Capital Structure Decision

Journal of Banking & Finance 25 (2001) 18571895 www.elsevier.
com/locate/econbase
An empirical analysis of incremental capital structure decisions under managerial entrenchment

Abe de Jong *, Chris Veld 1
Department of Finance, Tilburg University, P.O. Box 90153, NL-5000 LE Tilburg, Netherlands Received 1 February 1999; accepted 11 September 2000
Abstract We study incremental capital structure decisions of Dutch companies. From 1977 to 1996 these companies have made 110 issues of public and private seasoned equity and 137 public issues of straight debt. Managers of Dutch companies are entrenched. For this reason a discrepancy exists between managerial decisions and shareholder reactions. Conrming Zwiebel [American Economic Review (1996) 11971215] we nd that Dutch managers avoid the disciplining role of debt allowing them to overinvest. However, the market reactions show that this overinvestment behavior is recognized. We do not nd a conrmation of the adverse selection model of Myers and Majluf [Journal of Financial Economics (1984) 187221]. This is probably due to the entrenchment of managers and the prevalence of rights issues. 2001 Elsevier Science B.V. All rights reserved. JEL classication: G32
Corresponding author. Abe de Jong is an Assistant Professor at the Department of Finance and CentER at Tilburg University. Part of the research for this project was carried out when Abe de Jong was visiting Florida State University in Tallahassee. Tel.: +31-13-4662020; fax: +31-134662875. E-mail address: abe.dejong@kub.nl (A. de Jong). 1 Chris Veld is a Professor of Finance at the Department of Finance and CentER and a Professor of Personal Financial Planning at the Department of Fiscal Economics at Tilburg University. Part of the research for this project was carried out when Chris Veld was visiting the Faculty of Management at the McGill University in Montreal. 0378-4266/01/$ - see front matter 2001 Elsevier Science B.V. All rights reserved. PII: S 0 3 7 8 - 4 2 6 6 ( 0 0 ) 0 0 1 6 3 - 1
1858
A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895
Keywords: Capital structure; Managerial entrenchment; The Netherlands; Rights issues; Seasoned equity oerings; Announcement eects
1. Introduction A central question in the corporate nance literature relates to the choice of the rm between issuing equity or debt. The theoretical literature on this topic includes both static models, that date back to Modigliani and Miller (1958), and dynamic models. A common assumption in these models is that shareholders have the possibility to enforce managers to aim for shareholder value maximization. These theoretical models have been subject to empirical testing in a large number of studies. Some of these studies look at motives for the debtequity choice. Other studies investigate the wealth eects for this choice. Most of the empirical studies are done for the US. This is especially true for studies on the motives for the debtequity choice and for studies that both look at these motives and at the associated wealth eects. The assumption that shareholders can enforce shareholder value maximization is denitely true for the US. In this paper we study the debtequity choice for the Netherlands. This country is particularly interesting because shareholders in the Netherlands are hardly able to discipline managers. However, they are able to judge the decisions that managers make. This makes it interesting to compare the outcomes of studies on the motives for the debtequity choice and studies that investigate the wealth eects that are associated with this choice. In this paper we use data of Dutch rms to test the moral hazard models of Jensen (1986) and Zwiebel (1996) and to test the adverse selection model of Myers and Majluf (1984). In the model of Jensen (1986), moral hazard problems determine the choice for debt or equity. He explains that rms may engage in projects with negative NPV, because managers pursue growth. This overinvestment problem can be overcome by using debt instead of equity. In the model of Zwiebel (1996), managers only issue debt if they are forced by a discipliner. This discipliner is partially limited by managerial entrenchment. Managers trade-o empire building and the possibility to retain control. Empire building is constrained by a take-over and by the use of debt. The latter leads to an increase of the threat of a bankruptcy. Entrenched managers will avoid the disciplining role of debt. The adverse selection model of Myers and Majluf (1984) shows the impact of asymmetric information in case new investors are less informed about the value of the rm than existing shareholders. This gives rise to an underinvestment problem. The rm will not carry out a project with a positive NPV if the underpricing of the new capital, caused by the asymmetric information, will be higher than the value of the project. In order to deal with this problem the rm will use a pecking order of funds in
1859
which the least risky form of nancing is preferred. Entrenched managers without shareholdings are less inclined to transfer wealth from new investors to the current shareholders. Entrenchment may reduce the underinvestment problem. Bayless and Chaplinsky (1991) and Jung et al. (1996) have carried out studies on the motives for the debtequity choice using data for the US. The results of both studies are consistent with the adverse selection model. The results of Jung et al. (1996) are also consistent with the moral hazard model. This is also the case for a study by Berger et al. (1997) who study the relationship between leverage and corporate governance variables for US companies. A number of studies have investigated the wealth eects of security issues for the US. Review studies by Smith (1986) and Eckbo and Masulis (1995) present the following results. Common stock issues are associated with signicant negative abnormal returns. Issues of convertible bonds are also associated with negative abnormal returns, though the stock price decline is smaller than the stock price decline associated with the issuance of common stock. Issues of straight debt are not associated with an abnormal return. These results suggest a conrmation of the pecking order model. In this paper, we study the factors which induce Dutch companies to issue either stocks or bonds. A number of studies have shown that, unlike managers of American companies, Dutch managers are entrenched. Major Dutch companies all have a two-tier board system, consisting of an executive board and a supervisory board. The executive board coincides with the top-level executive board, while the supervisory board is completely composed of outside experts. In the Netherlands, the supervisory boards of large corporations are legally obliged to watch over the corporate enterprise as a whole, and not primarily or exclusively over the interests of shareholders or any other group of stakeholders (see Moerland, 1995). Even more important, Kabir et al. (1997) nd that practically all Dutch companies have adopted multiple takeover defenses. They argue that these measures are primarily directed to limit the power of common shareholders. La Porta et al. (1999) study shareholder rights in different countries. They use an index aggregating shareholder rights which is labeled as ànti-director rights'. Based on the calculation of this index for 27 countries, the authors conclude that the Netherlands belong to the group of low shareholder rights. Cools (1993), who interviewed the Chief Financial Ocers of 50 large Dutch rms, nds that for 38% of the rms, shareholders play no role at all or are ranked as the least important stakeholder. He contributes this to the Dutch institutional setting that has robbed the shareholder of Dutch corporations virtually all his voting power and control over the company (see Cools, 1993, p. 264). The conclusion that can be drawn from these studies is that Dutch managers enjoy a lot of freedom in taking nancing decisions. This leads to the objective of this paper, which is to study the security issue decision for the Netherlands. Empirical evidence for this country is particularly interesting because in the Netherlands shareholders are hardly able to
1860
discipline managers. The Dutch institutional setting makes it interesting to study the moral hazard and the adverse selection models. 2 The moral hazard model is interesting because managers' choices are likely to deviate from the interests of shareholders, but shareholders are still able to judge the decisions that managers make. This allows us to compare the outcomes of studies on motives for the debtequity choice with the wealth eects that are associated with this choice. The adverse selection model is interesting because, although it is often conrmed for the US, given the Dutch institutional setting with entrenched managers it is not expected to hold for the Netherlands. Finally, we also include control hypotheses from the static trade-o model. The reason for this is that this model has also received a lot of empirical conrmation in the corporate nance literature. According to the static trade-o theory rms have an optimal capital structure. Our sample of 110 issues of public and private equity by Dutch companies and 137 public issues of straight debt by Dutch companies gives the following results. We nd conrmation for the theory of Zwiebel (1996) that managers try to avoid debt. We also nd that rms with both positive free cash ow and low growth opportunities prefer to issue equity, and thus avoid debt. The results show that managers avoid debt when it has the largest disciplinary power, i.e., in case of low protability. We also use a model specication in which the movement towards the target debt/equity ratio is included. We nd a signicant positive eect for the coecient of the deviation from target. However, the explanatory power and the percentage of correct predictions are higher for specications that represent dynamic models. A study of the wealth eects of the security issues shows that the issuance of debt is associated with an insignicant abnormal return of 0.51% and that an equity issue is associated with a signicant negative abnormal return of )1.07%. These results are in accordance with the moral hazard and adverse selection models. A further analysis is carried out to distinguish between these models. This analysis shows that 19 Dutch companies announce the equity issue together with the news that the company has just completed an acquisition. These equity issues are associated with a signicant positive abnormal return of 2.08%. In 16 cases the issuing company indicates that it will use at least part of the funds for future, yet unspecied, acquisitions. These equity issues are associated with a signicant negative abnormal return of )2.72%. The dierence in announcement eects on an equity issue together with a completed acquisition and an equity issue to
2 It should be emphasized that in terms of tax and bankruptcy regulations Dutch companies are much like American companies. For example, both the Netherlands and the US have a system of double taxation. Taxes are levied on corporate prots before they are distributed to the shareholders. If these prots are distributed as cash dividends, they are taxed again as shareholder income. Like in the US depreciation is tax deductible. Also the consequences of bankruptcy are the same for Dutch managers as they are for American managers.
1861
be used for future acquisitions can be interpreted as additional evidence for the moral hazard model. A regression of the determinants of the returns on the abnormal returns leads to further evidence for the moral hazard model. Our study on motives for the debtequity choice does not render evidence for the adverse selection model. The signicantly negative abnormal return associated with stock issues and the non-signicant abnormal return associated with debt issues could be interpreted as evidence for the adverse selection model. However, an analysis of the determinants of the wealth eects indicates that the results are not likely to be driven by the adverse selection model. Therefore we conclude that both the studies on the motives for the debtequity choice and the study on the wealth eects do not provide evidence for this model. The remainder of this paper is organized as follows. In Section 2, a review of the previous theoretical and empirical review on this topic is presented. In Section 3 the data and the methodology are described. The results of this study are included in Section 4. Finally, a summary and conclusions are presented in Section 5. 2. Literature review 2.1. Theory on the debtequity choice: Hypotheses Moral hazard problems may determine the choice for debt or equity. Jensen (1986) explains that rms may engage in projects with negative NPV, because managers aim for growth of the rm. Debt disciplines the overinvestment behavior. For rms that have a relatively large amount of free cash ow and relatively low growth opportunities this problem is relevant. These rms are expected to issue debt, because of its disciplining role. Hypothesis 1a. Lack of growth opportunities and existence of free cash ow induce debt. 3; 4 Zwiebel (1996) argues that Jensen's theory requires a disciplinary device that forces managers to use debt. This implies that in case such a force is not present, managers will avoid the use of debt. Entrenched managers will avoid the disciplining role of debt. In the Netherlands, managers are entrenched
See also Stulz (1990). It is important to realize that our analysis regards debt and equity issues. Therefore, this hypothesis is conditional upon the company issuing new securities. We do not try to answer the question whether a company, which at the same time possesses free cash ow and a lack of growth opportunities, should issue securities at all.
4
1862
because disciplinary devices are largely absent. Therefore, these managers are likely to avoid debt. In our empirical model we assume that all Dutch companies lack a discipliner. This is based on the Dutch institutional setting where virtually all companies have adopted multiple takeover barriers. Therefore, it is almost impossible to distinguish dierent levels of disciplining. 5 Hypothesis 1b. Lack of growth opportunities and existence of free cash ow induce equity. Zwiebel describes a model in which pressure on managers to commit voluntarily to debt is derived from the constant pressure of a potential discipliner. The pressure is partially limited due to managerial entrenchment. Managers make a trade-o between expanding their empire and retaining control. In Zwiebel's model, empire building is constrained by a take-over and by debt. The constraint by debt is accomplished through the threat of a bankruptcy, and the associated loss of entrenchment. However, in the Netherlands the likelihood of a hostile take-over is negligible. Managers are expected to avoid debt if they are likely to overinvest, and if the possibility of a bankruptcy is larger. For this reason we expect that rms that face a threat of bankruptcy are more likely to issue equity. A bankruptcy threat is expected to occur for rms that have a low protability. Hypothesis 2. Threat of bankruptcy (i.e. the reverse of protability) induces equity. 6 Myers and Majluf (1984) describe an adverse selection model. They discuss the impact of asymmetric information in case investors are less informed about the value of the rm than insiders. In this case equity may be mispriced. If the rm has a project with a positive NPV available, the underpricing of the equity may be higher than the value of the project, and it will be passed. The underinvestment problem can be overcome by using a less risky form of nancing. For this reason rms have a preference to maintain slack 7 in order to have internal funds available for valuable projects. After internal funds, debt is preferred to outside equity. The ordering of preferences is called the pecking order, and leads to the hypothesis that debt would be preferred over outside
5 An additional reason is that dierent levels of disciplining are hardly measurable, due to lack of available data. 6 The threat of a bankruptcy is also inuenced by the amount of slack that the rm possesses. For this reason we also look at the interaction between slack and protability. 7 The dierence between slack and free cash ow is important. Slack is cash, liquid assets and unused borrowing power (Myers and Majluf, 1984). Free cash ow is the cash ow that remains after all positive NPV projects are undertaken (Jensen, 1986).
1863
equity. A number of factors inuence the ndings of the adverse selection model. The presence of slack increases adverse selection costs and makes an equity issue costly relative to a debt issue. Therefore, rms that have a relatively large amount of slack are more likely to issue debt. Hypothesis 3. Slack induces debt. 8 Cooney and Kalay (1993) re-examine the model of Myers and Majluf and allow rms to have projects with positive and negative NPVs. For rms with positive NPV projects the negative aspects of an equity issue may be overwhelmed by the good news of the acceptance of the project. According to this result, the costs of asymmetric information in an equity issue can be reduced by the expectation that the market has with regard to the protability of the projects. The expected prots of new projects are reected in the value of the growth opportunities. Hypothesis 4. Growth opportunities induce equity. Note that the combination of Hypotheses 3 and 4 seems similar to Hypothesis 1a from the moral hazard framework, i.e., lack of growth opportunities and free cash ow induce debt. However, both theoretical and empirical implications are dierent. Jensen discusses overinvestment behavior in a situation with excess capital for investments, while the discussions based on Myers and Majluf deal with underinvestment due to limited provision of capital. The empirical implications for overinvestment require the combination of cash and low growth prospects. In the underinvestment problem both cash and (a lot of) growth opportunities (with a non-penalized debt or equity issue) can overcome underinvestment. 9 Krasker (1986) presents a model in which insiders choose the size of the issue (i.e., the size of the investment project). In this model the decrease in the stock price due to the mispricing will increase with the relative issue size. For this reason, the larger the issue the less likely it is to be equity. Hypothesis 5. Relative issue size induces debt.
8 Like Hypothesis 1a, this hypothesis is conditional upon the company issuing new securities. We do not study the question whether a company, that possesses slack, should issue securities at all. 9 The dierence between Hypotheses 1a and 1b on one side and Hypotheses 3 and 4 on the other side is that in Hypotheses 1a and 1b we have taken two variables together (free cash ow and growth opportunities). In Hypotheses 3 and 4 we have included the variables (slack and growth opportunities) in two separate hypotheses. The reason for this is that in Hypotheses 3 and 4 the variables can separately inuence the choice between equity and debt. In Hypotheses 1a and 1b the variables jointly inuence this choice.
1864 A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895
Table 1 The hypotheses tested in this paper and the proxies that are used to test these hypothesesa Variables (1) Moral hazard models Jensen (1986): Lack of growth opportunities and existence of free cash ow Zwiebel (1996): Lack of growth opportunities and existence of free cash ow Zwiebel (1996): Protability Future acquisition Capital expenditures Repay short-term debt Adverse selection models Slack Growth opportunities H12: ) Issue choice (2) H1a: Debt H1b: Equity H2: Debt Wealth eect Debt (3) H8: + Equity (4) H9: ) Equitydebt (5) Proxy (6) Overinvestment dummy: 1 if Tobin's Q < 1 and free cash ow > 0, otherwise 0 Overinvestment dummy: 1 if Tobin's Q < 1 and free cash ow > 0, otherwise 0 Operating income/Total assets (BV) Dummy: 1 if purpose is future acquisition, otherwise 0 Dummy: 1 if purpose is capital expenditures, otherwise 0 Dummy: 1 if purpose is repay short-term debt, otherwise 0 H13b: ) H14b: + (Cash + liquid assets)/Total assets (BV) + Max [0, Average debt ratio industry ) debt ratio rm] Tobin's Q (MV equity + BV debt)/ replacement value assets
H10: ) H11: +
H3: Debt H4: Equity
H13a: ) H14a: +
H13a: ) H14a: +
Relative issue size Stock price run-up Rights oering Static trade-o model Deviation from target Movement toward target Control variables Frequent issuer Issue size
a
H5: Debt H6: Equity
H15a: ) H16a: )
H15a: ) H16a: ) H17: +
H15b: ) H16b: )
Issue size/total assets (BV) Stock return ) market return (12-month period preceding issue) Dummy: 1 if oering is rights issue, otherwise 0 Expected equity ratio of regression ) Actual equity ratio Dummy: 1 if issue moves equity ratio towards target, otherwise 0 Dummy: 1 if > 10 issues in 20-year period, otherwise 0 Log(issue size)
H7: Equity H18: + H18: +
This table includes the hypotheses that are tested in this paper as well as the proxies that are used to test these hypotheses. Column (1) presents the variables that are included. Column (2) refers to the choice between equity and debt. The terms `Debt' and Èquity' refer to the type of securities that will be issued if the corresponding variable has a higher value. The signs in columns (3)(5) refer to the eect that the variable has on the wealth eects. In columns (3) and (4), a positive sign (+) means that a positive eect is expected on the wealth eect and a negative sign ()) means the opposite. The signs in column (5) refer to the eect that the variable has on the dierence of the wealth eects between equity and debt. A positive sign (+) means that an equity issue is aected more positively than a debt issue and and a negative sign ()) means the opposite. The proxies that are used to test the hypotheses are included in column (6). Free cash ow is dened as operating income minus taxes, interest and dividends, divided by the market value of equity.
1865
1866
Lucas and McDonald (1990) construct a model in which rms `time' equity issues and information asymmetries are temporary. Undervalued rms will wait until the mispricing is reduced. The model implies that rms issue equity after a period of positive abnormal returns, in which the underpricing is possibly reduced. Hypothesis 6. Stock price run-up induces equity. A crucial assumption of the adverse selection model is that managers try to maximize wealth of current shareholders. Given the entrenchment of Dutch managers it is unlikely that this model will also hold for the Netherlands. Therefore we expect insignicant results for the factors that are associated with this model. 10 According to the static trade-o theory rms have an optimal capital structure, i.e. they aim for a target level of equity or debt to total capital. In the static trade-o framework the optimal capital structure is determined by the tax structure, the costs of bankruptcy, and agency problems. If this theory fully explains the choice between issuing new debt and new equity, rms would tend to move towards the optimal capital structure. Hypothesis 7. If the actual equity ratio is below the target equity ratio, an equity issue is induced. 11 Columns (1) and (2) in Table 1 summarize the hypotheses for the debt equity choice. 2.2. Theory on the wealth eects: Hypotheses 12 Jensen (1986) describes that the reaction to an issue depends on the expected purpose of the capital and the type of security that is issued. As managers have
10 Eckbo and Masulis (1992) argue that the adverse selection problems can be mitigated by the choice of the otation method. In a rights oering all shares may end up in the hands of the existing shareholders, therefore adverse selection between existing shareholders and new shareholders may become irrelevant. In a rights issue the adverse selection arises by the ratio of shares bought by existing shareholders to shares bought by new shareholders. We assume that the otation method does not inuence the debtequity choice. 11 There is no distinct separation between the determinants of the optimal capital structure and the determinants of the debtequity choice. However, according to theory, several variables specically inuence the marginal costs and benets of debt versus equity in the rm's capital structure. Empirical studies use these determinants of the capital structure, assuming that if a rm has more advantages of debt, it is also more likely to issue debt and vice versa. This is contradictory to the static trade-o model. 12 We thank an anonymous referee for very helpful suggestions with regard to these hypotheses.
1867
incentives to increase the size of the rms, even with negative NPV projects, the abnormal returns are determined by growth opportunities and free cash ow. If rms issue debt, they choose to discipline overinvestment behavior. Debt is valued mostly if a rm has a lack of growth opportunities and at the same time it has a large amount of free cash ow. Therefore we expect a positive relationship between the abnormal return associated with the issuance of debt and the combination of a lack of growth opportunities and the existence of free cash ow. Hypothesis 8. For debt issues: A lack of growth opportunities together with the existence of free cash ow aects abnormal returns positively. For equity issues the contrary story holds. Equity is least valuable if a rm has a lack of growth opportunities and at the same time it possesses free cash ow. In other words, a negative relationship is expected between the abnormal returns associated with the equity issues and the combination of a lack of growth opportunities and the existence of free cash ow. Hypothesis 9. For equity issues: A lack of growth opportunities together with the existence of free cash ow aects abnormal returns negatively. From a moral hazard point of view we expect that the use of funds from an equity issue to nance a future, yet unidentied, acquisition is bad news. Hypothesis 10. For equity issues: If the rm announces that it will use the proceeds to nance a future, yet unspecied, acquisition, abnormal returns are aected negatively. McConnell and Muscarella (1985) nd that announcements of capital expenditures by companies are associated with positive abnormal returns. Hypothesis 11. For equity issues: If the rm announces that it will use the proceeds to nance capital expenditures abnormal returns are aected positively. Short-term debt is more eective as a monitoring device for agency problems than long-term debt. The reason for this is that short-term debt gives the capital market more opportunities to monitor rms when they seek to renew it with other forms of capital. Hypothesis 12. For debt issues: If the rm announces that it will use the proceeds to pay back short-term debt, abnormal returns are aected negatively.
1868
In the Myers and Majluf (1984) framework issues are interpreted by investors as bad news, because the rm is expected to be overvalued. This results in a stock price decrease, which is larger if the security is riskier. In other words, if the adverse selection model holds, equity and debt lead to non-positive abnormal returns. According to the adverse selection model of Myers and Majluf (1984) the abnormal returns will be more negative for equity than for debt. It should be mentioned that a priori we do not expect the adverse selection model to hold for the Netherlands, based on the earlier mentioned entrenchment of Dutch managers. For this reason, we also expect non-significant results for the variables that drive this model. These variables are slack, growth opportunities, relative issue size and a stock price run-up. If the adverse selection model would hold, these variables would have the following coecients. If a rm with slack issues debt or equity, the investors are more likely to assume that the rm is overvalued, because slack provides an alternative source for nancing new projects. This eect is stronger for the issuance of equity. Hypothesis 13a. For debt and equity issues: Slack aects abnormal returns negatively. Hypothesis 13b. Slack aects abnormal returns more negatively for equity than for debt. The existence of growth opportunities decreases the problem for both equity and debt issues. This eect will be stronger for equity than for debt issues. Hypothesis 14a. For debt and equity issues: Growth opportunities aect abnormal returns positively. Hypothesis 14b. Growth opportunities aect abnormal returns more positively for equity than for debt. 13 Krasker (1986) generalizes the model of Myers and Majluf (1984) by making the issue size a continuous variable. He shows that the abnormal return for both equity and debt issues is negatively related to the relative size of the issue. The abnormal return is more negative for equity than for debt issues. Hypothesis 15a. For debt and equity issues: Relative issue size aects abnormal returns negatively.
Cooney and Kalay (1993) extend the Myers and Majluf (1984) model by assuming that managers may also accept negative NPV projects. In their model, growth opportunities aect the stock price reactions positively. This may even lead to a positive eect in equity issues.
13
1869
Hypothesis 15b. Relative issue size aects abnormal returns more negatively for equity than for debt issues. In the model of Lucas and McDonald (1990) rms time equity issues when equity is overvalued. This induces an adverse selection problem. It is expected that after a stock price run-up rms raise equity because of overvaluation. The issue reveals this overpricing, and the run-up will be negatively related to the stock price run-up. If a rm issues (risky) debt, the problem of perceived overvaluation still exists, but the adverse selection problem and thus the impact on abnormal returns will be smaller. 14 Hypothesis 16a. For debt and equity issues: A stock price run-up aects abnormal returns negatively. Hypothesis 16b. A stock price run-up aects abnormal returns more negatively for equity than for debt issues. Eckbo and Masulis (1992) describe that the adverse selection problems for equity can be reduced with the choice of the otation method. In the Netherlands, we nd three methods, i.e. standby rights oerings, 15 rm commitment oerings, 16 and private placements. Amongst these three methods rights issues are dominant. The wealth eects depend on the fraction of new shares that will be owned by existing shareholders. If the fraction is larger, overvaluation is less likely to be the motive for the issue. In a rights oerings the existing shareholders are in the position to buy the new shares, while in a rm commitment underwritten oering new investors are more likely to be attracted. In other words we expect that for equity issues, the choice for a rights oering aects abnormal returns positively.
14 See Lucas and McDonald (1990, pp. 10361037). Viswanath (1993) derives a counter hypothesis for equity issues. In the Viswanath (1993) model the stock price run-up is the result of an accumulation of growth opportunities. The issue enables managers to carry out the project. In this model the investors are expected to interpret the stock price run-up as evidence that the rm has a positive NPV project and they discount the value of this project. Therefore, Viswanath (1993) concludes that a stock price run-up aects abnormal returns from an equity issue positively because a larger run-up indicates a larger positive NPV project that can be realized. The implication from the Viswanath (1993) model is the opposite of the Lucas and McDonald (1990) model. The dierence between the two models consists of the fact that in the model of Lucas and McDonald (1990) the investors expect the project and react on the issue, while in the model of Viswanath (1993) investors react on both the news of the project and the news of the issue. 15 In a standby rights oering the underwriter guarantees the proceeds on any unsubscribed shares. All Dutch rights oerings are standby rights oerings. 16 In a rm commitment oering the underwriters buy the securities from the issuer and resell them to the public.
1870
Hypothesis 17. For equity issues: The choice of a rights oering aects abnormal returns positively. According to the static model a choice of either debt or equity decreases the dierence between the target and the actual equity ratio. For this reason an issue that moves the actual ratio towards the target will be associated with a positive abnormal return. An issue that moves the actual ratio from the target will be associated with a negative abnormal return. Hypothesis 18. The abnormal return is positively related to a movement towards the target equity ratio. Columns (3)(5) of Table 1 summarize the hypotheses for the debtequity choice. 2.3. Empirical evidence Empirical studies on the debtequity choice consist of studies that investigate motives for the choice between equity and debt and studies in which the wealth eects for this choice are investigated. Recent studies on motives for the debtequity choice are only available for the US. Bayless and Chaplinsky (1991) and Jung et al. (1996) carry out logistic regressions to examine the security issue decision. Bayless and Chaplinsky (1991) nd that slack and a large relative issue size induce debt and that an abnormal positive stock price performance induces equity. These facts can all be considered as evidence for the adverse selection model. Jung et al. (1996) nd that growth opportunities and an abnormal positive stock price performance induce equity and that a relative large issue size induces debt. This can also be considered as evidence for the adverse selection model. Besides that, Jung et al. (1996) nd that companies that issue equity, but were expected to issue debt, have poor investment opportunities. This can be considered as evidence for the moral hazard model. Berger et al. (1997) study the relationship between leverage changes and managerial entrenchment reducing shocks. Leverage changes are based on the annual ow of funds data from Compustat. They nd that leverage increases after managerial entrenchment reducing shocks. These shocks consist of unsuccessful tender oers, involuntary CEO replacements and the addition to the board of major stockholders. This result of Berger et al. (1997) is consistent with entrenched managers seeking to avoid debt. The wealth eects of security issues have extensively been studied for the US. Review studies by Smith (1986) and Eckbo and Masulis (1995) present the following results. Common stock issues are associated with signicant negative abnormal returns. Issues of convertible bonds are also associated with negative abnormal returns, though the stock price decline is smaller than the stock price
1871
decline associated with the issuance of common stock. Issues of straight debt are not associated with a signicant abnormal return. 17 Pilotte (1992) and Denis (1994) study the eect of growth opportunities on the abnormal returns of equity issues. They nd a positive relationship between the announcement eects and the existence of growth opportunities. Other studies investigate the relationship between the announcement eects and the choice of the otation method. Eckbo and Masulis (1995) show a large dierence in abnormal returns for US industrial rms that issue equity under dierent otation methods. The authors document a negative abnormal return of )3.1% for rm commitment oers, )1.5% for rights oers with a standby agreement, and )1.4% for rights oers without such an agreement. The more positive receipt of rights oers is conrmed by Bhren et al. (1997) who nd a positive abnormal return for rights oers in Norway. 3. Data description and methodology 3.1. Data description Our analysis concerns rms in the data set of the Dutch Central Bureau of Statistics containing data of listed non-nancial rms with annual reports since 1974. We study primary issues of stocks and issues of new bonds by listed nonnancial Dutch companies, which are made from 1 January 1977 to 31 December 1996. The issues are identied from two sources. Issues of shares and issues of domestic bonds are included in the monthly bulletin of the Dutch Central Bureau of Statistics (Maandstatistiek Financiewezen CBS). Issues of foreign bonds and Eurobonds can be identied from the quarterly review of the Dutch Central Bank (Kwartaalberichten Nederlandsche Bank). During this period a total of 110 public and private issues of common or preferred stock can be identied. These issues do not include initial public oerings. Besides that we can identify 137 issues of straight debt in the same period. For the issues to be included, data for at least one nancial year has to be available, before the nancial year in which the oering took place. The stock prices of the rms are obtained from Datastream. The market index chosen is the Datastream index for the Dutch market, since this is the only index for which data are available for the whole sample period.
Studies on announcement eects of stocks and hybrid debt for Japan give dierent results. Kang and Stulz (1996) and Kang et al. (1995) study announcement eects of stocks, convertible bonds and warrant-bonds issued by Japanese companies. They nd non-signicant or sometimes even signicantly positive abnormal returns. They attribute these abnormal returns to corporate governance dierences between Japan and the US.
17
1872
In order to investigate the motives for the choice for debt and equity we compare the characteristics of rms in the group of equity-issuers versus the group of debt-issuers. The characteristics are based on annual report data for the end of the nancial year before the issue. 18 In most proxies variables are taken relative to the rm's book value, instead of the market value. We have three reasons to do so. First, managers consider book values in decisionmaking. Second, as the stock price run-up is expected to be an important discriminating variable this has impact on the market value of rms. Third, other studies generally use book values. By using book values we are able to compare our results with these results for earlier studies. In the Netherlands some large rms issue debt or equity relatively frequently. Three rms have more than 20 issues in the 20-year period, and they account for 7% of the equity issues and 53% of the debt issues. In total seven rms have made more than 10 security issues each. The issues of these rms represent 24% of the equity issues and 77% of the debt issues. The announcement date is the rst date at which the announcement appears in the Dutch nancial newspaper Het Financieele Dagblad. 19 From our original sample of 110 issues of public and private equity and 137 issues of public debt, we can not nd an announcement date for 11 equity issues and 39 debt issues. From the remaining sample we eliminate 10 equity issues because the announcement of these issues was accompanied by the announcement of the issuance of other securities. Furthermore ve equity issues are eliminated because they are announced together with a complete nancial reorganization. Therefore we are able to use 98 debt issues and 84 equity issues. 3.2. Proxies The variables that are used in the analysis are related to the hypotheses mentioned in the previous section. Growth opportunities. Empirical studies generally measure growth opportunities by Tobin's Q. As Tobin's Q is not directly observable, most studies proxy Tobin's Q by the market-to-book ratio. In this study we use a more precise estimate of Tobin's Q. This approximation is described by Perfect and Wiles (1994). They dene Tobin's Q as the market value of equity plus the book value of debt divided by the replacement value of the assets. In the Netherlands rms either present replacement values directly in their annual reports or they present historical costs. If a replacement value is reported our
18 In general, a nancial year is from January until December. However, some rms have nancial years that start in February, April, July, or December. 19 We either use the electronic (on-line) version (for the issues from 1985) or the hard copy version (for the issues before 1985).
1873
measure for replacement value equals the book value of total assets. In case of historical costs we have to adjust the value to approximate the replacement value. This is relevant for plant and equipment, because these assets are depreciated over a longer horizon. For the same reason, the adjustment is not relevant for inventory and other current assets. We assume that in a base year the replacement value equals the historical costs. For each year's next year we adjust this replacement value by adding new investments and the growth in capital good prices and by subtracting depreciation. The base year is 1974 or the rst year for which rm data are available. Growth in capital good prices is based upon the price index of investment goods, as provided by the Dutch Central Bureau of Statistics. The replacement value of the assets is the book value of assets plus the dierence between the replacement value and historical value of plant and equipment. 20; 21 Free cash ow. Following Lehn and Poulsen (1989) we measure free cash ow as operating income minus taxes, interest expenditures and dividends paid, divided by the market value of equity. 22 Alternative proxies are the use of sales and total assets as denominator. The results for the alternative proxies are not presented as they give similar results. Overinvestment dummy. In the model of Jensen (1986) growth opportunities interact with free cash ow, i.e., an overinvestment problem occurs if a rm both has low growth opportunities and free cash ow. For this reason we have constructed a dummy variable that has a value of one if the rm has positive free cash ow and a Tobin's Q lower than one. Otherwise the dummy variable is zero. Protability. We measure protability as operating income over total assets (ROA). Operating income is the earnings before interest and taxes (EBIT). Total assets are measured in book values. We also use an alternative proxy for protability (earnings before depreciation, interest and taxes over total assets) but do not report the results here because our conclusions are insensitive to our choice of proxies. Relative issue size. Relative issue size is included as the issue size divided by the total assets (in book values).
20 We also calculated the market-to-book ratio of total assets. The correlation between the market-to-book ratio and our measure of Tobin's Q is very high, i.e., 0.991. 21 Note that the market value of a rm consists of assets in place and growth opportunities. Tobin's Q reects the relative importance, but a stock price run-up is also related to this. An increase in stock prices, with a constant value of assets in place, represents the market's recognition of the existence of future valuable projects (see Viswanath, 1993). 22 Our proxy for free cash ow is based on an accrual measure for operating income. Therefore, we have to assume that operating income equals the dierence between the cash inows and outows in the nancial year. Although this assumption will hold on average over longer horizons, short-term deviations are possible.
1874
Stock price run-up. The stock price run-up is measured as the stock return minus the market return over a period of 12-months preceding the issue. The six-month stock price run-up gives comparable results as the 12-month runup. For this reason the results of the six-month run-up are not reported here. Deviation from target capital structure. In order to test the static trade-o theory, we have to compare the target capital structure with the actual capital structure. Unfortunately, only the actual debt or equity ratio is observable, while the target ratio is unobservable. The rst possible approach is to approximate the target debt or equity ratio by its long-term average. In this approach it is assumed that the target ratio is relatively stable over time, and that the actual ratio uctuates around it. A second possible approach is to use the long-term average debt or equity ratio in the same industry instead of in the same rm. This approach assumes that rms in the same industry have similar target ratios. Opler and Titman (1997) suggest a third approach. They use a cross-section of rms to estimate the determinants of the actual debt ratio. In doing so they assume that the determinants perfectly determine the target debt ratio, and that the residual is the deviation from target. In our study we use all three approaches. We use market and book values for all approaches. Following Marsh (1982) and MacKie-Mason (1990) we measure the long-term average over the 10-year period preceding the year in which the actual equity ratio is measured. As our data-set only goes back to 1974 this is not always possible. In such cases we use the years after the issue instead of the years before the issue. For example, for an issue from 1980, we use the data for the period from 1974 to 1983. The industry average equity ratios are measured using the industry classication of the Dutch Central Bureau of Statistics. For each security issue of a rm we take the average equity ratio in the same industry in the year preceding the issue. The issuing rm is excluded from the average. 23 The market value of equity is calculated as the number of outstanding shares multiplied by the stock price at the end of the year preceding the issue. The market value of debt is assumed to be equal to its book value. In order to calculate the expected equity ratio we estimate the determinants of the optimal equity ratio from a cross-section of rms in the 19761995 time period. The expected equity ratio is the sum of the products of the coecients for the determinants and the actual rm characteristics. The dierence between the expected and the
The average book value equity ratio in the industry is calculated from the 2405 rm-year observations in the 19761995 data set discussed in Appendix A. We use the industry classication of the Dutch Central Bureau of Statistics, which consists of nine dierent industries. We exclude issuing rms from the averages. Because we require the presence of at least ve rms in an industry in a specic year, we leave out three issues in the analysis of industry target ratios.
23
1875
actual equity ratio is the residual in the regression. The determinants are chosen to represent the static trade-o theory and include the rms tax structure and bankruptcy risk and costs. The cross-sectional regressions are included in Appendix A. Slack. Slack consists of two components. The rst component is cash and liquid assets. This component is measured as the sum of cash and liquid assets divided by the book value of the rm. Liquid assets are the securities that the rm holds as investments. The second component is unused debt capacity (Brealey and Myers, 2000). This component is measured as the maximum of zero and the average debt ratio in the industry minus the rm's debt ratio. Slack is proxied as the sum of the two components. Control variables. While the previous variables were based on sound theoretical arguments, several variables are often used in other studies and prove to be relevant, but are not explicitly based on theory. We include a dummy for rms that frequently issue new capital on the public market. The dummy frequent issuer has a value of one for rms that have more than 10 issues in total of debt or equity in the 20-year period, and zero otherwise. We measure issue size by the log of the issue size. Variables from the static trade-o theory, that are also variables in dynamic models are often included in empirical studies. Opler and Titman (1997) state that this leads to a misspecication of the empirical model. For this reason these variables are not included in the logit regressions. Most of the variables that determine the choice for debt or equity are also relevant as determinants of the wealth eects of the security issue choice. The variables for slack, free cash ow, relative issue size, stock price run-up, growth opportunities, protability, the overinvestment dummy, frequent issuer and issue size are the same as the variables used in the logistic regressions. The earlier mentioned variable deviation from target measures the deviation of the actual equity ratio from the expected equity ratio, measured in book values in the year preceding the issue. This variable was shown to be relevant in the logit analysis. In this regression we include a dummy for the issuance towards target. This dummy is one if the issue moves the actual equity ratio towards the target and zero otherwise. Information provided with the announcement. De Roon and Veld (1998) study wealth eects for Dutch convertible bonds and warrant-bonds. They nd that in a large number of cases Dutch companies package announcements of hybrid debt issues in other (good) rm specic news. For this reason we check in the Dutch nancial newspaper Het Financieele Dagblad on the announcement day whether the issue was contaminated by the presentation of other rm specic news. The following types of contaminating information were identied: issue announcements made together with the presentation of the annual results of the company ( annual earnings);
1876
issue announcements accompanied by the presentation of the updated provisional results for the current year ( current earnings); issue announcements accompanied by news that the company has just completed an acquisition ( acquisition) and/or issue announcements made together with other rm specic news, such as a stock split, a large order, etc. ( strategic information). In all other cases the issue is classied as uncontaminated. We notice that eliminating all contaminating issues, like is done for other studies, is not possible since practically all equity issues in our sample are contaminated. Purpose of the issue. Following Jung et al. (1996) we also checked in Het Financieele Dagblad whether the purpose of the issue was mentioned. Jung et al. (1996) identify the following purposes: capital expenditures; repayment of long-term debt; repayment of short-term debt and/or repayment of bank debt. In addition we also include acquisition as a possible purpose. In a number of cases the company mentions that it will use at least a part of the proceeds of the issue to nance an acquisition. We include both the information provided with the announcement and the purpose of the issue in the regressions. 24 We include a dummy for acquisition information. This dummy is one if the announcement of the equity or debt issue is done together with the announcement of a recently completed acquisition and zero otherwise. Dummies are also included for two dierent purposes of the issue: capital expenditures and repayment of short-term debt. Finally we include a dummy if the company reveals that it wants to use at least a part of the proceeds for future acquisitions, which are yet unidentied. In all cases the dummy is one if the specic purpose is mentioned and zero otherwise. Flotation method. The dominant otation method in the Netherlands are rights issues (59% of the equity issues). We also have private placements of common equity (23%). The remainder are rm commitment underwritten offerings (17%), which are often foreign issues. Expected type or issues àgainst' type. On the basis of the logit analysis it is possible to compare the predicted security type with the type of security that is actually being used. 25 We include a dummy variable for correctly predicted security type that is one if the issue was predicted correctly with regression (2) in Table 3, and zero otherwise. The proxies are included as column (6) of Table 1.
Jung et al. (1996) also divide their sample according to the purpose of the issue. However, they do not include this information as variables in the regressions. 25 See e.g. Bayless and Chaplinsky (1991) and Jung et al. (1996).
24
1877
3.3. Methodology We have started by investigating an empirical model of the security issue choice for our sample rms. This model uses the earlier dened variables from the literature to predict the security issue choice. The model has the form of a logit regression in which the debtequity choice is measured by a dummy variable with a value of 1 for equity issues and 0 for debt. A positive estimate implies that rms that have higher values for the determinants are more likely to issue equity. The variables that are used in this model are specied in Section 3.2. The announcement eects of the equity and bond issues are measured using a standard event study methodology as described in Brown and Warner (1985). Excess returns are measured using an Ordinary Least Squares (OLS) market model regression: ^ ^ Ai;t Ri;t b0;i b1;i RM ; t where Ai;t is the abnormal return for rm i at day t; Ri;t denotes the return on security i at day t, dened as lnPi;t lnPi;t1 , and RM is the return on the t market index, that is measured in a similar way as Ri;t . The market index chosen is the Datastream index for the Dutch market, since this is the only index for which data are available for the whole sample period. The parameters ^ ^ b0 and b1 are estimated over the estimation period by running an OLS regression of the stock returns on a constant and the return on the market index. Denoting the announcement date as day 0, this estimation period ranges from day )120 to day )20. The event window ranges from day )1 to day +1. The test statistic is calculated using the method outlined by Brown and Warner (1985, p. 7) and is dened as At =Ât ; s where At is the average abnormal return over the N dierent rms on day t and Ât is the standard deviation of the average abnormal return obtained from s the estimation period. The null-hypothesis is that the abnormal return is zero. If the null-hypothesis holds and if the abnormal returns are independently identically distributed with nite variance, the test-statistic is asymptotically normally distributed. Besides calculating excess returns for each day in the event period, we also calculate cumulative average abnormal returns. 26
In calculating the test-statistic for the cumulative abnormal returns it is assumed that excess returns are not autocorrelated, so that the variance of a two-day excess return is just the sum of the variances of the corresponding one day returns. This test assumes cross-sectional independence among abnormal returns. This is a realistic assumption given the large number of dierent companies in our data-base.
26
1878
3.4. Moral hazard versus adverse selection Before discussing the results, it is useful to clearly distinguish the moral hazard and the adverse selection models in terms of underlying assumptions, predictions and relevant proxies. The moral hazard models of Jensen (1986) and Zwiebel (1996) both assume that managers only follow their own interests. The main interest for managers lies in the growth of the rm. Managers do not mind investing in bad projects as long as it leads to rm growth. The important dierence between the two moral hazard models is that Jensen (1986) assumes that rms which have a relative lack of growth opportunities and at the same time possess a relative large amount of free cash ow will issue debt. In the model of Zwiebel (1996) managers will avoid the disciplining role of debt and therefore issue equity. The reason is that Zwiebel (1996) assumes that the discipliner will be limited by managerial entrenchment. The adverse selection model assumes that managers act in the interests of existing shareholders. Therefore, they are expected to sell the security that benets these shareholders the most. An important reason for managers to sell equity in this model can be that equity is overpriced. The dierent assumptions and set-up of the moral hazard and adverse selection models lead to interesting dierences with respect to the hypotheses. The dierences in hypotheses show up in the predicted behavior of both the managers who issue debt or equity and the shareholders' reactions. This is shown in Table 1. Under the heading Moral hazard models the overinvestment dummy, protability, and information on the purpose of the issue (future acquisition, capital expenditures and repayment of debt) are mentioned as relevant variables. In column (2) the predicted inuence of these variables on the choice of the managers is shown (Hypotheses 1a, 1b, and 2). Moral hazard behavior is associated with shareholders' reactions as listed in columns (2) and (3) (Hypotheses 812). Similarly, the adverse selection models yield hypotheses on the managers' choices, which are reected in column (2) (Hypotheses 36). The predicted shareholders' responses are in columns (3)(5) (Hypotheses 13a17). The relevant variables in the adverse selection model are slack, growth opportunities, relative issue size, stock price run-up and rights oerings. Most proxies that we use are exclusively associated with either adverse selection (e.g. slack, relative issue size, stock price run-up) or with moral hazard (e.g. future acquisitions, protability, etc.). Growth opportunities is the only variable that plays a role in both the adverse selection model and the moral hazard models. However, in the adverse selection model it plays a role in itself. In the moral hazard models it only plays a role in the
1879
interaction with free cash ow. In these models growth opportunities are also included as a part of the overinvestment dummy. 27 4. Results In Panel A of Table 2 we present the characteristics of rms that have issued equity versus rms that have issued debt. In this table we present the standard test for the dierences in means. We also include a non-parametric test for dierences in medians. In the latter test, for both debt and equity issues we have counted the number of observations for each issue type below and above the median. The test describes whether debt or equity has a signicantly larger number of observations that is higher or lower than the median. 28 In Panel B the 12-month excess returns before and after the issues are presented. In Panel A of Table 2 it can be seen that the average return on assets for equity issuing rms is 7.0%. For debt issuing rms the average return on assets is 9.8%. This indicates that debt issuing rms are more protable than equity issuing rms. For equity issues we nd that the average dierence between the target equity ratio, determined using a cross-sectional regression specied in Appendix A, and the actual equity ratio is 0.058. This implies that the target was above the actual level of equity and that the new equity moves the rm towards its target. For the debt issuers the average is 0.019. As this is positive the rms are expected to issue equity. However, the deviation is much smaller than for equity and this dierence is signicant at the 1% level. 29 The results for target ratios based on long-term averages and on industry averages show that all these measures lack predictive power. The only exception is the long-term average book value measure. However, the signicance level here is only 10%, while the
27 There is an additional overlap in the proxies. Protability, measured as return on assets, is a variable in the moral hazard models. Slack, measured as the sum of cash and liquid assets divided by the total assets and unused debt capacity, only plays a role in the adverse selection model. However, in one of the logit models we have also included an interaction variable between protability and slack. This variable measures the avoidance of bankruptcy in the Zwiebel model. This variable is used in only one of the logit regressions. 28 We have performed an additional non-parametric test, the KruskalWallis test. This test incorporates the mean rank of the median of a group (debt or equity) and describes the probability that the ranks dier. In general the results of the two non-parametric tests are similar. These results are available from the authors. 29 If the ratios are measured in market values, we nd the opposite results. However, these results are insignicant. This nding is remarkable in terms of nance theory where market values are predominant. However, this result is in line with interview results of Cools (1993) who interviewed 50 CFOs of Dutch listed corporations. He nds that in 90% of the rms a target equity or debt ratio is formulated. Normally the ratio is total equity over total assets (52%). To the question of how a target equity ratio should be dened all respondents answered that this should be in book values.
1880 A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895
Table 2 Firm and issue characteristics for the 137 debt and 110 equity issues of Dutch companies from 1977 to 1996a Average debt issuing rm Panel A: Firm and issue characteristics Deviation of actual equity ratio from Expected eq. ratio (book) 0.019 Expected eq. ratio (market) 0.063 Long-term average (book) )0.003 Long-term average (market) )0.053 Industry average (book) 0.080 Industry average (market) 0.075 ROA (operation income over 9.844 total assets) Slack (cash and liquid assets/ 0.073 rm book value plus unused debt capacity) Free cash ow (operating 0.049 income minus taxes, interest and dividends/market value of equity) Log of issue size 5.085 Issue size over book value 0.025 of the rm Tobin's Q 1.044 Book value total assets 18,368 Median debt issuing rm Average equity issuing rm Median equity issuing rm Average difference debt and equity issuing rm p-value Median test: v2 Median test: p-value
0.026 0.076 0.005 )0.068 0.079 0.080 8.900 0.061 0.044
0.058 0.055 0.012 )0.066 0.085 0.043 7.007 0.071 )0.008
0.045 0.068 0.006 )0.063 0.065 0.027 8.700 0.046 0.050
)0.040 0.008 )0.015 0.012 )0.005 0.031 2.837 0.002 0.057
0.002 0.618 0.082 0.334 0.729 0.111 0.001 0.798 0.015
4.317 0.506 0.716 0.213 1.706 3.745 0.116 1.943 0.681
0.038 0.477 0.398 0.644 0.192 0.053 0.733 0.163 0.409
5.136 0.015 0.923 12,030
4.126 0.096 1.091 4626
4.024 0.061 0.979 1126
0.958 )0.072 )0.048 13,741
0.000 0.000 0.382 0.000
31.093 57.920 1.429 79.295
0.000 0.000 0.232 0.000
Panel B: Past and post 12-month excess returns Average debt issuing rm
Average equity issuing rm
Average dierence debt and equity issuing rm )0.207 0.023
p-value
Percentage positive returns debt issuers
Percentage positive returns equity issuers 76.4% 45.3%
Past 12-months excess return Post 12-months excess return

a
)0.047 0.006
0.160 )0.017
0.000 0.561
39.4% 52.3%
In Panel A rm characteristics are presented for the 137 debt and 110 equity issues of Dutch companies from 1977 to 1996. Expected equity ratios are from the OLS-regressions outlined in Appendix A. The long-term average equity ratios are based on the 10-year period preceding the issue. The industry average equity ratios are averages over all other rms in the industry (based the CBS industry classication) in a specic year. Return on assets is operating income over total assets. Total assets is measured in book values (Dutch guilders). The issue size is in Dutch guilders. Tobin's Q is the market value of equity plus the book value of debt divided by the replacement value of the assets. All book values are obtained from the data-set of the Dutch Central Bureau of Statistics containing data of listed non-nancial rms with annual reports since 1974. The test for dierences in medians is a non-parametric test. For both debt and equity issues the number of observations is counted for each issue type below and above the median. The test describes whether debt or equity has a signicantly larger number of observations that is higher or lower than the median. In Panel B the past and post 12-month excess returns are presented. The returns are measured as the dierence between the stock return and the market return over a period of 12 months before and after the issue.
1881
1882
regression approach shows a signicance at the 1% level. In general the results of the median test are similar to the comparison of the means. An interesting result from Panel B of Table 2 is that Dutch companies issue equity after periods of positive abnormal stock price performance. Of the equity issuers, 76.4% have a positive excess return, while for the debt issuers this percentage is 39.4%. Debt is generally issued after periods of negative abnormal stock price performance. This conrms earlier ndings by Asquith and Mullins (1986) and Jung et al. (1996) for US equity and by Dann and Mikkelson (1984), Mikkelson and Partch (1986) and De Roon and Veld (1998) for US and Dutch hybrid debt issues. The results are contrary to ndings of Kang and Stulz (1996) and Kang et al. (1995) for Japanese equity and hybrid debt issues. These studies do not nd that Japanese companies issue equity or hybrid debt after periods of positive abnormal returns. The post issuance stock price performance is not signicantly dierent between equity and debt issuing rms. The outcomes of the logit regressions in which the debtequity choice are explained are presented in Table 3. In regression (1) we nd that a higher return on assets signicantly leads to the issuance of debt. The coecient for slack is signicantly negative. The coecient for past returns is signicantly positive, indicating that rms time equity issues after a period of relative stock price increases. The coecients for Tobin's Q and free cash ow are insignicant. Issue size relative to rm size has a signicantly positive sign. In regression (1) we included several control variables. The dummy for frequent issuers shows a signicantly negative sign. These rms have a tendency to issue debt. The coecient for the log of the issue size has a signicantly negative sign, indicating that larger issues are more likely to be debt issues. In this model we did not nd a conrmation for the role of growth opportunities. However, in (1) growth opportunities are specied as hypothesized in the adverse selection models. In the moral hazard model of Jensen (1986), growth opportunities interact with free cash ow, i.e., an overinvestment problem occurs if a rm both has low growth opportunities and positive free cash ow. In regression (2) we include a dummy variable that has a value of one if the rm has a positive free cash ow and a Tobin's Q lower than one. Otherwise the dummy variable is zero. This overinvestment dummy indicates that rms both have free cash ow and low growth opportunities. The coecient for overinvestment is signicantly positive at the 10% level. Model (3) is an extension of model (2) with an interaction term of protability and slack. This term is included in order to test the relation between protability and the issuance decision. The reason for this term is that a bankruptcy can be avoided for a non-protable rm that possesses enough slack. We expect that a higher value for this interaction term induces debt. We nd that the interaction term is insignicant. This result may be caused by the fact that losses can only temporarily be compensated by a large amount of slack. Finally we check whether the results may be driven by the static trade-o theory. However, the correlation between the deviation
A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895 Table 3 Logit analysis for the debtequity choice of Dutch companiesa (1) Intercept Return on assets Slack Past 12-month excess return Tobin's Q Free cash ow Issue size/total assets Frequent issuer Log(issue size) Overinvestment dummy Return on assets Slack Deviation from target log L Pseudo-R2 % Correct % Equity correct % Debt correct Number of observations 3.84 (3.92) )0.11 ()1.73) )5.94 ()1.85) 2.42 (3.43) 0.61 ()1.01) )1.20 ()0.85) 34.13 (4.94) )1.07 ()2.32) )0.69 ()3.23) )89.47 47.28% 84.21% 81.82% 86.13% 247
1883
(2) 3.10 (2.97) )0.13 ()1.97) )6.26 ()1.93) 2.62 (3.58) )0.28 ()0.44) )1.95 ()1.21) 35.24 (5.03) )1.18 ()2.49) )0.64 ()3.10) 0.89 (1.93) )87.53 48.43% 84.26% 81.82% 86.86% 247
(3) 3.23 (2.99) )0.15 ()1.82) 8.99 ()1.31) 2.61 (3.58) )0.16 ()0.24) )1.90 ()1.14) 34.71 (4.87) )1.18 ()2.46) )0.65 ()3.13) 0.91 (1.97) 0.30 (0.46) )87.43 48.48% 85.02% 82.73% 86.86% 247
(4) 2.09 (2.62) 2.00 (3.05) 27.72 (4.53) )0.73 ()1.68) )0.73 ()3.81) 3.52 (1.71) )99.22 41.54% 83.40% 76.36% 89.05% 247
a Logistic regressions in which the dependent variable takes the value one for equity issues and zero for debt issues. The sample has 110 equity issues and 137 debt issues by Dutch companies from 1977 to 1996. Deviation from target is the deviation of the actual equity ratio from the expected equity ratio according to the OLS-regression in Appendix A. Both the actual equity ratio and the target equity ratio are in book values. Return on assets is operating income over total assets. Total assets is measured in book values. Slack is cash and liquid assets over total assets plus unused debt capacity. The 12-month excess return is measured as the dierence between the stock return and the market return. Tobin's Q is the market value of equity plus the book value of debt divided by the replacement value of the assets. Free cash ow is operating income minus taxes, interest expenditures and dividends paid divided by the market value of equity. Frequent issuer is a dummy variable that has a value of one for rms that have more than 10 issues in the 20-year period and zero otherwise. The overinvestment dummy has a value of one if free cash ow is positive and Tobin's Q is below one, and zero otherwise. All book values are obtained from the data-set of the Dutch Central Bureau of Statistics containing data of listed non-nancial rms with annual reports since 1974. The pseudoR2 equals 1 ) (log-likelihood at convergence/log-likelihood at zero). t-statistics are in parentheses. * Signicant at 10% level. ** Signicant at 5% level. *** Signicant at 1% level.
1884
from target and return on assets is )0.36. Because of this correlation we cannot include both the deviation from target and protability in a regression. Therefore we specify an alternative model in regression (4). Here we include the inuence of a deviation from the target capital structure. In regression (4) the deviation from target has a signicantly positive coecient. This conrms the static trade-o theory. 30 In our analysis we nd a conrmation for the moral hazard model of Zwiebel (1996) in a situation in which managers are entrenched. A lack of growth opportunities and the existence of free cash ow leads to an issuance of equity (Hypothesis 1b). In other words, Dutch managers prefer to issue equity if they are most likely to overinvest. This conrms the Zwiebel (1996) model. We also nd that protability induces debt (Hypothesis 2). Our result is most likely driven by moral hazard behavior of entrenched Dutch managers. 31 It conrms the avoidance of debt, i.e., there is a relationship between the threat of a bankruptcy and the preference for equity. In this theory protable rms issue debt because they are less likely to lose control through a bankruptcy. Following our expectations we do not nd evidence for the adverse selection model. Slack is marginally signicant in the rst two models. However, it is non-signicant in the third model (Hypothesis 3). No relationship is found between growth opportunities and security issue choice (Hypothesis 4). Relative issue size is signicantly positive, while it was hypothesized to be negative (Hypothesis 5). The only result that may be interpreted as evidence for the adverse selection model is the nding that stock price run-ups are consistently related to equity issues (Hypothesis 6). However, the Lucas and McDonald (1990) model is not the only explanation for the phenomenon that rms issue equity after a stock price run-up. Opler and Titman (1997) oer an alternative explanation for this result that is also consistently found for the US. Based on conversations with CEOs, CFOs and investment bankers they consider the possibility that managers issue equity after stock price run-ups because they believe that market prices are too volatile in relation to fundamentals. By timing the equity issues just after a stock price run-up managers rationally respond to inecient markets (Opler and Titman, 1997). Finally we have checked whether our results could have been driven by the static trade-o theory. We do nd a signicantly positive sign for the deviation from target (Hypothesis 7). However, if we look at the explanatory power and
30 It also conrms earlier results by De Jong and Van Dijk (2001). They have studied the capital structure of Dutch companies by sending out questionnaires to these companies in order to ask about the rm's characteristics. Following this they have estimated a structural equations model with latent variables to test capital structure theories. One of their most important ndings is that the static trade-o model oers a good explanation for the capital structure of Dutch companies. 31 An alternative explanation that we did not investigate further is that this result is driven by the signalling model of Ross (1977).
1885
the percentage of correct predictions we see that they are higher for the specications that represent dynamic models. Model (2) shows a pseudo-R2 of 48.43% and classies 84.26% of the issues correctly. Model (4) that includes the static trade-o theory has a pseudo-R2 of only 41.54% and classies 83.40% correctly. 32 In Table 4 univariate results are described for the wealth eects. For equity and debt issues we compare the cumulative average abnormal returns of the announcements. Announcements of equity issues on average lead to a slightly signicant negative abnormal return of )1.07%, while debt issues lead to an insignicantly positive abnormal return of 0.51%. These results are in accordance with the moral hazard and adverse selection models. A further investigation of the stock price reactions should help us to distinguish between these models. First it can be remarked that practically all debt issues in our sample are uncontaminated. The few contaminated debt issues do not give other results than the uncontaminated issues. Contrary to the debt issues, the equity issues are practically all contaminated. The equity issues that are uncontaminated and the equity issues that are announced together with the annual results of the previous nancial year of the company are associated with negative abnormal returns of, respectively, )2.35% and )3.00%. Equity issues announced together with an update of the provisional results of the current nancial year are associated with a non-signicant negative abnormal return of )0.17%. This is an interesting result since in principle updates of provisional results can either be good or bad news. Although we have not tested for this explicitly, we conjecture that a potential explanation for the dierence is that Dutch rms time security issues in such a way that they are announced when good news about provisional results is available. De Roon and Veld (1998) nd a similar result in their study of issues of warrant-bond loans and convertible bond loans by Dutch companies. In 19 cases, the equity issue is announced together with the completion of an acquisition. This joint event is associated with a signicant positive abnormal return of 2.08%. This return can be explained by the fact that 13 out of the 19 acquisitions are cross-border acquisitions. This result is in line with Corhay and Tourani (2000) who nd that cross-border acquisitions by Dutch companies are wealth creating activities. 33 A closer look at the
32 Note that the explanatory power of our models is good compared to other studies. For example, Jung et al. (1996) have pseudo-R2 s between 26% and 41%. In their models 74.6% and 80.8% of the issues are correctly classied. 33 It should also be noticed that the international evidence on wealth eects of bidder rms in cross-border acquisitions is not conclusive. For example, Kang (1993), Pettway et al. (1993) and Eun et al. (1996) nd that Japanese acquirers of American companies experience wealth increases. However, Eun et al. (1996) also nd that British rms that acquire American rms experience wealth reductions. Bodnar et al. (1999) conclude that rms involved in international operations are more valuable than comparable domestic rms.
1886
Table 4 Wealth eects of the debt and equity issues by Dutch companiesa Debt Total sample 0.51 (0.95) [)0.20] 0.43 (0.79) [)0.20] 2.408 (1.57) [0.46] 1.84 (1.02) [0.46] 0.29 (0.49) [)0.22] 10 8 82 89 Obs. 98 Equity )1.07 ()1.66) [)0.98] )3.00 ()2.49) [)2.51] )0.17 ()0.24) [)0.16] 2.08 (1.90) [1.50] )2.35 ()1.09) [)1.49] )0.22 ()0.28) [0.34] )1.24 ()0.63) [)2.89] )1.66 ()1.96) [)0.98] )1.46 ()1.97) [)0.97] )0.52 ()1.15) [0.21] )0.41 ()0.28) [)1.49] Obs. 84
Contaminated Annual earnings Current earnings Acquisitions Uncontaminated
17 51 19 8
Purpose Acquisition Capital expenditures Repay short-term debt No purpose
35 13
39
Flotation method (only equity) Rights issue Private placement Underwritten oering
51 16 17
The sample includes 98 announcements of bond issues and 84 announcements of stock issues by Dutch companies between 1 January 1977 and 31 December 1996. The table contains cumulative average abnormal returns for the period of days )1 and +1. Day 0 is the day on which the announcement is published in Het Financieele Dagblad. The abnormal returns are calculated using the standard event study methodology as outlined by Brown and Warner (1985). z-values are in parentheses, medians are in square brackets. Subsample results are reported for the nature of the other news reported together with the issue and the purpose of the issue. Results are not reported for cells smaller than 8. * Signicant at 10% level. ** Signicant at 5% level.
1887
purpose of the acquisition reveals that in 35 cases the company indicates that it will use (part of) the proceeds of the issue in order to nance acquisitions. These issues are associated with a non-signicant negative abnormal return of )0.22%. This result is remarkable if we look at the positive reaction on the simultaneous announcement of a completed acquisition and an equity issue. A further investigation of the 35 issues reveals that in 16 cases the company indicates that it wants to use at least a part of the proceeds for future acquisitions. The cumulative average abnormal return for these 16 cases is )2.72% with a t-value of )2.29. As this is a clear case of managerial discretion, this negative return can be interpreted as evidence for the moral hazard model. The otation method of the equity issue yields minor dierences between the subsamples. As explained before, this may be due to a lack of information. The information is either missing on the otation method at the announcement date of the issue or, in case of a rights issue, on the selling of rights in the future by existing shareholders. In order to test for other characteristics that inuence abnormal returns, we perform regressions in which the cumulative abnormal return is explained by rm and issue characteristics. As discussed in De Roon and Veld (1998), OLS regression are not likely to be ecient, because the abnormal returns are calculated as residuals from the market model. Therefore, the abnormal returns actually measure rm-specic risk, which diers between rms. We reduce the inuence of heteroskedasticity in order to improve the eciency of the estimates by applying weighted least squares regressions. We weigh by rm-specic risk, i.e. the dependent and independent variables are divided by the standard error from the market model in the estimation period. The regression results are included in Table 5. The determinants of abnormal returns associated with the debt issues are in column (1). The existence of slack leads to a signicantly positive coecient for debt. This is contrary to the hypothesis for the adverse selection model. The repayment of short-term debt gives a signicantly positive coecient. This is possibly caused by the fact that shareholders are able to judge the purpose of the issue. The abnormal returns associated with the equity issues in column (2) are signicantly positively inuenced by protability (return on assets), relative issue size and the announcement of a completed acquisition. The last mentioned eect is in accordance with the results presented in Table 4. In regressions (4) and (5) we add a dummy variable with a value of one if the rm has a positive free cash ow and a Tobin's Q below one, and zero otherwise. In Section 3.2 we already explained that this overinvestment dummy indicates that rms both have free cash ow and low growth opportunities. As expected, we nd that the overinvestment dummy is signicantly positive in debt issues and signicantly negative in equity issues. In columns (3) and (6) we report the results for a test of the dierences between the coecients of debt and equity issues. We compare, respectively, columns (1) and (2) and columns (4) and (5),
1888
Table 5 Weighted least squares regression results of rm and issue characteristics on abnormal return of debt and equity issuesa Debt (1) Intercept Return on assets Slack Past 12-month excess return Tobin's Q Free cash ow Issue size/total assets Log(issue size) Frequent issuer Acquisition announcement Future acquisitions Purpose is capital expenditures Purpose is repayment of short-term debt Towards target Predicted correctly Overinvestment dummy Adjusted R2 Number of observations
a
Equity (2) 0.019 (0.42) 0.003 (2.38) (2.38) )0.020 ()0.17) 0.034 (1.58) )0.031 ()0.95) )0.072 ()1.44) 0.190 (2.86) )0.006 ()1.01) 0.012 (0.46) 0.052 (2.61) )0.008 ()0.41) )0.008 ()0.41) )0.002 ()0.14) )0.027 ()1.11) 0.21 84
Dierence (1) vs. (2) (3) 0.024 (0.46) 0.005 (1.85) (1.85) )0.193 ()1.29) 0.029 (1.19) )0.048 ()1.39) )0.138 ()1.50) 0.374 (1.48) )0.007 ()0.73) 0.011 (0.39) )0.019 ()0.62) 0.009 (0.44) )0.026 ()0.89)
Debt (4) )0.037 ()1.20) )0.003 ()1.61) ()1.61) 0.185 (2.26) 0.008 (0.67) 0.029 (1.70) 0.002 (0.03) )0.257 ()1.49) 0.007 (1.01) )0.002 ()0.18) )0.001 (0.01) 0.050 (2.70) )0.013 ()1.43) )0.004 ()0.27) 0.022 (1.93) 0.12 98
Equity (5) 0.074 (1.36) 0.003 (2.29) (2.29) )0.058 ()0.51) 0.040 (1.87) )0.060 ()1.67) )0.032 ()0.59) 0.182 (2.78) )0.006 ()0.98) 0.005 (0.19) 0.054 (2.72) )0.005 ()0.26) )0.009 ()0.43) )0.007 ()0.41) )0.034 ()1.38) )0.035 ()1.93) 0.23 84
Dierence (4) vs. (5) (6) 0.110 (1.77) 0.006 (2.03) (2.03) )0.248 ()1.67) 0.033 (1.33) )0.088 ()2.36) )0.038 ()0.37) 0.439 (1.72) )0.012 ()1.14) 0.007 (0.25) )0.010 ()0.32) 0.006 (0.32) )0.030 ()1.04) )0.055 ()2.51)
)0.005 ()0.20) )0.002 ()1.23) ()1.23) 0.167 (2.03) 0.005 (0.44) 0.017 (1.04) 0.063 (1.11) )0.182 ()1.07) 0.002 (0.25) 0.001 (0.07) 0.008 (0.43) 0.048 (2.54) )0.011 ()1.20) )0.002 ()0.11) 0.09 98
The sample includes 98 announcements of bond issues and 84 announcements of stock issues by Dutch companies between 1 January 1977 and 31 December 1996. Day 0 is the day on which the announcement is published in Het Financieele Dagblad. The dependent variable is the cumulative abnormal return for announcement i for day )1 to day +1, which is calculated using the standard event study methodology as outlined in Brown and Warner (1985). Columns (1) and (4) are regression results for debt issues and columns (2) and (5) are regression results for equity issues. In columns (3) and (6) we compare, respectively columns (1) and (2) and columns (4) and (5), by regressing the abnormal returns on the explanatory variables and interaction terms of the
1889
Table 5 (Continued) explanatory variables and a dummy variable with the value of one for equity. In columns (3) and (6) we report the interaction terms. Return on assets is operating income over total assets. Total assets is measured in book values. Slack is cash and liquid assets over total assets plus unused debt capacity. The 12-month excess return is measured as the dierence between the stock return and the market return. Tobin's Q is the market value of equity plus the book value of debt divided by the replacement value of the assets. Free cash ow is operating income minus taxes, interest expenditures and dividends paid divided by the market value of equity. Frequent issuer is a dummy variable that has a value of one for rms that have more than 10 issues in the 20-year period and zero otherwise. The target ratio is dened as the target based on book values according to the crosssectional regression outlined in Appendix A. Information on whether the issue is predicted or not is derived from the logistic regression (Eq. (2) in Table 3). The overinvestment dummy has a value of one if free cash ow is positive and Tobin's Q is below one, and zero otherwise. t-statistics are in parentheses. * Signicant at 10% level. ** Signicant at 5% level. *** Signicant at 1% level.
by regressing the abnormal returns for the full sample on the explanatory variables and interaction terms of the explanatory variables and a dummy variable with the value of one for equity. The interaction terms measure the dierence of equity issues, relative to debt issues. We report the coecients and t-values of these interaction terms. An interpretation in terms of the capital structure models shows that the Jensen (1986) model seems to have explanatory power. For debt issues, we nd that a lack of growth opportunities and the existence of free cash ow is associated with signicantly positive abnormal returns (Hypothesis 8). For equity issues we see a signicantly negative coecient for the overinvestment dummy (Hypothesis 9). Both hypotheses can be interpreted as evidence for the moral hazard model. The dierence between the coecients is signicant at the 5% level. For equity issues we also see that the returns are positively aected if the rm announces the issue together with the completion of an acquisition. 34 The reason for this is that investors are able to judge the purpose of the issue. This purpose is less likely to be overinvestment, i.e. empire building. The use of funds for future acquisitions gives the expected negative sign, however this eect is not signicant (Hypothesis 10). The dierence in eects between announcing a completed acquisition and announcing that at least part of the funds will be used to nance a future acquisition can be considered as evidence for the moral hazard model. The announcement that the rm is going to use the funds for capital expenditures is not associated with a signicant abnormal return (Hypothesis 11). Finally we see, for debt issues, that if the rm
Abhyankar and Dunning (1999) study issues of convertible bonds and convertible preferred shares in the UK. The results in their paper also show that issues that are used to nance specic acquisitions are associated with less negative abnormal returns than the remainder of their sample.
34
1890
announces that it will use the proceeds to repay short-term debt, the announcement is associated with a signicantly positive abnormal return (Hypothesis 12). This result is contrary to our expectations. Following our expectations we do not nd a conrmation for the adverse selection model. Slack gives a signicantly positive coecient for debt while, according to the hypothesis for the adverse selection model, a negative sign would be expected (Hypothesis 13a). The role of growth opportunities contrasts with the prediction of the model, because the coecient for Tobin's Q is signicantly negative for equity issues. For debt issues, this coecient is signicantly positive, as expected in the model (Hypothesis 14a). Relative issue size gives an insignicant negative sign for debt. In contrast to the expectation for the adverse selection model it gives a positive signicant sign for equity (Hypothesis 15a). The insignicant sign for stock price run up for debt and equity and the dierence between debt and equity is associated with a rejection of the Lucas and McDonald (1990) model (Hypotheses 16a and 16b). 35 We have also tested the dierences between debt and equity issues (Hypotheses 13b16b). For slack, the dierence between debt and equity issues gives the expected negative sign. However, this is caused by the signicantly positive coecient for debt that is contrary to our hypothesis. Therefore, the hypothesis that slack aects abnormal returns more negatively for equity than for debt is rejected (Hypothesis 13b). The coecient for growth opportunities is signicant, but the sign is opposite to the hypothesized sign (Hypothesis 14b). 36 The hypothesis that relative issue size aects abnormal returns more negatively for equity than for debt issues is also rejected (Hypothesis 15b). Finally, the stock price run-up does not inuence debt issues dierently from equity issues (Hypothesis 16b). These results conrm our rejection of the adverse selection model for the Dutch situation. Table 4 already showed that rights issues are not associated with a higher abnormal return than underwritten oerings (Hypothesis 17). Therefore, according to our expectations for the Dutch market, we do not nd a conrmation for the adverse selection model. Finally we nd that our analysis does not provide a conrmation for the static trade-o model, since the dummy variable for movement towards target is insignicant for both equity and debt issues (Hypothesis 18).
35 Note that the positive coecient for the stock price run-up for equity issues conrms the model of Viswanath (1993). However, this coecient is only signicant in column (5) and is insignicant in column (2). 36 In line with most empirical studies we interpret Tobin's Q as a measure for growth opportunities. An alternative interpretation for Tobin's Q is as a measure for overvaluation. In such a case, the negative sign for Tobin's Q could be interpreted as evidence for the adverse selection model. However, we believe that the stock price run-up in the pre-event period is a better measure of overvaluation. This coecient is insignicant.
1891
5. Summary and conclusions A large amount of empirical evidence on capital structure theory is available in the nance literature. However, the evidence is largely limited to the US. In terms of managerial decision making, the US is characterized by managers who are strongly disciplined by the capital markets. If these managers do not succeed in maximizing shareholder value, their companies may be taken over by other companies. Such a situation does not exist in the Netherlands. In this country practically all companies have adopted multiple takeover barriers. These are primarily directed to limit the inuence of common shareholders. Besides that the Dutch governance structure is characterized by the presence of large shareholders. This situation makes the case for capital structure decisions of Dutch companies particularly interesting. We study two capital structure theories in detail: the moral hazard model and the adverse selection model. Besides that we use control hypotheses based on the static trade-o theory. The models are tested on a sample of 110 issues of public and private seasoned equity and 137 public issues of straight debt. We look at the motives for the company to choose equity or debt and we study the wealth eects of these security issues. The moral hazard model is relevant for Dutch rms, because managers are likely to have discretionary space to overinvest. This behavior will be reected in the debtequity choice, but the shareholders' valuation will be shown in the announcement eects. The model of Myers and Majluf (1984) assumes that managers act in the interest of existing shareholders. Because of the existence of entrenched managers and because of the dominant role of rights issues in the Netherlands, it can be expected that the adverse selection model will not be relevant for the Dutch situation. We do not nd evidence for the disciplining role of debt in the overinvestment theory. On the contrary, the results indicate that managers avoid debt when this is most disciplining. This occurs when rms either have free cash ows and low growth opportunities or when they have a low protability. As expected we do not nd evidence for the adverse selection model. Finally, we also nd evidence that Dutch companies issuing equity move towards a target equity ratio (based on book values). This can be considered as evidence for the static trade-o theory. A study of the wealth eects shows that the issuance of debt is associated with an insignicant return. This conrms empirical research for the US. An equity issue is associated with a signicant negative abnormal return of )1.07%. A further investigation of the determinants of the abnormal returns gives evidence for the moral hazard models. We do not nd evidence for the adverse selection model. A comparison of the motives for the debtequity choice and the market's reaction reveals interesting results. First, moral hazard behavior emerges in the
1892
debtequity choice as managers who are most likely to overinvest tend to avoid the disciplinary force of debt. This behavior is recognized by the market and the resulting stock price reactions reveal the market's suspicions of overinvestment behavior. Second, the evidence for the adverse selection model is absent in both the debtequity choice and the market's reaction.
Acknowledgements The authors are indebted to Rudolf Bosveld, Niklas Darijtschuk, Hans Degryse, Vihang Errunza, Piet Moerland, Steven Ongena, Peter Roosenboom, Janette Rutterford, Ronald van Dijk, Yulia Veld-Merkoulora, Mark Walker, and especially two anonymous referees for helpful comments and suggestions. The paper has also benetted from comments of participants at the Financial Management/European Financial Management Conference in Lisbon (June 1998), the Conference of the German Finance Association in Hamburg (September 1998), the Conference on Raising Capital in Dierent National Markets in Frankfurt (November 1998), the FinBeldag in Rotterdam (November 1998), the Southern Finance Association Meeting in Key West (November 1999) and of seminar participants at McGill University, Tilburg University and York University. Furthermore they are grateful to the Dutch Central Bureau of Statistics (CBS) for providing data and to Grzegorz Pawlina for his excellent research assistance. The usual disclaimer applies.
Appendix A. Determinants of the equity ratio Our data set contains annual report and stock market data of the Dutch listed non-nancial rms in the CBS database for 19761995 for which the variables in Table 6 are available. The variables in Table 6 are chosen with the importance of preventing distress and using tax benets in mind. The variable for tangibility of the assets is inversely related to bankruptcy costs. Unlevered beta measures business risk, which invokes bankruptcy risks. Size, i.e. the log of sales is inversely related to risk. For corporate tax reasons, depreciation can be deducted from prots. Therefore depreciation can be considered to be a non-debt tax shield. We perform OLS regressions with dummies for the years and industries in order to take time-specic and industry-specic elements (tax, regulation, macroeconomic factors, etc.) into account. The dummies are not reported. The t-statistics are heteroscedasticity-consistent. The results are in Table 7. The results in Table 7 conrm the impact of distress. Tangible assets, business risk and size are signicantly related to the equity ratio, with a sign as
A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895 Table 6 Equity ratio and determinants EQ_BV EQ_MV TGA_BV TGA_MV BETA_UNL LN_SLS DEP_BV DEP_MV
1893
Equity ratio in book values: book value of equity divided by book value of total assets Equity ratio in market values: market value of equity divided by market value of total assets Tangible, xed assets plus inventory, divided by book value of total assets Tangible, xed assets plus inventory, divided by market value of total assets Unlevered beta based on weekly returns and Datastream index for the Dutch market Log of annual sales Depreciation divided by book value of total assets Depreciation divided by market value of total assets
Table 7 Pooled regression results (t-values in parentheses) EQ_BV CONSTANT TGA_BV TGA_MV BETA_UNL LN_SLS DEP_BV DEP_MV Adjusted R Observations
2
EQ_MV (17.47) ()5.05) (8.46) ()10.81) (7.09) 0.4107 )0.2742 0.3984 )0.0207 0.3957 0.62 2405 (22.21) ()17.01) (18.80) ()13.75) (3.97)
0.3962 )0.1186 0.1994 )0.0239 0.9852 0.21 2405
expected. Depreciation approximates the non-debt tax shields. With the coefcients in Table 7 we estimate the target equity ratio, by multiplying the coecients with the values for the variables in the year preceding the issue.
References
Abhyankar, A., Dunning, A., 1999. Wealth eects of convertible bond and convertible preference share issues: An empirical analysis of the UK market. Journal of Banking and Finance, 1043 1066. Asquith, P., Mullins, D.W., 1986. Seasoned equity oers. Journal of Financial Economics, 6189. Bayless, M.E., Chaplinsky, S., 1991. Expectations of security type and the information content of debt and equity oers. Journal of Financial Intermediation, 195214. Berger, P.G., Ofek, E., Yermack, D.L., 1997. Managerial entrenchment and capital structure decisions. The Journal of Finance, 14111438. Bodnar, G.M., Tang, C., Weintrop, J., 1999. Both sides of corporate diversication: The value impacts of geographic and industrial diversication, Working Paper, Johns Hopkins University, Baltimore, MD.
1894
Bhren, ., Eckbo, B.E., Michalsen, D., 1997. Why underwrite rights oerings? Some new evidence. Journal of Financial Economics, 223261. Brealey, R.A., Myers, S.C., 2000. Principles of Corporate Finance, sixth ed. Irwin/McGraw-Hill, Boston. Brown, S.J., Warner, J.B., 1985. Using daily stock returns: The case of event studies. Journal of Financial Economics, 331. Cools, K., 1993. Capital structure choice; confronting: (Meta) theory, empirical tests and executive opinion, Ph.D. Dissertation. Tilburg, The Netherlands. Cooney, J.W., Kalay, A., 1993. Positive information from equity issue announcements. Journal of Financial Economics, 149172. Corhay, A., Tourani Rad, A., 2000. International acquisitions and shareholder wealth: Evidence from the Netherlands, International Review of Financial Analysis, 163174. Dann, L.Y., Mikkelson, W.H., 1984. Convertible debt issuance capital structure change and nancing-related information: Some new evidence. Journal of Financial Economics, 157186. De Jong, A., Van Dijk, R., 2001. Determinants of leverage and agency problems, Working Paper, Tilburg University. Denis, D.J., 1994. Investment opportunities and the market reaction to equity oerings. Journal of Financial and Quantitative Analysis, 159177. De Roon, F., Veld, C., 1998. Announcement eects of convertible bond loans and warrant-bond loans: An empirical analysis for the Dutch market. Journal of Banking and Finance, 14811506. Eckbo, B.E., Masulis, R.W., 1992. Adverse selection and the rights oer paradox. Journal of Financial Economics, 293332. Eckbo, B.E., Masulis, R.W., 1995. Seasoned equity oerings: A survey. In: Jarrow, R., Maksimovic, V., Ziemba, W.T. (Eds.), Finance (Series of Handbooks in Operations Research and Management Science). ElsevierNorth Holland, Amsterdam, pp. 10171072. Eun, C.S., Kolodny, R., Scheraga, C., 1996. Cross-border acquisitions and shareholder wealth: Tests of the synergy and internalization hypotheses. Journal of Banking and Finance, 1559 1582. Jensen, M.C., 1986. Agency costs of free cash ow, corporate nance, and take-overs. American Economic Review, 323329. Jung, K., Kim, Y., Stulz, R., 1996. Timing, investment opportunities, managerial discretion, and the security issue design. Journal of Financial Economics, 159185. Kabir, R., Cantrijn, D., Jeunink, A., 1997. Takeover defenses, ownership structure and stock returns in the Netherlands: An empirical analysis. Strategic Management Journal, 97109. Kang, J.K., 1993. The international market for corporate control: Mergers and acquisitions of US rms by Japanese rms. Journal of Financial Economics, 345371. Kang, J.K., Kim, Y.C., Park, K.J., Stulz, R.M., 1995. An analysis of the wealth eects of Japanese oshore dollar-denominated convertible and warrant bond issues. Journal of Financial and Quantitative Analysis, 257270. Kang, J.K., Stulz, R.M., 1996. How dierent is Japanese corporate nance. An investigation of the information content of new security issues. Review of Financial Studies, 109139. Krasker, W.S., 1986. Stock price movements in response to stock issues under asymmetric information. The Journal of Finance, 93105. Porta, R., Lopez-de-Silanes, F., Shleifer, A., 1999. Corporate ownership around the world. The Journal of Finance, 471518. Lehn, K., Poulsen, A., 1989. Free cash ow and stockholder gains in going private transactions. The Journal of Finance, 771787. Lucas, D.J., McDonald, R.L., 1990. Equity issues and stock price dynamics. The Journal of Finance, 10191043. MacKie-Mason, J.K., 1990. Do taxes aect corporate nancing decisions. The Journal of Finance, 14711493.
1895
Marsh, P., 1982. The choice between equity and debt: An empirical study. The Journal of Finance, 121144. McConnell, J.J., Muscarella, C.J., 1985. Corporate capital expenditure decisions and the market value of the rm. Journal of Financial Economics, 399422. Mikkelson, W.H., Partch, M.M., 1986. Valuation eects of security oerings and the issuance process. Journal of Financial Economics, 3160. Modigliani, F., Miller, M., 1958. The cost of capital corporation nance and the theory of investment. American Economic Review, 261297. Moerland, P.W., 1995. Alternative disciplinary mechanisms in dierent corporate systems. Journal of Economic Behavior and Organization, 1734. Myers, S.C., Majluf, N.S., 1984. Corporate nancing and investment decisions when rms have information that investors do not have. Journal of Financial Economics, 187221. Opler, T., Titman, S., 1997. The debtequity choice. Working Paper No. 97-9. Fisher College of Business, The Ohio State University, Columbus, OH. Perfect, S.B., Wiles, K.W., 1994. Alternative constructions of Tobin's Q: An empirical comparison. Journal of Empirical Finance, 313341. Pettway, R.H., Sicherman, N.W., Spiess, D.K., 1993. Japanese foreign direct investment: Wealth eects from purchases and sales of US assets. Financial Management, 8295. Pilotte, E., 1992. Growth opportunities and the stock price response to new nancing. Journal of Business, 371394. Ross, S.A., 1977. The determination of nancial structure: The incentive-signalling approach. The Bell Journal of Economics, 2340. Smith, C.W., 1986. Investment banking and the capital acquisition process. Journal of Financial Economics, 329. Stulz, R.M., 1990. Managerial discretion and optimal nancing policies. Journal of Financial Economics, 327. Viswanath, P.V., 1993. Strategic considerations, the pecking order hypothesis, and market reactions to equity nancing. Journal of Financial and Quantitative Analysis, 213234. Zwiebel, J., 1996. Dynamic capital structure under managerial entrenchment. American Economic Review, 11971215.

Capital Structure Decision

Încărcat de

Informații document

Descriere originală:

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Capital Structure Decision

Încărcat de

Drepturi de autor:

Formate disponibile

Journal of Banking & Finance 25 (2001) 18571895 www.elsevier.

An empirical analysis of incremental capital structure decisions under managerial entrenchment

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

1864 A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

H3: Debt H4: Equity

H5: Debt H6: Equity

H15a: ) H16a: ) H17: +

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

H7: Equity H18: + H18: +

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

1880 A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

0.026 0.076 0.005 )0.068 0.079 0.080 8.900 0.061 0.044

0.058 0.055 0.012 )0.066 0.085 0.043 7.007 0.071 )0.008

0.045 0.068 0.006 )0.063 0.065 0.027 8.700 0.046 0.050

)0.040 0.008 )0.015 0.012 )0.005 0.031 2.837 0.002 0.057

0.002 0.618 0.082 0.334 0.729 0.111 0.001 0.798 0.015

4.317 0.506 0.716 0.213 1.706 3.745 0.116 1.943 0.681

0.038 0.477 0.398 0.644 0.192 0.053 0.733 0.163 0.409

5.136 0.015 0.923 12,030

4.126 0.096 1.091 4626

4.024 0.061 0.979 1126

0.958 )0.072 )0.048 13,741

0.000 0.000 0.382 0.000

31.093 57.920 1.429 79.295

0.000 0.000 0.232 0.000

Average equity issuing rm

Average dierence debt and equity issuing rm )0.207 0.023

Percentage positive returns debt issuers

Percentage positive returns equity issuers 76.4% 45.3%

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

Past 12-months excess return Post 12-months excess return

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

Contaminated Annual earnings Current earnings Acquisitions Uncontaminated

Purpose Acquisition Capital expenditures Repay short-term debt No purpose

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895

0.3962 )0.1186 0.1994 )0.0239 0.9852 0.21 2405

A. de Jong, C. Veld / Journal of Banking & Finance 25 (2001) 18571895