information in executive information during the period 2002–2008. They distinguish between two sub-periods excluding the year of IFRS adoption 2005. One sub-period contains the pre–mandatory adoption period (2002-2004) and the other represents the post–mandatory adoption period (2006-2008). Ozkan et al. (2012) focus on the change in executive cash compensation.2 Their sample includes 13,505 executive-year observations for 3,046 firm-years of 892 publicly listed companies from 15 European countries. In a first estimation, they analyse if compensation committees base executive compensation more on an accounting measure of the firm’s performance after mandatory IFRS adoption. In a second regression, they explore if the general use of accounting based on relative performance evaluation from the pre- to post-IFRS adoption period has changed. In a third regression, they estimate the sensitivity of executive pay-to-stock-return 1

For a summary see Br¨ uggemann, Hitz, and Sellhorn (2013). Their paper investigates cash compensation because contractual usefulness of accounting performance measures, which is the focus of their analysis, refers to the cash component rather than to the equity component of executive compensation. 2

7

performance and the change in this sensitivity after the adoption of IFRS. Ozkan et al. (2012) find that European companies increased the use of accounting-based pay-performance sensitivity as compensation committees placed a higher weight on accounting earnings in compensation contracts when the adoption of IFRS entailed a major accounting change. All in all, our presentation of the theoretical and empirical literature has shown that, on the one hand, there is the expectation based on analytical results that optimal valuation information does not automatically give rise to optimal information for stewardship purposes. However, there is less theoretical evidence on how pervasive the possible conflict between the two purposes actually is. Recent analytical literature suggests that the (firm and governance) context plays a role for the nature of the relationship. On the other hand, several empirical studies that have explicitly focused on the relationship between stewardship and valuation, using compensation data from the US, have shown significant positive associations. Thus, for our empirical analysis we do not form any specific expectation regarding the relationship.

3 3.1

Research design Empirical model

Our study investigates the association between the stewardship and valuation roles of financial reporting. Following Bushman et al. (2006), we examine this relation in a three-step regression model. We establish a model that estimates two single regressions interrelated with a third regression. The first regression estimates valuation relevance (VEC), the second regression estimates stewardship relevance (CEC), while the third regression combines both estimated coefficients and estimates the valuation on stewardship coefficient (VSC). We use industry and firm specific regressions, controlling for year fixed effects in our industry specific regressions. Each regression estimates VECs, CECs and VSCs for each industry/firm with aggregated firm year observations (industry/firm specific). The first regression estimates the effect of earnings (EARN) on the market adjusted stock returns (MRET) for each industry/firm. EARN defines the change in net income between year t and t-1 deflated by the market capitalization at the end of year t-1. The coefficient that results from a change in EARN by one unit on MRET is saved as the valuation earnings coefficient for each industry/firm. This signifies that the firms’ valuation earnings coefficients are aggregated within an industry/for each firm.

8

The same approach is applied for regression (2). The second equation describes the regression we run to estimate the industries’/firms’ compensation earnings coefficients, again aggregating firm years within an industry or for one specific firm. The compensation earnings coefficients represent stewardship relevance by estimating the association between a change in EARN and the percentage change in CEOs’/executives’ cash compensation (C) controlling for the 12-months cumulative stock market return of i’s fiscal year (RET). After running the first two regressions we obtain industry and firm specific valuation and compensation earnings coefficients that are included as independent and dependent variables in the third regression. The third regression then estimates the association between valuation relevance (VEC) and stewardship relevance (CEC), aggregated by industries/firms. Our resulting association (VSC) indicates the relationship between stewardship and valuation based on an industry/firm specific estimation. This industry/firm specific aggregation is necessary to have sufficient variation in the year observations within an industry/for each firm to estimate a valuation and compensation earnings coefficient. Different industry/firm specific valuation and compensation earnings coefficients are important for the variation in the third regression stage and for estimating an overall association between the two objectives.3 We extend our model to a multivariate setting in that we include firm and ownership variables in our third regression model (3). Following Kaserer and Moldenhauer (2008), our third regression model (3) controls for different ownership variables. We use free float, a distinction between inside and outside investors and six different groups of shareholders. The third equation denotes these variables with CorpGovi . Our six shareholder groups are manager, family, bank, strategic, institutional and other investors. Inside investors comprise manager and family, while outside investors include bank, strategic, institutional and other investors. In line with Engelen (2015) and Rapp and Wolff (2010), we include firm variables for growth opportunities (market to book ratio), risk (leverage) and firm size (logarithm of total assets). All firm and ownership variables are medians of firm/industry i that consider i’s firm year observations. In our first regressions we distinguish between three different calculations for the market adjusted stock return (MRET) that differ in including/excluding the reaction of the market on earnings in financial reporting. Our main results concentrate on the 12-months market adjusted stock returns calculated from the beginning to the end of the 3 In contrast, if we ran the first and the second regression over the whole sample, we would obtain exactly one valuation earnings coefficient from the first regression and one compensation earnings coefficient from the second regression. A third regression would thus not provide any meaningful results.

9

fiscal year t (MRET-12m) according to Bushman et al. (2006). For robustness tests we apply two other market adjusted stock return calculation: MRET-15m and MRET-12m3. With MRET-15m that uses 15-months market adjusted stock returns calculated from the beginning of the fiscal year t and ending 3-months after the end of the fiscal year t, we follow Perotti and Wagenhofer (2014). With MRET-12m3 that describes 3-months shifted 12-months returns beginning and ending 3-months after the start/end of the fiscal year, we follow Perotti and Wagenhofer (2014) and Banker et al. (2009).

Empirical model 1. First regression estimates valuation relevance (VEC): M RETi = α0 + V ECi · EARNi + α1 D(t) + εi

(1)

2. Second regression estimates stewardship relevance (CEC): Ci = β0 + CECi · EARNi + β1 RETi + β2 D(t) + εi

(2)

3. Third regression estimates the effect of VEC on CEC (VSC): CECi = γ0 + V SC · V ECi + γ1 F irmi + γ2 CorpGovi + εi

where: i M RETi RETi EARNi Ci V ECi CECi V SCi F irmi CorpGovi D(t)

= = = = = = = = = = =

(3)

Industry (two-digit SIC-codes) or firm The 12-months cumulative market adjusted stock market return of i’s fiscal year; The 12-months cumulative stock market return of i’s fiscal year; The change in deflated earnings between year t and t-1 of i; The percentage change in compensation between year t and t-1 of i; Valuation earnings coefficient of firm i or industry i; Compensation earnings coefficient of firm i or industry i; Valuation on stewardship coefficient of firm i or industry i; Firm variables of firm i or industry i; Governance variables of firm i or industry i; Year dummies for fixed year effects.

10

3.2

Sample and data

We use hand-collected data and data from the commercial databases Compustat Global and Datastream Worldscope for firm years between 2006 and 2013. The hand-collected data contains executive compensation and corporate data. The data from the commercial databases provides financial statement and stock market information for the firms in our sample. The executive compensation information was derived from the annual reports of German listed stock corporations. The dataset on compensation differentiates between the Chief Executive Officer’s (CEOs’) compensation and the average board members’ compensation that is defined as the board members’ compensation (including the CEO) divided by the total number of board members. Furthermore, the dataset distinguishes between annual cash compensation and total compensation. Cash compensation is the sum of the fixed salary, fringe benefits and short term incentives (bonus), while total compensation also includes long term (equity) incentives. Following other empirical studies (e.g. Bushman et al., 2006; Banker et al., 2009), we focus on cash compensation and use total compensation for robustness tests. In line with Bushman et al. (2006), we exclude firm years with a CEO change (Panel A) and include these observations for robustness tests (Panel B). Firm year observations where we have no information about a CEO change are also excluded in Panel A and included in Panel B. We focus on an industry specific (two-digit SIC-specific) and a firm specific analysis of up to 844 German firm year observations. -Insert Table 1 hereTable 1 explains our sample selection. In our analysis we focus on firms listed in the main indices of the German stock exchange, DAX, MDAX, TecDAX and SDAX, in the fiscal years 2006 to 2013. The total number of firms that form part of these indices between 2006 and 2013 is 1280 (160 per year). From this number of firms we excluded foreign firms, firm year observations with incomplete compensation data, firms applying US-GAAP and firm year observations with a lack of data availability (missing earnings, stock market returns, total assets or market capitalization). This reduction leads to 844 firm year observations (159 firms) with available compensation data for the board members. Another 146 firm year observations lack individualized CEO compensation data. Therefore, we have 698 firm year observations with the percentage change in CEO compensation between year t and t-1 for our empirical analysis for Panel B. Our main analysis observes firm years without a CEO change and firm years where we have no information about a CEO change. Therefore, our main sample selection Panel A contains

11

658 firm year observations for the percentage change in average board compensation and 615 firm year observations for the percentage change in CEO compensation. Solely focussing on CEO compensation data may thus give rise to a bias, considering the self-selection of firms not publishing individual compensation of board members. While it is, in general, mandatory to present separate compensation for the members of the management board, companies do have the opportunity to avoid this individualised disclosure if the annual shareholder meeting supports this proposal with a majority of at least 75%.4 Therefore, our analysis also uses the average board compensation in order to check whether this possible bias influences our results regarding the valuation-stewardship relationship. Our industry classification follows the two-digit Standard Industrial Classification (SIC). Table 2 represents the arranged industries naming the two-digit SIC codes. Table 3 provides an overview of our sample data presenting the variable abbreviation, the description and the source. The difference in NICON between year t and t-1 is deflated by the market capitalization at the beginning of year t (MCAP) and represents our independent variable EARN. Our data allows us to calculate cash and total compensation of the CEO and the average among board members (AVG). Dividing the cash/total compensation by the number of board members results in the average cash/total compensation among board members including the CEO. Leverage, firm size and market to book value are additional firm variables. All share variables and free float indicate different groups of shareholders that we include for ownership control purposes. -Insert Table 2 and Table 3 here-

3.3

Descriptive statistics and correlations

Table 4 provides summary statistics of the main sample variables. We present an industry and firm specific overview that includes 698 firm year observations for the CEOs’ compensation and 844 for the board members’ compensation. This leads us to fifteen industry and at least 144 firm specific observations for each variable. This table provides an overview of Panel B that includes firm year observations with a CEO change and with no information about a CEO change. Panel A in contrast excludes these firm year observations. Leverage, firm size, market to book value, free float and all share variables are medians of i’s industry/firm year observations. 4 Hitz and Werner (2012) analyse factors that explain companies’ decisions to avoid individualised disclosure of compensation.

12

-Insert Table 4 hereTable 5 shows the correlation coefficients of the median variables according to Table 4. We can observe positive and highly significant correlation coefficients. The change in cash compensation is not significantly correlated with the stock return (MRET and RET) for CEOs’ firm specific and for average boards’ industry specific median variables. Intervening variables can exist and disrupt the relation between these variables. -Insert Table 5 here-

4 4.1

Results Valuation and stewardship relevance

Table 6 presents the results for the first regressions (equation (1)) using industry and year fixed effects. Although we are not using industry fixed effects in our industry specific regressions that we run for saving the coefficients and that lead to the third industry specific regression, we use industry fixed effects in Table 6, where we present regressions without the aggregation for illustrative purposes. The results for equation (1) with industry and year fixed effects indicate a significant and positive valuation earnings coefficient for all market adjusted stock return measures. In line with Bushman et al. (2006) that use 12-months market adjusted returns, we can state a significant positive association of earnings on market adjusted stock returns. -Insert Table 6 hereTable 7 displays the results of the second regression stage as explained in equation (2) including industry and year fixed effects. We find highly significant positive compensation earnings coefficients for all compensation measures. The 12-months stock market return has a positive effect on the percentage change in compensation. Our results are consistent with Bushman et al. (2006), indicating positive values for the compensation earnings coefficients (CECs). -Insert Table 7 hereTable 8 provides a summary statistic of the estimated industry and firm specific valuation and compensation earnings coefficients. The estimated industry valuation earnings coefficients lie between -344.12 and 1720.92 for the 12-months market adjusted

13

returns. The industry compensation (stewardship) earnings coefficients for the CEOs’ cash compensation are between -771.26 and 22614.45. After deleting the industry Agriculture, because we have only 7 firm year observations (one firm) representing this industry, we include 837 instead of 844 firm year observations in our empirical model. The firm specific valuation and stewardship coefficient lie in a higher range than the industry specific ones. At least six firm year observations estimate on firm specific coefficient. The third regression stage uses these estimated values to regress valuation and stewardship coefficients on each other to obtain a valuation on stewardship coefficient (VSC). -Insert Table 8 here-

4.2

Effect of valuation on stewardship relevance

In Table 9, we show the Pearson and Spearman Rank correlation table to compare the relation of our proxies for stewardship and valuation. While the Pearson correlation analyses linear relationships, the Spearman rank correlation table examines monotonic relationships. The correlation table indicates a positive relation. The Pearson and the Spearman rank correlation coefficients are positive significant for the CEO’ cash compensation for our industry specific regressions. Our firm specific regressions show significant positive correlation coefficients for the CEOs’ and average boards’ compensation in the Pearson correlation table. -Insert Table 9 hereTable 10 summarizes the industry and firm specific associations between valuation and stewardship relevance of earnings. Our industry and firm specific findings indicate a significant positive relationship between valuation and compensation earnings coefficients for CEOs’ cash compensation. Furthermore, our firm specific regressions indicate a significant positive relationship for average boards’ cash compensation. We cannot support these findings for industry specific boards’ cash compensation. Figures 1 and 2 illustrate the industry and firm specific relationship between valuation and stewardship. -Insert Table 10, Figure 1 and Figure 2 here-

14

4.3

Multivariate analysis

Table 11 presents our firm specific results for the third regression stage controlling for different shareholder types, firm size, leverage and market to book ratio. The association between valuation earnings coefficients and stewardship earnings coefficients remains significant positive, while including control variables for ownership, growth opportunities, risk and firm size. The first and the second column of Table 11 summarize the results controlling for free-float, firm size, leverage and market to book value. The third and the fourth column of Table 11 indicate results controlling for inside and outside investors’ shares. The fifth and the sixth column of Table 11 show the results controlling for different shareholder groups’ shares instead of inside and outside investors’ shares. These six shareholder groups are manager, family, strategic, bank, institutional and other shareholder, while inside shareholder are managers or family and outside shareholders are strategic, bank, institutional or other. All control variables are medians for firm i of i’s firm year observations. With CEOs’ and average boards’ cash compensation we can support our prior results of a positive association between valuation and stewardship coefficients of earnings in a multivariate setting. These results hold for our industry specific regressions with the use of CEOs’ cash compensation that we provide in Table 12. In line with prior results for Panel A, average boards’ cash compensation remains not significant in our industry specific univariate and multivariate analyses. This might be the case because we have more aggregation of firm year observations on industry level in contrast to firm level basis. Furthermore, we can observe that free float is significantly positively influencing stewardship coefficients of earnings (CEC), while an increase in outside, family or strategic investors significantly decreases stewardship relevance (CEC). For average boards’ cash compensation CEC is higher in the regression model that includes free float. Therefore, free float might result a stronger connection between accounting and compensation. In contrast, this connection is weaker with a high degree of outside investors. -Insert Table 11 and 12 here-

15

4.4

Robustness tests

We run several additional regressions to verify our results. These results are summarized in Tables 13, 14 and 15. Our industry and firm specific regressions with MRET-12m and total compensation support our significant positive findings for MRET-12m and cash compensation (Table 13). Table 14 uses Panel B instead of Panel A. Panel B also includes firm year observations with a CEO change and firm years for which we have no information about a CEO change. We cannot support our prior findings on an industry specific basis but on a firm specific one with CEOs’ total compensation. The industry specific findings indicate a significant positive relationship for average boards’ total compensation, while our firm specific findings indicate a significant negative association for average boards’ total compensation. These additional findings suggest that the association tends to be stronger when CEO changes are taken into account and not included in our regression models. Table 15 presents industry and firm specific findings for Panel A and B and with market adjusted return over 15 months (MRET15m) and 12 months by 3 months shifted (MRET12m3). Panel A’s industry and Panel A’s and B’s firm specific regressions support our prior findings of a significant positive association for CEOs’ cash and total compensation and for both market adjusted return measures. Average boards’ compensation only shows significant positive associations for our firm specific regressions in Panel A and B but not for our industry specific regressions. Again we can observe one negative significant finding for average boards’ total compensation in Panel B with MRET12m3. This again shows the importance to consider CEO changes when analysing the relationship of stewardship and valuation. -Insert Tables 13, 14 and 15 here-

5

Conclusion

In this paper, we empirically investigate the association between the stewardship and valuation roles of financial reporting. Observing aggregate valuation and stewardship coefficients for industries and firms with all firms using IFRS, we find that stewardship and valuation relevance of earnings are significantly positively related applying CEOs’ cash and total compensation data from the period 2006 to 2013 based on industry and firm specific regressions and on three different market adjusted stock return measures. For CEOs’ cash compensation, our firm specific findings are robust to a different panel that includes years with a CEO change and years with no information about a CEO change. Furthermore, we can confirm our positive findings for CEOs’ and boards’ cash compensation in firm specific regression models that additionally control for ownership, 16

growth opportunities, risk and firm size. With average boards’ cash compensation we can find support on firm level only. To sum up our findings, we find generally supportive evidence for a positive association between stewardship and valuation relevance of earnings for a sample of firms from the German setting. Extending the scarce empirical research literature on the current debate about the relationship between stewardship and valuation, our results show evidence for firms preparing their financial statements according to IFRS, as one of the first studies analysing this relationship outside the US. Our study provides support for prior results that were based on US data (Bushman et al., 2006; Banker et al., 2009). We note that our data not only extends current research by focusing on a different geographical area but also considers compensation of the whole management board and is not limited to CEO compensation data. At least in general, this should enable a more comprehensive view on the relation between stewardship and valuation relevance. Moreover, in a multivariate setting we include firm and governance variables to control for context in our analysis of the relationship. As we show that these factors do not generally alter the relationship, this provides further evidence on a positive relationship between stewardship and valuation in our sample.

References Banker, Huang, and Natarajan. Incentive Contracting and Value Relevance of Earnings and Cash Flows. Journal of Accounting Research, 47(3):647–678, 2009. Barth, Beaver, Hand, and Landsman. Accruals, cash flows, and equity values. Review of Accounting Studies, 4:205–229, 1999. Br¨ uggemann, Hitz, and Sellhorn. Intended and Unintended Consequences of Mandatory IFRS Adoption: A Review of Extant Evidence and Suggestions for Future Research. European Accounting Review, 22(1):1–37, 2013. Bushman and Smith. Financial accounting information and corporate governance. Journal of Accounting & Economics, 32:237–333, 2001. Bushman, Engel, and Smith. An Analysis of the Relation between the Stewardship and Valuation Roles of Earnings. Journal of Accounting Research, 44(1):53–83, 2006. Cascino, Osma, Gassen, Imam, and Jeanjean. Who Uses Financial Reports and for What Purpose? Evidence from Capital Providers. Accounting in Europe, 11(2):185–209, 2014. Choi, Peasnell, and Toniato. Has the IASB Been Successful in Making Accounting

17

Earnings More Useful for Prediction and Valuation? UK Evidence. Journal of Business Finance & Accounting, 40(7-8):741–768, 2013. Conyon and Schwalbach. Executive compensation: Evidence from the UK and Germany. Long Range Planning, 33:504–26, 2000. Core, Guay, and Verrecchia. Are Performance Measures Other than Price Important to CEO incentives? The Accounting Review, 78:957–981, 2003. Devalle, Onali, and Magarini. Assessing the value relevance of accounting data after the introduction of IFRS in Europe. Journal of International Financial Management & Accounting, 21(2):85–119, 2010. Engel, Hayes, and Wang. CEO Turnover and Properties of Accounting Information. Journal of Accounting and Economics, 36:197–226, 2003. Engelen. The effects of managerial discretion on moral hazard related behavior: German evidence on agency costs. Journal of Management & Governance, 19:927–60, 2015. Gassen. Are Stewardship and Valuation Usefulness Compatible or Alternative Objectives of Financial Accounting? Working Paper, Humboldt University Berlin, 2008. Gjesdal. Accounting for stewardship. Journal of Accounting Research, 19:208–231, 1981. Heinle and Hofmann. Soft-Information and the Stewardship Value of Accounting Disclosure. Operations Research Spectrum, 33(2):333–358, 2011. Hitz and Werner. Why do Firms Resist Individualized Disclosure of Management Remuneration? Available at SSRN: http://ssrn.com/abstract=1588186, 2012. IASB. The Conceptual Framework for Financial Reporting-ED/2015/3. IFRS Foundation, Oct. 2015. London. Kaserer and Moldenhauer. Insider ownership and corporate performance: evidence from Germany. Review of Managerial Science, 2(1):1–35, 2008. Kuhner and Pelger. On the Relationship of Stewardship and Valuation — An Analytical Viewpoint. Abacus: A journal of accounting, finance and business studies, 51(3): 379–411, 2015. Lambert. Contracting Theory and Accounting. Journal of Accounting and Economics, 32:3–87, 2001. O’Connell. Reflections on stewardship reporting. Accounting Horizons, 21:215–227, 2007.

18

Ozkan, Singer, and You. Mandatory IFRS Adoption and the Contractual Usefulness of Accounting Information in Executive Compensation. Journal of Accounting Research, 50(4):1077–1107, 2012. Paul. On the Efficiency of Stock-Based Compensation. Review of Financial Studies, 5: 471–502, 1992. Pelger. Practices of standard-setting – An analysis of the IASB’s and FASB’s process of identifying the objective of financial reporting. Accounting, Organizations and Society, 50:51–73, 2016. Perotti and Wagenhofer. Earnings Quality Measures and Excess Returns. Journal of Business Finance and Accounting, 41(5-6):545–71, 2014. Rapp and Wolff. Determinanten der Vorstandsverg¨ utung [Determinants of executive pay]. Zeitschrift f¨ ur Betriebswirtschaft, 80:1075–1112, 2010. Voulgaris, Stathopoulos, and Walker. IFRS and the Use of Accounting-Based Performance Measures in Executive Pay. International Journal of Accounting, 49(4):479–514, 2014. Walker. Accounting for Varieties of Capitalism: The Case against a Single Set of Global Accounting Standards. British Accounting Review, 42:137–142, 2010. Watts. Conservatism in accounting, Part I: Explanations and implications. Accounting Horizons, 17:207–221, 2003. Watts. What has the Invisible Hand Achieved? Accounting & Business Research, Special Issue, 36:51–61, 2006. Whittington. Harmonisation or Discord? The Critical Role of the IASB Conceptual Framework Review. Journal of Accounting and Public Policy, 27:495–502, 2008. Zeff. The Objectives of Financial Reporting: A Historical Survey and Analysis. Accounting and Business Research, 43(4):262–327, 2013.

19

A

Tables Table 1 Construction of the sample

Sample selection DAX, MDAX, TecDAX and SDAX firms during the fiscal year 2006 and 2013 Less foreign firm year observations Less firm year obs. that are eliminated due to missing/incomplete compensation data Less firm year obs. applying US-GAAP Less firm year obs. with lack of data availability Less firm year obs. with missing lagged board compensation data Panel B: Total firm year obs. with available change in average board compensation Less firm year obs. with missing lagged CEO compensation data Panel B: Total firm year obs. with available change in CEO compensation Panel B: Total firm year obs. with available change in average board compensation Less firm year obs. with a CEO replacement or with no information about a CEO replacement Panel A: Total firm year obs. with available change in average board compensation Panel B: Total firm year obs. with available change in CEO compensation Less firm year obs. with a CEO replacement or with no information about a CEO replacement Panel A: Total firm year obs. with available change in CEO compensation

1280 80 49 30 91 186 844 146 698 844 186 658 698 83 615

Table 2 Industry membership (SIC codes-9 different SIC-groups)

One-digit SIC code 01 15,16 20,23,24,25,27 28,29 30,32 33,34,35 36 37,38 42,45 47,48,49 50,51,52,53,59 60,61,62 63,64 65,67 73,75,79,80,87 99

Description of industry code Agriculture Construction Manufacturing - Products (Food, Apparel, Furniture etc.) Manufacturing - Chemicals, Petroleum Refining Manufacturing - Plastic Products, Stone etc. Manufacturing - Metal, Machinery Manufacturing - Electronic Manufacturing - Transportation, measuring equipment Transportation - Motor freight, by air Transportation - Services, Communications Trade - Wholesale and Retail Finance Insurance Real Estate Services Public Administration Total

20

Frequency 7 14 38 111 40 149 37 70 21 73 61 56 20 51 80 19 847

Table 3 Data description

Variable MCAP MRET-15m MRET-12m MRET-12m3 LEV Firm size MTB BVPS NICON RET-12m SIC FIX-BOARD FB-BOARD STI-BOARD LTI-BOARD FIX-CEO FB-CEO STI-CEO LTI-CEO NOBM Share-Man Share-Fam Share-Bank Share-Strat Share-Inst Share-Other Free Float CC-CEO TC-CEO CC-AVG TC-AVG EARN

Description The natural logarithm of market capitalization Market-adjusted RET 15-months Market-adjusted RET 12-months Market-adjusted RET 12-months (3-months later shifted) Leverage (Total debt to common equity) Logarithm of total assets Market to book ratio Book value per share Net income (loss) - consolidated Stock return (Change in Total Return Index) 12-months Standard Industry Classification code Fixed salary board (in thousand) Fringe benefits board (in thousand) Short term incentives board (in thousand) Long term incentives board (in thousand) Fixed salary CEO (in thousand) Fringe benefits CEO (in thousand) Short term incentives CEO (in thousand) Long term incentives CEO (in thousand) Number of board members Percentage of shares managers are holding of total firm shares Percentage of shares family is holding of total firm shares Percentage of shares banks are holding of total firm shares Percentage of shares strategic investors are holding of total firm shares Percentage of shares institutional investors are holding of total firm shares Percentage of shares other investors are holding of total firm shares Percentage of shares holding in free float of total firm shares Percentage change in CEOs’ cash compensation Percentage change in CEOs’ total compensation Percentage change in average boards’ cash compensation Percentage change in average boards’ total compensation Difference in NICON t and t-1 deflated by MCAP of year t-1

21

Source Datastream Datastream Datastream Datastream Datastream Datastream Datastream Datastream Compustat Datastream Datastream hand-collected hand-collected hand-collected hand-collected hand-collected hand-collected hand-collected hand-collected hand-collected hand-collected hand-collected hand-collected hand-collected hand-collected hand-collected hand-collected calculated calculated calculated calculated calculated

Table 4 Summary statistics Variable Industry specific CC-CEO TC-CEO CC-AVG TC-AVG EARN MRET-12m MRET-15m MRET-12m3 RET-12m Free Float Share-Insider Share-Outsider Share-Manager Share-Family Share-Bank Share-Strategic Share-Institutional Share-Other Leverage Firm size Market to book value Firm-specific CC-CEO CT-CEO CC-AVG CT-AVG EARN MRET-12m MRET-15m MRET-12m3 RET12m Free Float Share-Insider Share-Outsider Share-Manager Share-Family Share-Bank Share-Strategic Share-Institutional Share-Other Leverage Firm size Market to book value

Obs

Mean

Std. Dev.

Min

Max

15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15

0.02 0.04 0.01 0.04 4.76e-06 -0.06 -0.04 -0.04 0.08 0.54 0.02 0.19 0.001 0 0.02 0 0.08 0.02 1.00 8.29 1.63

0.05 0.04 0.04 0.03 8.36e-06 0.10 0.10 0.08 0.10 0.12 0.05 0.07 0.004 0 0.10 0 0.03 0.08 1.14 1.62 0.41

-0.09 -0.05 -0.06 -0.03 -0.00001 -0.30 -0.30 -0.19 -0.16 0.41 0 0.10 0 0 0 0 0.03 0 0.15 6.85 0.87

0.11 0.11 0.08 0.10 0.00002 0.06 0.14 0.07 0.20 0.78 0.12 0.36 0.02 0 0.04 0 0.13 0.31 4.76 12.32 2.22

144 144 159 159 159 159 159 159 159 183 182 182 182 182 182 182 181 182 157 183 179

0.02 0.06 0.04 0.06 -9.92e-06 -0.02 -0.03 -0.05 0.12 0.51 0.16 0.23 0.08 0.08 0.06 0.01 0.10 0.06 1.47 7.61 2.18

0.15 0.16 0.22 0.18 0.0001 0.30 0.35 0.32 0.30 0.21 0.23 0.20 0.18 0.17 0.17 0.06 0.11 0.13 4.19 2.08 1.78

-0.51 -0.51 -0.46 -0.49 -0.0004 -0.93 -1.04 -0.99 -0.79 0.06 0 0 0 0 0 0 0 0 -1.81 4.11 -2.79

0.65 0.65 2.06 1.05 0.0001 1.30 1.55 1.44 1.44 1 0.94 0.89 0.94 0.71 0.89 0.80 0.61 0.71 37.56 14.51 10.49

CC (TC) defines the %-change in cash (total) compensation between year t and t-1. AVG is the average board compensation. MRET names the market-adjusted stock return, while RET names the stock returns. MRET is different for Panel A and B, because the market differs. This table provides an overview for Panel B. Free float, leverage, firm size, market to book value and all share variables are medians of i’s industry/firm year observations.

22

Table 5 Pearson correlation table Panel B-Industry specific EARN MRET-12m RET12m CC-CEO CC-AVG Panel B-Firm specific

EARN 1.0000 0.3107*** 0.3107*** 0.4767*** 0.3654***

MRET-12m

RET-12m

CC-CEO

CC-AVG

1.0000 1.0000*** 0.1082*** 0.0018

1.0000 0.1082*** 0.0018

1.0000 0.5184*** 1.0000

EARN MRET-12m RET-12m CC-CEO CC-AVG EARN 1.0000 MRET-12m 0.2886*** 1.0000 RET-12m 0.2886*** 1.0000*** 1.0000 CC-CEO 0.1972*** -0.0151 -0.0151 1.0000 CC-AVG 0.2295*** 0.1144*** 0.1144*** 0.6790*** 1.0000 The table shows correlations. Panel A describes the data that excludes firm years with a CEO change, while Panel B includes these years. This table uses Panel B. The correlation for Panel B are in line with Panel A. This table distinguishes between industry and firm specific median variables. Significance Levels: *p< 0.05, ** p < 0.01, ***p < 0.001

Table 6 Effect of earnings on stock market return Dep. Var.: MRET MRET-12m MRET-15m MRET-12m3 EARN 550.5*** 625.9*** 339.7*** (3.99) (3.91) (2.62) Obs. 658 658 658 R-sq. 0.388 0.395 0.453 Adj. R-sq. 0.367 0.374 0.434 First regression: M RETi = α0 + V ECi · EARNi + α1 D(t) + εi Notes: The t-statistics are in parentheses. The sample period is 2006-2013. The Panel excludes obs. with a CEO change. All regressions control for industry (15) and year fixed effects, while the effects saved for the industry specific third regression only control for year fixed effects. Significance Levels: *p< 0.1, ** p < 0.05, ***p < 0.01

23

Table 7 Effect of earnings on compensation Dep. Var.:

CC CC CEO AVG EARN 516.4* 414.4** (1.76) (2.36) RET12m 0.501 0.131*** (1.42) (2.66) Obs. 615 658 R-sq. 0.087 0.078 Adj. R-sq. 0.052 0.045 Second regression: Ci = β0 + CECi · EARNi + β1 RETi + β2 D(t) + εi Notes: The t-statistics are in parentheses. The sample period is 2006-2013. The Panel excludes obs. with a CEO change. All regressions control for industry (15) and year fixed effects, while the effects saved for the industry specific third regression only control for year fixed effects. Significance Levels: *p< 0.1, ** p < 0.05, ***p < 0.01

Table 8 Summary statistics of the VECs and CECs Variable Obs Mean Std. Dev. Min Max Industry specific Valuation coeff. (MRET-12m) 15 727.72 630.82 -334.12 1720.92 Stewardship coeff. CASH-CEO 15 2718.32 5725.02 -771.26 22614.45 CASH-AVG 15 730.19 1255.47 -1135.39 3140.23 Firm specific Valuation coeff. (MRET-12m) 93 2946.29 19932.22 -22118.7 153119.7 Stewardship coeff. CASH-CEO 74 -1073.5 33947.7 -283921.8 28673.83 CASH-AVG 80 2166.89 6859.11 -35256.71 23230.37 Notes: The table shows a summary statistics of the estimated values for stewardship and valuation relevance of earnings (excluding obs. with a CEO change). At least six firm year observations estimate one firm specific coefficient. The abbreviation CASH stands for cash compensation.

24

Table 9 Pearson - Spearman rank - correlation Pearson Industry CEO 0.5810** (0.0231) AVG 0.1930 (0.4907)

Spearman rank Pearson Spearman rank Industry Firm Firm 0.8036*** 0.4316*** 0.0975 (0.0003) (0.0001) (0.4083) 0.1571 0.3494*** 0.1313 (0.5760) (0.0016) (0.2488)

Notes: The table shows correlations. The p-values are in parentheses. We use fixed time effects for industry specific regressions only, the 12-month market adjusted stock return and cash compensation. This table applies the earnings coefficients of Table 8. Significance Levels: *p< 0.1, ** p < 0.05, ***p < 0.01

Table 10 Industry and firm specific association between valuation and stewardship relevance (with cash compensation and MRET-12m) CEO AVG Industry Industry Valuation 5.272** 0.384 (2.57) (0.71) Obs. 15 15 R-sq. 0.338 0.037 Adj. R-sq. 0.287 -0.037

CEO AVG Firm Firm 2.777*** 0.451*** (4.06) (3.27) 74 79 0.186 0.122 0.175 0.111

All regressions estimate this equation: CECi = γ0 + V SC · V ECi + εi Notes: The t-statistics are in parentheses. The sample period is 2006-2013. These regressions do not control for time fixed effects. The coefficients used in these third regressions are estimated in regression one and two industry and firm specific controlling for time fixed effects in industry specific regressions only. Valuation earnings coefficient are estimated with MRET-12m in regression one. Significance Levels:*p< 0.1,** p < 0.05,***p < 0.01

25

Table 11 Firm specific association between valuation and stewardship relevance for MRET12m and cash compensation including controls (1) CEO 2.866*** (3.88) 22650.4 (1.17)

CEC for VEC Free float

(2) AVG 0.409*** (2.93) 6725.1* (1.87)

share-in share-out

(3) CEO 2.966*** (4.20)

(4) AVG 0.418*** (3.03)

–18080.4 (-0.78) -65068.3*** (-2.82)

-5658.3 (-1.31) -10711.2** (-2.49)

share-man share-fam share-strat share-bank share-inst share-other Firm size Leverage Market to book value Obs. R-sq. Adj. R-sq.

390.3 (0.17) 848.9 (0.35) 1340.1 (0.47) 68 0.219 0.156

-286.6 (-0.68) -88.52 (-0.19) 728.6 (1.36) 74 0.180 0.120

33.45 (0.02) 1472.3 (0.62) 11.41 (0.00) 68 0.297 0.228

-277.3 (-0.66) -12.01 (-0.03) 616.3 (1.12) 74 0.211 0.141

(5) CEO 2.314*** (3.48)

(6) AVG 0.383*** (2.68)

-604.7 (-0.02) -23689.9 (-1.00) -159136.2*** (-5.08) -9413.8 (-0.26) -15192.0 (-0.34) -21884.0 (-0.81) -644.8 (-0.30) 595.0 (0.24) -899.4 (-0.31) 68 0.474 0.382

4294.9 (0.60) -9083.3* (-1.83) -15565.4** (-2.54) -4945.6 (-0.63) -7377.3 (-0.77) -6172.4 (-1.10) -149.4 (-0.33) -272.4 (-0.51) 214.7 (0.36) 74 0.264 0.147

All regressions estimate this equation: CECi = γ0 + V SC · V ECi + γ1 F irmi + γ2 CorpGovi + εi Notes: The t-statistics are in parentheses. The sample period is 2006-2013. The Panel excludes obs. with a CEO change. We include several firm and ownership variables, where each of them indicates the median for firm i (calculated from i’s firm year observations). Firm variables included in these regressions are firm size, leverage and market to book value. Corporate governance controls are free float, percentage firm share of in- and outside investors and percentage firm share of different shareholder types. These shareholder types are manager (man), family (fam), strategic investors (strat), bank (bank), institutional investors (inst) and other investors (other). Furthermore, we do not control for year fixed effects. Significance Levels:*p< 0.1,** p < 0.05,***p < 0.01

26

Table 12 Industry specific association between valuation and stewardship relevance for MRET12m and cash compensation including controls (1) CEO 5.642* (2.09) -6500.5 (-0.42)

CEC for VEC Free float

(2) AVG 0.394 (0.82) 6497.3** (2.35)

share-in share-out

(3) CEO 5.761* (2.13)

(4) AVG -0.0399 (-0.07)

-17558.9 (-0.49) 9090.1 (0.38)

-3583.5 (-0.45) 1987.6 (0.37)

share-man share-fam share-strat share-bank share-inst share-other Firm size Leverage Market to book value Obs. R-sq. Adj. R-sq.

-260.2 (-0.20) 279.1 (0.17) 2399.9 (0.58) 15 0.404 0.073

237.0 (1.01) -55.07 (-0.19) 243.4 (0.33) 15 0.604 0.385

-574.6 (-0.49) 281.8 (0.18) 3841.9 (0.77) 15 0.424 -0.008

528.5* (2.01) -370.8 (-1.02) 739.3 (0.66) 15 0.391 -0.065

(5) CEO 7.021* (2.13)

(6) AVG 0.434 (0.81)

-52051.4 (-0.09) 0 (.) 0 (.) -49613.0 (-0.23) -53247.9 (-0.49) 7472.7 (0.30) -869.1 (-0.63) 1180.4 (0.58) 759.1 (0.11) 15 0.470 -0.238

-81984.0 (-0.89) 0 (.) 0 (.) -76538.4* (-2.17) 26115.8 (1.49) 6731.6 (1.65) 426.5 (1.89) -390.7 (-1.19) 1619.0 (1.45) 15 0.710 0.322

All regressions estimate this equation: CECi = γ0 + V SC · V ECi + γ1 F irmi + γ2 CorpGovi + εi Notes: The t-statistics are in parentheses. The sample period is 2006-2013. The Panel excludes obs. with a CEO change. We include several firm and ownership variables, where each of them indicates the median for industries i (calculated from i’s firm year observations). Firm variables included in these regressions are firm size, leverage and market to book value. Corporate governance controls are free float, percentage firm share of in- and outside investors and percentage firm share of different shareholder types. These shareholder types are manager (man), family (fam), strategic investors (strat), bank (bank), institutional investors (inst) and other investors (other). Furthermore, we do not control for year fixed effects. Significance Levels:*p< 0.1,** p < 0.05,***p < 0.01

27

Table 13 Industry and firm specific association between valuation and stewardship relevance (with MRET-12m and total compensation) CEO AVG Industry Industry Valuation 4.874** 0.700 (2.45) (1.50) Obs. 15 15 R-sq. 0.316 0.147 Adj. R-sq. 0.263 0.081

CEO AVG Firm Firm 2.956*** 0.369*** (4.10) (2.77) 74 79 0.189 0.091 0.178 0.079

All regressions estimate this equation: CECi = γ0 + V SC · V ECi + εi Notes: The t-statistics are in parentheses. The sample period is 2006-2013. These regressions do not control for time fixed effects. Valuation earnings coefficient are estimated with MRET-12m in regression one and regression two uses total compensation. Significance Levels:*p< 0.1,** p < 0.05,***p < 0.01

28

Table 14 Industry and firm specific association between valuation and stewardship relevance (with Panel B, MRET12m, cash and total compensation) Industry C CEO Valuation 0.771 (1.10) Obs. 15 R-sq. 0.085 Adj. R-sq. 0.014 Firm C CEO Valuation 0.628*** (3.79) Obs. 74 R-sq. 0.166 Adj. R-sq. 0.154

C AVG 0.951* (1.91) 15 0.220 0.160

T CEO 0.261 (0.50) 15 0.019 -0.057

T AVG 0.372 (0.66) 15 0.032 -0.042

C T T AVG CEO AVG 0.00850 0.754*** -0.239** (0.12) (4.07) (-3.04) 95 74 95 0.000 0.187 0.090 -0.011 0.176 0.080

All regressions estimate this equation: CECi = γ0 + V SC · V ECi + εi Notes: The t-statistics are in parentheses. The sample period is 2006-2013. These regressions do not control for time fixed effects. Valuation earnings coefficient are estimated with MRET-12 in regression. C and T stand for cash and total compensation. These regression use Panel B that includes firm year observations with a CEO change and years where we have no information about a CEO change. Significance Levels:*p< 0.1,** p < 0.05,***p < 0.01

29

Table 15 Industry and firm specific association between valuation and stewardship relevance for MRET15m and MRET12m3 Industry Panel A

Valuation Obs. R-sq. Adj. R-sq. Panel B

Valuation Obs. R-sq. Adj. R-sq. Firm Panel A

Valuation Obs. R-sq. Adj. R-sq. Panel B

Valuation Obs. R-sq. Adj. R-sq.

C CEO 3.558** (2.61) 15 0.343 0.292

with MRET-15m C T AVG CEO -0.0989 3.433** (-0.27) (2.64) 15 15 0.006 0.350 -0.071 0.300

T AVG 0.438 (1.39) 15 0.129 0.062

C CEO 2.341* (1.80) 15 0.200 0.139

with MRET-12m3 C T AVG CEO -0.0896 2.267* (-0.28) (1.83) 15 15 0.006 0.206 -0.070 0.145

C CEO -0.0392 (-0.06) 15 0.000 -0.077

with MRET-15m C T AVG CEO 0.447 -0.277 (0.91) (-0.59) 15 15 0.060 0.026 -0.012 -0.049

T AVG 0.195 (0.38) 15 0.011 -0.065

C CEO -0.559 (-1.08) 15 0.082 0.011

with MRET-12m3 C T T AVG CEO AVG 0.181 -0.350 0.504 (0.44) (-0.92) (1.26) 15 15 15 0.015 0.061 0.108 -0.061 -0.011 0.040

C CEO 1.639*** (2.93) 73 0.108 0.095

with MRET-15m C T AVG CEO 0.263** 1.735*** (2.44) (2.94) 80 73 0.071 0.108 0.059 0.096

T AVG 0.220** (2.02) 80 0.050 0.037

C CEO 1.255** (2.04) 73 0.055 0.042

with MRET-12m3 C T AVG CEO 0.277** 1.349** (2.36) (2.08) 80 73 0.067 0.057 0.055 0.044

T AVG 0.307** (2.63) 80 0.081 0.069

C CEO 0.386*** (2.89) 74 0.104 0.091

with MRET-15m C T AVG CEO -0.00545 0.497*** (-0.10) (3.34) 95 74 0.000 0.134 -0.011 0.122

T AVG -0.200*** (-3.18) 95 0.098 0.088

C CEO 0.353** (2.61) 74 0.087 0.074

with MRET-12m3 C T AVG CEO 0.000918 0.418*** (0.02) (2.74) 95 74 0.000 0.095 -0.011 0.082

T AVG -0.207*** (-3.48) 95 0.115 0.106

T AVG 0.448 (1.70) 15 0.181 0.119

All regressions estimate this equation: CECi = γ0 + V SC · V ECi + εi Notes: The t-statistics are in parentheses. The sample period is 2006-2013. The abbreviations C and T stand for cash (C) and total (T) compensation. The left table part uses MRET-15m in regression one, while the right uses MRET-12m in regression one. Panel A excludes obs. with a CEO change, while Panel B includes them and years where we have no information about a CEO change. Significance Levels:*p< 0.1,** p < 0.05,***p < 0.01

30

B

Figures

Figure 1: Industry specific valuation-stewardship associations (with MRET12m)

Figure 2: Firm specific valuation-stewardship associations (with MRET12m)

31