Managerial entrenchment and matching leverage preferences

Transcription

1 Managerial entrenchment and matching leverage preferences Timo Korkeamäki, Eva Liljeblom, and Daniel Pasternack * January 4, 2016 ABSTRACT We study the effect of managerial entrenchment on CEOs tendency to imprint their personal leverage preferences upon the firms they manage. Using a unique data source that allows us to measure personal wealth and indebtness in great detail, we find a connection between CEOs personal leverage and that of their firms. The connection is driven by the CEOs with a longer tenure and the CEOs who serve in a dual role. The connection is significantly weaker for those CEOs who have a proportion of their personal wealth tied to the firm. Presence of blockholders also weakens the connection. * Corresponding author: TimoKorkeamaki, Hanken School of Economics, P.O. Box 479, Helsinki, Finland, Tel , timo.korkeamaki@hanken.fi 1

2 Managerial entrenchment and matching leverage preferences January 4, 2016 ABSTRACT We study the effect of managerial entrenchment on CEOs tendency to imprint their personal leverage preferences upon the firms they manage. Using a unique data source that allows us to measure personal wealth and indebtness in great detail, we find a connection between CEOs personal leverage and that of their firms. The connection is driven by the CEOs with a longer tenure and the CEOs who serve in a dual role. The connection is significantly weaker for those CEOs who have a proportion of their personal wealth tied to the firm. Presence of blockholders also weakens the connection.

3 1. Introduction Individual attributes of the CEO have been shown to influence corporate capital structure decisions (see e.g. Bertrand and Schoar 2003, and Frank and Goyal 2009b). 1 Besides demographic attributes, behavioral traits may also affect the connection between the CEO and the financial decisions of the firm. Recent studies have indeed suggested that the behavioral characteristics of the CEO play an important role e.g. for the firm s leverage decisions. Graham, et al. (2013) report survey evidence of a connection between optimistic CEOs and high leverage. Similarly, Ben-David, et al. (2013) find that CEOs who overestimate accuracy of their information regarding future prospects of the firm tend to use more financial leverage. Both findings in essence confirm Hackbarth s (2008) theoretical expectation that optimistic and overconfident managers use financial leverage more aggressively. Our unique data source allows us to measure both personal assets and liabilities in much greater depth than what is done in prior studies. We test for whether CEO entrenchment affects the connection between CEO s leverage and that of the firm s, and whether the effect can be mitigated by the CEO s stock ownership in the firm. We find strong support for more powerful and entrenched CEOs having a leverage level more similar to that of the firm. CEO stock ownership and presence of blockholders seem alleviate the connection (the potential agency problem). The results of our study are of interest for the debate on whether firm s seek CEOs with matching (risk) preferences, or whether the CEO is over time able to imprint his preferences on the decisions of the firm. If CEO personal traits affect firm behaviour, it is natural to expect that CEOs personal actions would align with corporate decision-making. Indeed, evidence is emerging 1 Bertrand and Schoar (2003), who document a connection between manager fixed effects and firm leverage. They suggest managers business education and age as potential factors behind their observation. Frank and Goyal (2009b) find further support for the importance of CEO fixed effects, but note that in their rich set of CEO observable characteristics, they fail to find consistent evidence regarding any particular CEO variable s effect on leverage. 1

4 on matching behaviour between CEOs personal actions and those done by the firms that they run. Chyz (2013) finds that tax evasion at the personal level of the manager is connected to tax sheltering by the firm. Cronqvist, et al. (2012) suggest that the CEO s personal attitude towards financial risk is an important determinant of corporate capital structure. They report a strong connection between CEOs personal borrowing choices in conjunction with their home purchases, and leverage used by the firms they manage. In another related study, Kallunki and Pyykkö (2013) find that prior personal defaults of CEOs of Finnish private firms increase the likelihood of financial distress for firms that they manage. Managerial entrenchment may have a direct effect on firm leverage. Berger, et al. (1997) point out that entrenched managers have more discretion over their firms capital structure, and suggest that entrenched managers avoid leverage for agency motives, such as to protect their human capital and to avoid committing free cash flow. On the other hand, Karolyi (2014) reports that CEOs with longer tenure have better access to debt financing, which may work against the Berger, et al. (1997) prediction. We follow Cronqvist, et al. (2012), and study a more subtle agency issue, that is whether discretion leads CEOs to follow their personal leverage preferences more closely in capital structure decisions regarding their firm. They note that the CEO s tendency to imprint his/her debt preferences upon the firm results in potential agency problems, as the corporate capital structure is driven by CEO personal views instead of strictly value maximizing motives. Our main hypothesis is that CEOs who possess more decision-making power are more likely to exhibit matching between personal leverage and firm leverage. Our choices of proxies for CEO discretion follow prior literature. A longer CEO tenure increases the CEO s ability to shape the corporate management structure to enhance his/her power and entrenchment (Chava, et al., 2010). Hermalin and Weisbach (1998) suggest that CEOs with

5 longer tenure can affect board composition, and thus the board may become less independent as CEO tenure increases. Our second main measure of CEO entrenchment is CEO duality, where the CEO is also serving as the chairman of the board. Jensen (1993) points out the potential agency issues arising from CEO duality. We argue that CEOs with long tenure and/or dual role are more likely to imprint their personal preferences upon the firm s capital structure. We also study whether CEOs stock ownership and presence of blockholders have effects on the tendency to match preferences. We expect agency problems to be less severe both when CEOs have a higher proportion of their wealth invested in the firm, and when blockowners monitor the firm. In such cases, the CEOs are more likely to make capital structure decisions where their first priority is maximization of firm value. In a more general sense, we join prior literature in arguing that CEO personal characteristics matter in corporate capital structure decisions. In a recent paper, Fee, et al. (2013) question the above-mentioned findings by Bertrand and Schoar (2003), and urge researchers to pay more attention to firm attributes than to CEO attributes. Fee, et a. (2013) find that CEO changes fail to affect corporate financing and investment policies within the three years following the change. Our arguments are not in conflict with their findings, as we posit that CEOs need to gain power within the organization before they can imprint their own preferences upon the firm. We use a unique Finnish dataset that allows us to measure CEO personal leverage more precisely than what is done in prior literature. As we mention above, Cronqvist, et al. (2012) employ US data and consider the CEO s most recent home purchase. However, using mortgage borrowing upon a new home purchase as a proxy for personal leverage may contain a bias. Since interest payments on mortgages are tax-deductible, home buyers may have an incentive to borrow beyond their true leverage preference, for example to pay off their other

6 loans. Also, home purchases occur infrequently, which introduces challenges in measurement of leverage decisions consistently in time between personal leverage and firm leverage. Our data come from the Finnish Tax Authority, and it includes the total indebtness of the CEOs in our sample. We also obtain detailed records on our sample CEOs assets, which allows us to estimate their personal leverage in a way that is consistent with the typical measurement of corporate leverage. We find, consistent with Cronqvist, et al. (2012), that both book leverage and market leverage are connected to personal leverage in our dataset. In line with our expectations, we find a strong connection between agency issues and the matching relationship between personal leverage and firm leverage. Firms that have a CEO with a long tenure are more likely to exhibit matching between firm leverage and that of the CEO. CEOs with a dual role are also significantly more likely to show matching behavior. Furthermore, the matching relationship is significantly weaker for the CEOs who have a higher proportion of their personal wealth tied to the firm. Finally, we find that the matching relationship is weaker in firms with concentrated ownership. Our results thus support the notion that matching of personal leverage and firm leverage is an outcome of managerial entrenchment, and the issue is alleviated by stock ownership by the CEO and by presence of blockholders. Endogenous matching is an issue commonly addressed in the literature that studies the connections between CEO characteristics and firm characteristics. The direction of causality can be questioned, as a correlation between CEO personal traits and firm actions could also be explained by firms hiring managers whose personal characteristics match with those of the firm. The concern that endogenous matching is behind our results is alleviated by two factors. First, our results indicate that matching between CEO leverage and firm leverage occurs primarily in firms where the CEO has been in place for more than three years. This suggests that firms do not specifically seek CEOs with leverage preferences that match with the

7 current capital structure of the firm. Secondly, as Finland is a small and distant country with a peculiar language, the local market for individuals with CEO capabilities is likely to be limited. 2 This should constrain firms ability to endogenously match firm characteristics with those of a new CEO. The rest of the paper is structured as follows. Section 2 presents our data sources, along with descriptive statistics for the data items used in our econometric analysis. Section 3 describes our main results, and Section 4 concludes. 2. Data and descriptive statistics Our data source for CEO wealth and indebtness is the Finnish tax return data, which we have obtained from the Finnish tax authority for CEOs of all companies listed on the Helsinki Stock Exchange between 2002 and The Finnish tax authority kept complete wealth records until the wealth tax was abolished in Finland in As Holmen, et al. (2007) note, the public availability of the Finnish tax data resembles the Swedish system, and its origins date back to 1766, when Finland was part of Sweden. In the wealth taxation, the tax authorities used so-called taxation values for wealth items. These values were, however, converging towards actual market values at the time of our study period. We take the taxation values as given for real estate and other wealth such as forest ownership, but recalculate actual market value for equities, based on the exact information of stock portfolio composition for each CEO at each year-end. The lack of market value data on specific real estate items introduces some bias in our personal leverage estimates, as individuals with a 2 Yonker (2014) shows that in the US, the potential pool for CEOs grows with population density and sunny environment. Such factors may further limit the size of the CEO job market in Finland. 3 Finnish tax data can be considered very reliable. Firstly, tax records are public in Finland, and they generate a lot of public scrutiny and media attention. For reputation concerns, CEOs are thus likely to report truthfully. Secondly, Finns are characterized as being highly ethical, and they tend to conform to rules. In the 2007 Trusted Brand Survey by Reader s Digest, 72% of the Finns viewed themselves as conformists, whereas the global average was 19%. Transparency International s Corruption Perceptions Index ranked Finland in top 2 in the world during each of our sample years.

8 high proportion of real estate are likely to have an overestimated leverage. 4 We also obtain executive option data come from Alexander Corporate Finance Oy, a dominant consultant of companies in Finland regarding ESOPs. For financial statement data, we rely on Worldscope. The variables used in our analysis are defined in Appendix A. Since we study capital structure, we exclude financial firms from our sample. The number of non-financial firms listed on the Helsinki Stock Exchange during our sample years is 124 (2002), 118 (2003), 117 (2004) and 113 (2005), which sums up to 472 potential data points. After we exclude a small number of firms with either missing observations or book leverage greater than one, we are left with a final full sample consists of 431 firm-year observations for 125 different firms. 5 Descriptive statistics for various measures of CEO wealth and wealth composition, along with other firm characteristics, are reported in Table 1. The table provides information regarding CEO characteristics in Panel A, and firm characteristics in Panel B. Insert Table 1 We can note from Table 1 that Finnish CEOs are not particularly wealthy, with the median (gross) wealth of about 263,000 euros. It is also interesting to note that the median real estate holdings are zero, which suggests that many CEOs rent rather than own their residence. Their median age is 50, which is slightly lower than the US median of 55 among S&P 1500 firms (this and the following US comparisons come from Custodio and Metzger, 2014). The median tenure of four years compares to the US median of five years. About 21% of the CEOs of Finnish publicly traded firms have a Master s degree in business, whereas 4 However, including a control for the proportion of real estate only strengthens our reported results. 5 To avoid further deterioration, we use a substitution method to replace missing data points for firm characteristics that we use as control variables. In the substitution method we replace a missing data point by the cross sectional median for the data item in that year. Despite the resulting smaller sample size, our results are virtually identical when firms with any missing variables are dropped from the regressions.

9 37% of the US CEOs have an MBA. About 64% of our sample CEOs live in the Helsinki metropolitan area. Finally, serving a dual role of CEO and Chairman is not as common in Finland as it is in the US. The Dual Role indicator receives the value of one for 37% of our Finnish sample. Custodio and Metzger (2014) report that in their sample of S&P1500 firms in years , 60.1% of the CEOs are also Chairmen of the Board. Panel A of Table 2 provides further information on our main variables of interest, namely personal leverage of the CEOs, and both book-, and market leverage of the firms in our sample, by year. Table 2 indicates that personal leverage of the Finnish CEOs varies quite a bit from year to year during our four-year sample period. Market leverage shows a decreasing trend. Given the observed variation through time, we control for year fixed effects in all of our regressions in Section 3. Insert Table 2 In Panel B of Table 2, we segregate our leverage measures by our two main measures of managerial entrenchment, CEO duality and CEO tenure, and by whether the CEO s stock portfolio contains shares of the firm that she manages. The High tenure = 1 column includes firm-years when CEO s tenure is greater than three years. As we mention above, Berger, et al. (1997) expect entrenchment to be connected to lower firm leverage. However, our results in Panel B of Table 2 are inconsistent with their expectations. 6 Firms with either CEO duality or a CEO with high tenure are slightly more levered than firms with no CEO duality, which goes against Berger, et al. (1997) prediction. Personal leverage is significantly higher in firms with short CEO tenure, and firms where the CEO does not own any of the firm s stock. We consider the connection between personal leverage and firm leverage in the next section. 6 Recall Karolyi (2014) result of a positive correlation between CEO tenure and access to debt financing. It is possible that the effect of ease of access counteracts the entrenchment effect, and thus explains why there is no difference in leverage between the high tenure and the low tenure group.

10 3. Regression results In this section, we test the effect of personal leverage on firm leverage, and whether managerial entrenchment influences the relationship between the two variables. 3.1 Connection between firm leverage and CEO personal leverage We begin by observing the connection between personal leverage and firm leverage in a multivariate setting. We employ empirical models similar to those in previous studies. Our baseline model is defined as in equation (1). Firm lev α α Pers lev α Firm controls α Personal controls α Year effects ε, (1) wherefirm lev is either book leverage or market leverage, depending on the specification. Our firm controls are motivated by prior studies on capital structure determinants (see e.g. Rajan and Zingales 1995, and Frank and Goyal 2009a). Ln(assets) controls for effects of firms size on leverage, ROA controls for firm profitability, and M/B (Market-to-Book) controls for the presence of growth opportunities. Tangibility (property, plant and equipment over total assets) is set to capture tangible assets that may be used as collateral. We also include Diversified, which is an indicator variable for firms with more than one four-digit SIC code to control for potential effects of industrial diversification on leverage, and Concentr. Own, which is the percentage of the firm s equity owned by the five largest owners. The latter will control for the effect of block holders as monitors. Our Personal controls include TotalWealth, WealthOwnFirm (percentage of assets invested in own firm s stock), CEO_Age, CEO_Tenure, Ln(optionvalue), Helsinki_region, and

11 Business_degree. Of the personal controls, age and business degree are motivated by Bertrand and Schoar (2003). Tenure captures the effect of managerial entrenchment, as discussed above. TotalWealth, WealthOwnFirm, and Ln(optionvalue) control for the effects of the CEO s personal wealth and its composition. 7 We control for CEOs residing in the Helsinki region, as the Helsinki metropolitan area has housing market characteristics that differ significantly from the rest of the country. In particular, Robb and Robinson (2014) note that elasticity of housing supply affects its usability as collateral, and Oikarinen (2015) reports that in Finland, cities in the Helsinki region differ in that respect from more rural cities. 8 Following Cronqvist, et al. (2012), we report OLS results, while Tobit regressions produce very similar inferences. In Table 3, we report regression results where the dependent variable is market leverage, defined as total debt/(total assets book value of equity + market capitalization). In the first column of Table 3 we use Personal_leverage, the year dummies, and the constant as the only independent variables. The results confirm a statistically strong positive relation between personal leverage of the CEO and the firm leverage. In column (2), we add firm accounting controls and the dummy variable for the Helsinki region. The specification in column (2) is similar to the base line model in Cronqvist, et al. (2012), with the exceptions that we include the Helsinki area dummy, and we do not control for median industry leverage due to the small number of publicly-traded firms in the Finnish market. Our control variables enter with same signs as what Cronqvist, et al. (2012) report in their Table 3, Panel A, column (4). In contrast to their findings, the coefficients for market to book, and profitability fail to reach statistical significance at conventional levels, whereas 7 Using Total Income instead of Total Wealth yields identical inferences. 8 It is possible that local elasticity of residential housing supply affects both individuals and firms in the area (Cvijanovic, 2015). The Helsinki region dummy is set to capture any systematic effects that CEOs residing in the area would exhibit in the relation between their personal leverage and firm leverage.

12 wefind tangibility to have a very significant effect on leverage. In Cronqvist, et al. (2012) baseline model, it enters with an insignificant coefficient. Our main variable of interest, Personal_leverage, remains positive and significant at the 5% level. The dummy for the Helsinki region enters with a negative coefficient, indicating that firms with CEOs residing in the Helsinki area have lower leverage. Industry fixed effects for the 12 industry classes used by the Helsinki stock exchange weaken our evidence on Personal leverage in column (3), but the coefficient continues to be statistically significant at the 10% level. Notably Tangibility continues to have exhibit a strong positive coefficient, so the result regarding that variable in column (2) is not only due to the industrial structure of Finland. When we add controls for diversified firms, and concentrated ownership in column (4), the coefficient on Personal_leverage gains strength again. The coefficient strengthens further when we add controls for personal characteristics of the CEO in column (5). It is interesting to note that Personal_leverage is the only CEO characteristic to affect firm leverage. Namely, none of the CEO personal controls enters with a statistically significant coefficient. This is consistent with US findings by Frank and Goyal (2009b). Also, while Serfling (2014) finds that younger CEOs take more risk in terms of R&D intensity, operating leverage, and diversification of operations, our findings are consistent with his result that CEO age is not connected to leverage. Furthermore, if CEOs with a longer tenure are more entrenched, we fail to support Berger, et al. (1997) expectation that entrenched managers use lower leverage to safeguard their entrenched position in the firm. Finally, Lewellen (2006) provides a rationale for an inverse relation between manager s wealth tied to his/her own firm, and leverage. We fail to find support for that notion as well. Insert Table 3

13 In Table 4, we repeat the analysis of Table 3, now with book leverage as the dependent variable. The results are fairly similar to those reported in Table 3. Our test variable Personal_ leverage exhibits statistically weaker coefficients when book leverage is the dependent variable. Also, the positive coefficients for Tangibility and Ln(assets)are weaker in Table 4, and ROA now exhibits a stronger negative effect on leverage. Insert Table 4 New firms may face constraints in their access to debt financing (Peterson and Rajan, 1994). This is a potential cause of bias in our results, as young high tech firms are well presented in our data set. At the same time, a number of Finnish listed firms are founded in the 1800s, so our sample exhibits a wide variation in firm age. However, in unreported results, we find that firm age affects neither book leverage nor market leverage of our sample firms, while its inclusion leaves our results throughout the paper intact. 3.2 The matching relationship and managerial entrenchment In this sub-section, we study whether managerial entrenchment affects the above-documented link between firm leverage and CEO s personal leverage. Our two main measures of managerial entrenchment are CEO tenure and CEO duality, which both are commonly used proxies of entrenchment (see e.g. Chava, et al., 2010).The two measures are interrelated, but not as highly as one might think. In our sample, about 78% of the CEOs in a dual role have tenure greater than three years. However, 46% of the CEOs with tenure above three years are not in dual role. Correlation between the two dummies is 37.85%.

14 We alter equation (1) by adding an interaction term between personal leverage and entrenchment proxies, as indicated in equation (2). (2) Our first entrenchment proxy is Dual_role, which takes the value of one for CEOs who serve in a dual role of the CEO and the Chairman. As mentioned above, it is much less common for Finnish firms to have the CEO in a dual role than for US firms. 9 Out of all our firm-year observations, 37% have Dual_role =1.It is likely that a CEO in a dual role has more control over firm s leverage, and we therefore expect a α 3 to be positive and significant. The results are reported in Table 5. Our main interest in Table 5 is on the interaction variable, where we interact Dual_role with Personal_leverage to observe whether the relationship between personal leverage and firm leverage is affected by CEO duality. We report various specifications, with either market leverage or book leverage as the dependent variable, as indicated in the table. Table 5 results suggest that the relationship between personal leverage and firm leverage is almost entirely driven by those firms that have a CEO serving in a dual role. Except for the case of book leverage in combination with both year, and industry effects (column six of Table 5), the interaction term PerslevXdual is positive and statistically significant, regardless of the set of controls that we use, and of whether we use market leverage or book leverage as the dependent variable. It should also be noted that Dual_role 9 This is partially explained by the Finnish self regulation system, which draws from the Corporate Governance Code for listed companies. During our sample period, it included a recommendation for board independence. Firms that deviate from the recommendations are expected to explain why. Board independence was later even more strongly emphasized in the new versions of the code from e.g and 2010, leading to a continued reduction in Chairman-CEO duality in Finnish listed firms.

15 enters with a positive sign in all specifications of Table 5. When we leave the interaction term out in unreported regressions, the effect of Dual_role becomes also highly statistically significant in all specifications. This finding conflicts Berger, et al. (1997) prediction that entrenched managers avoid leverage. Insert Table 5 Our second measure of managerial entrenchment is CEO tenure. Chava, et al. (2010) note that longer tenure allows the CEO to increase her power within the organization, both through organizational changes and manager-specific investments. In Panel A of Table 6, we interact CEO tenure with Personal_leverage. Columns (1) and (2) include our firm-, and personal controls. The interaction term receives a positive coefficient that is highly statistically significant, which indicates that the connection between personal leverage and firm leverage is more pronounced for CEOs with longer tenure. Tenure is obviously related to CEO age. While our personal controls include CEO age, we add an interaction also for CEO age and personal leverage in columns (3) and (4) of Table 6, Panel A. 10 As the results indicate, the evidence in columns (1) and (2) is not affected by CEO age. In columns (5) and (6) we employ the interaction with age, without the interaction with CEO tenure. Even in those regressions, the interaction term fails to reach statistical significance at conventional levels. We thus conclude that our findings on the 10 Age may also directly affect the CEO s risk appetite. As Serfling (2014) indicates, two opposite hypotheses are at work. On one hand, younger executives have greater career concerns (Holmström, 1999). On the other hand, younger CEOs may signal their quality by making more aggressive decisions (Prendergast and Stole, 1996).

16 stronger connection between personal leverage and firm leverage among CEOs with longer tenure are not driven by age of those CEOs. Insert Table 6 In Panel B of Table 6, we rerun our regression model from column (5) of Table 3 with sub-samples that are split on the basis of our entrenchment proxies. While we only report results of regressions with market leverage as the dependent variable in Panel B of Table 6, using book leverage yields very similar results. The first column of Panel B only includes firm years with CEO-Chairmen, and the second column includes firm-years with no dual role CEOs. As the coefficient on Personal_leverage indicates, the connection between personal leverage and firm leverage is only present in firms with CEO duality. In columns (3) and (4) of Panel B, we split our full sample by CEO tenure, so that in column (3), we include only firms with CEO tenure less than four years. Column (4) sample includes firms where CEO tenure is greater than three years. Again, the connection between firm leverage and personal leverage is present only in firms with entrenched CEO, in other words firms where the CEO has a longer tenure. CEO tenure of three years is slightly lower than the median of four years. However, we purposefully select the three-year cutoff to contrast our findings against some of the results reported by Fee, et al. (2013). They observe firms investment and financing policies in conjunction with changes in leadership. They fail to find changes in investment and financing patterns within three years following CEO changes, and conclude that CEO styles are secondary to firm effects in explaining corporate behavior. Our results in Table 6 suggest

17 a long lag between CEO changes and subsequent effects on firms financing patterns, and we argue that the lag is explained by the time required for the new CEO to gain a position where she can imprint her personal traits on the firm. This is also consistent with literature on capital structure adjustment speed. Graham and Leary (2011) summarize adjustment speed findings of various authors. According to their summary, firms annually close 10% to 30% of the gap between their current leverage and book leverage. Our Table 6 results are also relevant for the debate on the effects of CEO characteristics and the potential endogenous CEO-firm matching. If the main direction of causality between personal leverage and firm leverage is such that the CEOs determine their firms leverage and imprint their personal preferences on firms capital structures, then we expect CEOs with a longer tenure with the firm to exhibit a stronger relationship between personal leverage and firm leverage. The alternative hypothesis would build on endogenous matching, where firms seek CEOs whose personal characteristics match with those of the firm. Our Table 6 results indicate that indeed, the matching relationship is significantly stronger for CEOs with longer tenure, and personal leverage does not correlated with firm leverage for firms with recent CEO hires. In Table 7, Panel A we interact personal leverage with the proportion of CEO s wealth invested in the firm. If CEOs tendency to follow their personal preferences in corporate capital structure decision reflects an agency issue, this agency issue is likely to be alleviated by CEO s personal stakes in the firm. Also, Cronqvist, et al. (2012) hypothesize that CEOs may be affected by hedging considerations, so that they would counter-balance high firm leverage with low personal leverage, or vice versa. Such hedging considerations should be more important for CEOs who have a higher proportion of their wealth tied to the firm in form of equity ownership, which would speak against matching personal leverage

18 with firm leverage, assuming that such matching is sub-optimal from the purely economic stand point. Alderson, et al. (2014) find that CEO tenure and ownership in own firm are related, and their results suggest that lower stock ownership by CEOs with shorter tenure affects their appetite for risk. However, in our data set, CEO tenure appears to be relatively unrelated to CEO stock ownership. We find that 30.3% CEOs with tenure greater than three years have their own company s shares, in comparison to 29.8% for CEOs with shorter tenure. The proportion of total wealth invested in own shares is also indistinguishable between the two groups. We, again, present results of regressions with both market leverage and book leverage as dependent variables. Table 7 results indicate that the connection between personal leverage and firm leverage is considerably weaker for firms with higher CEO ownership. We thus find support for the notion that managerial stock ownership reduces the ill effects of managerial entrenchment. As Panel B of Table 2 indicated earlier, about 41% of the CEOs in our sample own shares in the company that they manage. When we include option holdings into the position in the own firm, or consider them separately, a picture that is different from Table 7 findings emerges (in un-tabulated results). Namely, option holdings increase the CEO s tendency to match firm leverage with personal leverage. Our interpretation for the difference in the effects between stock ownership and option positions is the following. Stock ownership creates a more stable bond between manager and the firm. The Finnish tax code treats income arising from option exercises as ordinary income, which leads the CEOs in our sample to face a tax liability of up to 60% on their option-related income (Liljeblom, et al., 2011). This liability is likely to force many of the CEOs into immediately selling a large proportion of the

19 shares that they receive upon option exercise, just in order to pay for the resulting income tax. We therefore view equity ownership as a more efficient form of aligning long-term motives for Finnish CEOs, and our evidence supports that view. Finally, in Panel B of Table 7, we consider concentrated ownership as an additional corporate governance factor that potentially alleviates problems related to managerial entrenchment. We define an indicator variable Block Own, that takes on the value of one if the five largest shareholders of the firm own more than 25% of the firm s equity, combined. We then interact personal leverage with the concentrated ownership proxy, similarly to what we do with our other agency-related proxies in Tables 5-7. The results in Panel B of Table 7 indicate that concentrated ownership also weakens the connection between personal leverage and firm leverage. This suggests that block holders perform a monitoring function, and thereby reduce the CEO s tendency to follow his personal leverage preferences in the firm s capital structure decisions. As noted earlier, Cronqvist, et al. (2012) rely on the CEO s most recent home purchase when measuring her personal leverage preferences. In unreported results, we confirm that in our Finnish sample also, the CEO s personal leverage in 2005 correlates positively with the firm s leverage in a panel that includes subsequent nine years, from 2006 to However, the correlation is present both among firms that retain the CEO in office in 2005, and among firms that change their CEO. In another set of unreported tests, we consider the Cain and McKeon (2014) suggestion that both use of leverage and type of M&A activity are spurred by a sensation seeking personal treat of the CEO. We observe acquisitions of more than 25% of the target firm s shares, resulting in higher than 50% ownership, and being greater than USD 5 million in value, that occur during tenure of the CEO who is in place in our sample firms in 2002.

20 158 acquisitions meet these criteria. We find that neither engaging in M&A activity nor making foreign acquisitions correlates with the level of personal leverage. Making diversifying acquisitions (outside own firm s primary 2-digit SIC code) is weakly related to lower personal leverage, but the regression coefficient of such indicator variable falls short of conventional levels of statistical significance. 4. Conclusions We study the effects of CEOs personal leverage on firm leverage. Under the assumption that the CEO s personal view on indebtness may deviate from the optimal policy for the firm, we view evidence of matching as a potential agency issue, similar to Cronqvist, et al. (2012). Our results are consistent with the agency view. We find that in the overall sample, a matching relationship exists between personal leverage and firm leverage, but that relationship is concentrated to firms where the CEO is more entrenched. Our proxies of entrenchment are CEO tenure and CEO duality. We further find that matching is less pertinent in firms where the CEO has a proportion of her wealth invested in the firm s equity. We contribute to the literature by a more accurate proxy for personal leverage. Our unique data from the Finnish tax authorities allow us to generate a balance sheet for each CEO in our sample. We also contribute to the debate on the importance of CEO personal characteristics in explaining firm behaviour. Our evidence suggests that CEO personal traits are reflected in firm behaviour only after the CEO has secured a power position within the firm.

21 References Alderson, M.J., N. Bansal, and B.L. Betker, 2014, CEO turnover and the reduction in price sensitivity, Journal of Corporate Finance 25, Ben-David, I., J. Graham, and C.R. Harvey, 2013, Managerial miscalibration, Quarterly Journal of Economics, forthcoming. Berger, P.G., E. Ofek, and D.L. Yermack,1997, Managerial entrenchment and capital structure decisions, Journal of Finance 52, Bertrand, M., and A. Schoar, 2003, Managing with style: The effect of managers on firm policies, Quarterly Journal of Economics 118, Cain, M.D., and S.B. McKeon, 2014, CEO personal risk-taking and corporate policies, Journal of Financial and Quantitative Analysis, forthcoming. Chava, S., P. Kumar, and A. Warga, 2010, Managerial agency and bond covenants, Review of Financial Studies 23, Chyz, J.A., 2013, Personally tax aggressive executives and corporate tax sheltering, Journal of Accounting and Economics 56, Cronqvist, H., A.K. Makhija, and S.E. Yonker, 2012, Behavioral consistency in corporate finance: CEO personal and corporate leverage, Journal of Financial Economics 103, Custodio, C., and D. Metzger, 2014, Financial expert CEOs: CEO s work experience and firm s financial policies, Journal of Financial Economics 114, Cvijanovic, D., 2015, Real estate prices and firm capital structure, Review of Financial Studies, forthcoming. Fee, C.E., C.J. Hadlock, and J.R. Pierce, 2013, Managers with and without style: Evidence using exogenous variation, Review of Financial Studies, forthcoming. Frank, M.Z., and V.K. Goyal, 2009a, Capital structure decisions: Which factors are really important?, Financial Management 38, Frank, M.Z., and V.K. Goyal, 2009b, Corporate leverage: How much do managers really matter?, unpublished working paper, University of Minnesota. Graham, J.R., C.R. Harvey, and M. Puri, 2013, Managerial attitudes and corporate actions, Journal of Financial Economics 109, Graham, J.R., and M.T. Leary, 2011, A review of empirical capital structure research and directions for the future, Annual Review of Financial Economics 3, Hackbarth, D., 2008, Managerial traits and capital structure decisions, Journal of Financial and Quantitative Analysis 43,

22 Hermalin, B.E., and M.S. Weisbach, 1998, Endogenously chosen boards of directors and their monitoring of the CEO, American Economic Review 88, Holmen, M., J.D. Knopf, and S. Peterson, 2007, Trading-off corporate control and personal diversification through capital structure and merger activity, Journal of Business & Accounting 34, Holmstrom, B., 1999, Managerial incentive problems: a dynamic perspective, Review of Economic Studies 66, Jensen, M.C., 1993, The modern industrial revolution, exit, and the failure of internal control systems, Journal of Finance 48, Kallunki, J.-P., and E. Pyykkö, 2013, Do defaulting CEOs and directors increase the likelihood of financial distress of the firm?,review of Accounting Studies 18, Karolyi, S.A., 2014, Personal lending relationships, Carnegie Mellon University working paper. Lewellen, K., 2006, Financing decisions when managers are risk averse, Journal of Financial Economics 82, Liljeblom, E., D. Pasternack, and M. Rosenberg, 2011, What determines stock option contract design?,journal of Financial Economics 102, Oikarinen, E., R. Peltola, and E. Valtonen, 2015, Regional variation in the elasticity of supply of housing, and its determinants: The case of a small sparsely populated country, Regional Science and Urban Economics 50, Peterson, M., and R. Rajan, The benefits of lending relationships: evidence from small business lending,,journal of Finance 49, Prendergast, C., and L. Stole, 1996, Impetuous youngsters and jaded old-timers: acquiring reputation for learning, Journal of Political Economy 104, Rajan, R.G., L. Zingales, 1995, What do we know about capital structure? Some evidence from international data, Journal of Finance 50, Robb, A.M., and D.T. Robinson, 2014, The capital structure decisions of new firms, Review of Financial Studies 27, Serfling, M.A., 2014, CEO age and the riskiness of corporate policies, Journal of Corporate Finance 25, Yonker, 2014, Geography and the market for CEOs, unpublished working paper, Indiana University.

23 Appendix A. Variable definitions for key variables The table describes our key variables for CEO (Panel A) and the firm (Panel B). Wealth and income data are from the Finnish tax authority; financial statement data, industry codes and stock return and price data are from Datastream. Firm ownership data are from Pörssiyhtiöt books and the Finnish Central Securities Depository; dividend and share repurchase data are from the Helsinki Stock Exchange, and ESOP data are from Alexander Corporate Finance Oy. All nominal variables ( values ) are denoted in euros. By own firm we refer to the firm employing the CEO. The time period is 2002 to 2005, and the sample covers 532 firm-years for 162 firms. Panel A. CEO characteristics CEO_Age The age of the CEO at year-end. CEO_Tenure Tenure with company in years, rounded to the closest full year. Dual_role An indicator value that takes the value of one for CEOs who also serve as the Chairman of the Board. TotalWealth The total taxable value of the CEO wealth (gross of debt). Equities are calculated at the market value at the year-end, while for other assets, their taxational values determined by the tax authorities have been used RealEstateValue The total value of real-estate holdings. Personal_leverage CEO s personal total debt divided bytotalwealth. WealthOwnFirm Value of gross holdings in the firm employing the CEOdivided bytotalwealth. Ln(optionvalue) Natural logarithm of the Black-Sholes-value of the CEOs ESOP portfolio at yearend, plus 1 euro. Business_degree An indicator variable that takes the value of one for CEOs with a Master s degree in business. Helsinki_ region An indicator variable that takes the value of one for CEOs whose residence according to tax authorities is in Helsinki, Espoo, Vantaa, or Kauniainen. Panel B. Firm characteristivs Ln(assets) Natural logarithm of Total_Assets. M/B Market value of equity/book value of equity. ROA EBIT/Total Assets. Book_ leverage Total debt / Total assets. Mkt_leverage Total debt / (Total debt + equity market capitalization). Tangibility Property, plant and equipment over total assets. Concentr. Own % of shares held by 5 largest shareholders. Block Own An indicator variable that takes the value of one in firm-years when Concentr. Own > 25% Diversified Anindicator for firms with more than 2 SIC-codes at the 2-digit level.

24 Table 1. Descriptive statistics This table reports descriptive statistics for key variables describing CEOs in Panel A, and firms characteristics in Panel B. See Appendix A for variable definitions. Panel A. CEO characteristics Mean Median Min Max Obs CEO_Age CEO_Tenure TotalWealth Wealthownfirm RealEstateTotal Personal_leverage Option_value Business degree Helsinki region Dual role Panel B. Firm characteristics Mean Median Min Max Total Assets Market_to_Book ROA Tangibility Debt_to_Assets Market Leverage Concentr. Own Diversified

25 Table 2.Personal leverage and firm leverage This table reports descriptive statistics for our measures of personal leverage and firm leverage, by year in Panel A. Panel B differences in means of the leverage variables, where the sub-samples are divided by whether the firm has a CEO in dual role or not, whether the CEO has a tenure greater than 3 years (= High tenure) or not, and whether the CEO has any wealth invested in the firm s stock. For variable definitions, see Appendix A. Panel A Year Variable Mean Median Max Min 2002 Personal_leverage Mkt_leverage Book_leverage Personal_leverage Mkt_leverage Book_leverage Personal_leverage Mkt_leverage Book_leverage Personal_leverage Mkt_leverage Book_leverage Panel B Dual role High tenure Wealth own firm Personal_leverage ** *** Book_leverage ** * Mkt_leverage * N

26 Table 3. Determinants of market leverage This table reports the results from regressing firm leverage (measured as market leverage) on CEO s personal characteristics and firm controls. For variable definitions, see Appendix A. Varibles significant at the 10%, 5%, and 1% level are denoted with *, **, and ***, respectively. (1) (2) (3) (4) (5) Variables Mkt_leverage Mkt_leverage Mkt_leverage Mkt_leverage Mkt_leverage Personal_leverage ** ** * *** *** (2.519) (2.558) (1.975) (2.624) (2.885) Helsinki_region *** *** *** *** (-3.057) (-2.782) (-3.183) (-3.405) Ln(assets) ** ** *** ** (2.370) (2.110) (3.064) (2.582) M/B * (-1.434) (-1.421) (-1.557) (-1.689) ROA *** (-1.271) (-1.245) (-1.216) (-7.728) Tangibility *** *** *** *** (5.530) (5.541) (5.276) (5.532) Diversified (0.619) (0.787) Concentr. Own ** ** (2.061) (2.130) TotalWealth (-0.332) WealthOwnFirm (-0.727) CEO_Age (1.202) CEO_Tenure (-0.542) Ln(optionvalue) (-0.516) Business_degree (1.195) Constant *** (9.794) (1.519) (0.647) (-0.739) (-1.299) Industry effects no no yes no no Year effects yes yes yes yes yes Observations Adj R

27 Table 4. Determinants of book leverage This table reports the results from regressing firm leverage (measured as book leverage) on CEO s personal characteristics and firm controls. For variable definitions, see Appendix A. Varibles significant at the 10%, 5%, and 1% level are denoted with *, **, and ***, respectively. (1) (2) (3) (4) (5) Variables Book_leverage Book_leverage Book_leverage Book_leverage Book_leverage Personal_leverage ** * ** ** (2.065) (1.962) (1.107) (2.162) (2.378) Helsinki_region *** *** *** *** (-2.839) (-2.863) (-3.037) (-3.547) Ln(assets) * (1.110) (1.135) (1.407) (1.803) M/B * (-1.301) (-1.252) (-1.414) (-1.720) ROA * * *** (-1.668) (-1.955) (-1.647) (-6.685) Tangibility ** * (1.631) (2.455) (1.477) (1.848) Diversified * (1.549) (1.660) Concentr. Own * * (1.731) (1.864) TotalWealth (-0.329) WealthOwnFirm (-1.461) CEO_Age (0.866) CEO_Tenure (-0.384) Ln(optionvalue) (-0.032) Business_degree ** (2.107) Constant *** ** (7.321) (2.034) (1.374) (1.402) (0.335) Industry effects no no yes no no Year effects yes yes yes yes yes Observations Adj R

28 Table 5. Corporate governance and matching preferences This table reports the results from regressing firm leverage (measured as market or book leverage) on CEO s personal financial characteristics, firm controls, and corporate governance variables (Dual_role, and its interaction with Personal_leverage, i.e. the variable PerslevXdual ). For variable definitions, see Appendix A. Varibles significant at the 10%, 5%, and 1% level are denoted with *, **, and ***, respectively. Mkt _leverage Book _leverage Mkt _leverage Book _leverage Mkt _leverage Book _leverage Variables Personal _leverage (0.754) (0.671) (0.571) (0.669) (0.446) (0.872) PerslevXdual *** ** *** *** *** (2.717) (1.981) (3.951) (2.657) (2.879) (1.349) Dual_role ** (1.344) (0.884) (2.071) (1.182) (1.010) (1.104) Firm controls Yes yes yes yes no no Personal controls No no yes yes no no Industry effects No no no no yes yes Year effects Yes yes yes yes yes yes Observations Adj R