Value Creation from Asset Sales: New Evidence from Bond and Stock Markets *

Transcription

1 Value Creation from Asset Sales: New Evidence from Bond and Stock Markets * Matthew J Clayton Kelley School of Business Indiana University 1309 East Tenth Street Bloomington, IN (812) mjclayto@indiana.edu and Natalia Reisel Edwin L. Cox School of Business Southern Methodist University PO Box Dallas, TX (214) nreisel@smu.edu January 2010 * We thank David Mauer, and seminar participants at University of Colorado, University of Virginia, and Indiana University for valuable comments. The index returns are kindly provided by William Maxwell. We thank Miji Lee for research assistance. The usual disclaimer applies.

2 Value Creation from Asset Sales: New Evidence from Bond and Stock Markets Abstract This paper posits that asset sales create value by allowing a firm to adjust its capital structure after it has become over-leveraged. We than examine this hypothesis by analyzing wealth effects of the asset sales on both equity and debtholders over a period from 1990 to Consistent with existing literature, we find that equityholders experience positive announcement period returns. We show, however, that significantly positive equity returns are concentrated in the sub-sample of highlyleveraged firms. Further, we find that highly-leveraged firms also experience positive announcement period bond returns. The low-leverage firms show no consistent significant excess equity or bond returns. Additionally, highly-leveraged firms have significantly higher equity and bond returns when proceeds from assets sales are used to retire debt. When proceeds are not paid out the firm value may be destroyed. Overall this evidence suggests that altering capital structure to lower leverage is a significant source of value creation from asset sales.

3 1. Introduction Asset sales are a frequent and important form of corporate reorganization. Empirical research has documented that asset sales create value for equityholders by consistently showing positive mean excess equity returns on the announcement of voluntary assets sales. 1 The mechanism of the value creation from asset sales, however, is not fully understood. The existing literature has suggested several possible channels though which assets sales can create value for equityholders: efficient redistribution of assets to value-maximizing users (neoclassical explanation), increase in focus, and wealth transfer from bondholders. 2 In this paper, we emphasize a capital structure adjustment as an alternative and important means by which asset sales create value and then examine the cross-section of the wealth effects of assets sales on both equity and debtholders to test if changes in capital structure are significantly related to excess returns from asset sales. Specifically, we argue that asset sales create value by altering the capital structure towards the optimal level. Following the establishment of an optimal capital structure, if the value of assets in place deteriorates, the firm will have excess leverage. By selling assets and retiring debt, the firm adjusts towards its optimal capital structure which will increase firm value. A simple model is provided that illustrates these ideas. The model predicts asset sales will create value for both debt and equityholders when the firm has high leverage. Further, the model highlights the importance of the use of proceeds from asset sales. 3 Value is created when the firm uses the proceeds to retire debt because the debt payout insures that the manager acts in the interest of other stakeholders. When the proceeds are not paid-out both equity and bond value may be destroyed 1 See Hite, Owers, and Rogers (1987), Jain (1985), Alexander, Benson, and Kampmeyer (1984), Lang, Poulson, and Stulz (1995), and John and Ofek (1995). 2 See Section 2 for a more detailed discussion of the literature. 3 Asset sales often result in significant cash remuneration as documented by this and earlier studies (see, for example, Bates (2006)). 1

4 because managers may misuse cash generated by asset sales to pursue their own objectives (Lang, Poulsen, and Stulz (1995), and Bates (2006)). The predictions of the model are investigated in a sample of large asset sales transactions conducted between 1990 and Unlike previous studies that focus on equity returns, we analyze cross-sectional excess returns of both bondholders and shareholders. There is limited evidence on the wealth effects of asset sales on bondholders although existing theories differ in their predictions. 4 The results show that highly leveraged firms, proxied by firms with market leverage above sample median, experience significant excess returns in the month of the announcement of the asset sales. The returns are positive for both equityholders and debtholders. Mean excess bondholder return for highly leveraged firms is 50 basis points and mean excess equity return is 2.07 percent. 5 These findings are confirmed when non-investment grade bond (junk) status is used to capture highly leveraged firms: mean excess junk bonds return is 99 basis points and mean excess equity return for firms with junk bonds outstanding is 4.00 percent. In contrast, mean excess bondholder and shareholder returns for firms with low leverage are not significantly different from zero. The strong effect of leverage on excess returns remains after controlling for focus increasing transactions, bond covenant structure, managerial ownership structure, industry structures as well as firm financial characteristics such as growth opportunities, liquidity, profitability and size. Further, the empirical analysis demonstrates the importance of the use of proceeds in explaining value creation from asset sales. The firms choose either to pay proceeds to 4 Exceptions are Datta, Iskandar-Datta, Raman (2003) and Datta and Iskandar-Datta (1996). 5 We use two approaches to capture bond excess return; see p. 8 for details. 2

5 stakeholders, retain the proceeds in the firm, or not to announce the use of proceeds. Our findings suggest that significant positive bondholder and equityholders excess returns are concentrated in the sample of firms that use the proceeds to retire debt. These results are robust to the notion that the use of proceeds is unlikely to be assigned randomly across firms but rather is endogenously determined as a function of firm characteristics. Further, we find that, after correcting for endogeneity, not-announcing the use of proceeds results in significantly negative bondholder and equityholder monthly excess returns. Additionally, we find that mainly debt payout by highly leveraged firms results in positive excess returns. Collectively, this evidence suggests that the adjustment of capital structure is a significant determinate of the value created from asset sales. This sheds new light on firms motives for conducting value enhancing asset sales. The previous research mainly attributed value creation from assets sales to redistribution to a more efficient use (neoclassical explanation). In contrast, we do not find that neoclassical model alone or wealth transfer between bondholders and shareholders explains value creations from assets sales. The results also suggest that agency cost of managerial discretion can be an importance source of cross-sectional variation in value creation from asset sales, consistent with Lang, Poulsen, and Stulz (1995). Further, while traditionally the literature views security issuance as the mechanism of capital structure adjustment, this paper shows that sale of assets is another important way to adjust the capital structure for highly leveraged firms. The remainder of the paper is organized as follows. Section 2 discusses existing motives for asset sales and their effect on both equity and bond values. Section 3 describes a simple model of asset sales based on capital structure considerations and agency costs and discusses the empirical implications. The data selection criteria and process is described in Section 4. Section 5 3

6 develops the empirical analysis and presents the results. Section 6 concludes the paper and summarizes the main findings. 2. Discussion of Existing Motives for Asset Sales and Wealth Effects Existing motives for assets sales can be grouped into two major categories: Efficiencybased and agency-based. Efficiency based explanations suggest that asset sales are mainly undertaken to increase firm value; while agency based explanations suggest that asset sales are undertaken to transfer wealth among different parties such as equityholders, bondholders and managers Efficiency based explanations for asset sales Neo-classical explanation Asset sales may be undertaken to redistribute assets to a more efficient use (this explanation can be traced back to Lucas (1978) paper on the optimal firm size). 6 Under this theory firms always act optimally and will sell an asset when it is more productive (and thus more valuable) to another user. The seller can capture part of the additional value generated from the redeployment of the asset in the sale price thus increasing firm value. Additionally, cash generated from the asset sale may be used to finance unfunded positive NPV project. If the firm was capital constrained, this further increases firm value and an increase in equity valuation should be observed. 6 More recent papers, for example, include Maksimovic and Phillips (2002), Warusawitharana (2007), and Yang (2008). 4

7 Bonds have priority over equity for firm cash flows. Ceteris paribus an increase in firm value should either increase bond values or leave bond values unchanged. 7 If the debt level is low relative to firm cash flows so that the bond payments are secure, a further increase in firm value will not have a large effect on bond values. If the payment of the bond cash flow is in doubt, then an increase in firm value should increase the bond value Increase in Focus Another possible source of value creation from assets sales is due to increase in focus (see John and Ofek(1995)). Under this theory firms act to undo previous inefficient investments (e.g. inefficient diversification). Through the sale of unrelated assets the firm can eliminate negative synergies and allow management to operate core assets more efficiently. Assuming the assets are sold in an efficient market, the increase in operating efficiency of the remaining assets will increase firm value and both equityholders and bondholders can benefit Agency based explanations for asset sale Agency Costs of Managerial Discretion Asset sales may be used to generate funds allocated at the managerial discretion (Lang, Poulsen, and Stultz (1995)). Under this theory, managers value control and firm size above maximizing stakeholder wealth. Asset sales will only occur when they are the cheapest source of financing available to carry out managers own objectives which are not necessarily aligned with the objectives of equityholder or bondholders. A successful asset sale is good news, because it signals positive information about the value of the asset (if the asset turns out to be less valuable than expected or if the firm must take a large discount on the asset to facilitate a sale it is less likely that the asset sale will be the cheapest source of funds). However, this good news is 7 This assumes no significant increase in the risk of the assets. Selling assets for cash should not generally result in an increase in risk. It is possible, if the cash is reinvested in alternative assets, these assets could increase the risk of the firm relative to the assets which were sold (assets substitution). 5

8 moderated by the agency costs of free cash flow. Under this theory, the change in value from asset sales depends on the use of proceeds. Only when proceeds are paid out (to equityholders or bondholders), is an increase in equity value observed because the agency costs are avoided. If the proceeds are paid out to debtholders, then debt value may increase because overall firm leverage decreases Shareholder-bondholder conflict Asset sales may be used to transfer funds from bondholders to shareholders. Bondholders may experience a decrease in the value due to the loss of the collateral, co-insurance effect and risk-shifting. The possible mechanisms of wealth transfer are the payout of the proceeds to equity holders, increase in focus, and re-investing proceeds from the asset sales in higher risk projects. When debt is risky, the equity can be thought of as a call option on the firm s assets. Using proceeds from asset sales to invest in higher risk projects may increase the risk of the firm s assets and, hence, the value of the call option. Debt is on the other side of the call option, and the increase in risk lowers the value of the debt. Bondholders recognize potential for the shareholder-bondholder conflict and include restrictive covenants in the debt contracts which often require the pay-out of the proceeds from asset sales to bondholders (see, for example, Smith and Warner (1979)). Some firm actions, on the other hand, may lower the firm risk benefiting bondholders at the expense of shareholders. Amihud and Lev (1981), Eckbo and Thorburn (2003) provide evidence consistent with managerial risk-aversion. 3. Implications of the Capital Structure and Agency Cost Model of Asset Sales 6

9 In this section, we present a simply model to direct our empirical analysis. We build on the insights provided by previous research: 1) we recognize in line with the efficiency based explanations that asset sales may allow firms to take advantage of positive NPV projects; 2) we also, however, recognize that managers interests are not always aligned with the interests of stakeholders in line with the agency based explanations for assets sales. The formal mode is developed in Appendix A. We demonstrate how asset sales can arise to adjust capital structure and the resulting effects on debt and equity values. Debt arises endogenously in the model to restrict empire building managers from access to free cash flow. 8 This prevents managers from using cash flows for negative NPV projects when positive NPV projects are unavailable. Since leverage forces the payout of firm cash flows, if a positive NPV project becomes available, the firm must raise capital to fund the project. Once an optimal level of leverage is established, if there is a deterioration in expected future cash flows the firm will be over levered. Too much leverage can decrease firm value in various ways. For instance, debt overhang may leave a firm unable to raise funds even for investment in positive NPV projects. This leads to the benefit of an asset sale (before the firm is in financial distress). There are several possible methods by which the firm can alleviate the under-investment problem through an asset sale. For example, the firms may use the proceeds to retire debt and lower the leverage. This will increase firm value by mitigating the problems associated with excess leverage. Alternatively, the firm could retain the cash from the asset sale that could be used to finance positive NPV project. In the latter case, however, the empire building manager 8 While this is a common rational in the finance literature for leverage, there are several other commonly accepted rationales for why firms take on debt. For example, to take advantage of tax shields, or due to asymmetric information about asset value, or even as a commitment to another strategic decision. The qualitative results with respect to asset sales in the model are not dependent on why the firm chooses initial leverage. 7

10 may use the cash to undertake a negative NPV project if the positive NPV project does not materialize. This highlights why debt payout is important. There may be other methods to mitigate the value decrease caused by excess leverage. For instance, if under investment is the cost of too much leverage, why can t the firm just wait to see if it has a positive NPV project and then use the asset sale to generate cash to take the project directly? The formal model demonstrates circumstances where it is necessary for the firm to adjust its capital structure before debt overhang becomes a problem: If it waits until it is in a debt overhand situation an asset sale may not generate enough capital to take the positive NPV project. There are several empirical implications of the capital structure model of asset sales. First, only firms with excess leverage will create value from the asset sales. Second, this value will be created when the proceeds are used to retire existing debt. Third, value will be created for both debt and equityholders. 4. Data and Descriptive Statistics Asset sales are collected from the Securities Data Corporation (SDC) mergers and acquisitions database with initial announcement date from and transaction values of at least 75 million US dollars. The SDC data includes the announcement date, company identification number (cusip), transaction value, SIC code of the division of the asset sales, SIC code of the selling company, and type of remuneration. The initial sample contained 4,318 asset sales announcements. The sample is further limited to public industrial companies and requires non-missing total assets and long-term debt data on Compustat. The availability of bond data is also required. The bond data is compiled from three sources: Lehman Brother Bond Database 8

11 (LBBD), National Association of Insurance Companies Database (NAIC) and Fixed Income Securities Database (FISD). LBBD reports month-end bid quotes and matrix prices from Lehman Brother s dealers through 1997 as well as monthly bond returns. There are well known issues with using matrix prices for event studies (see Warga and Welch, 1993), thus only monthly bond returns based entirely on quote prices are used. Since LBBD is only available through 1997, NAIC is used to compute monthly bond returns from 1998 to NAIC reports transaction prices provided by insurance companies. Both LBBD and NAIC have been widely used in empirical studies to determine abnormal bond returns. Bond characteristics are collected from FISD. FISD reports a wide-range of issue-specific information including issuer, maturity date, call date, coupon, bond rating, and seniority level as well as bond specific features such as put and convertible. All convertible and putable bonds are excluded from the sample. 9 FISD also provides information about asset sales covenants associated with a bond issue. 10 Lexis-Nexis and Factiva are used to verify information on each transaction and collect data on the use of proceeds from the asset sales. Use of proceeds is classified following Lang, Poulsen and Stulz (1995). In addition, asset sales from financial distressed firms are classified similar to Brown, James, and Mooradian (1994) and eliminated. A total of 12 transactions are excluded due to either inconsistent data or the fact that the asset sale was associated with a firm in financial distress. Standard & Poor s Executive Compensation database is used to collect managerial ownership data and equity data is collected from CRSP. The final sample consists of 439 asset sales and 1,527 bond issues. 9 It is also verified that none of the bonds matured or were called during the month of the asset sales transaction. 10 In some case, covenant data is unavailable for the bonds in our sample. For example, FISD did not report covenant data for issues that lack subsequent data from prospectuses and other more detailed documents (see Reisel, 2008) for details about covenant data in FISD). The quality of the covenant data is also lower before 1989 then onward that is why we start our sample in

12 Table 1 provides descriptive statistics for firms, transactions, and bonds in the sample. The typical sample firm is large with mean total assets of nearly $32 billion (median is nearly $14 billion). The mean transaction value is $ million (median is $ million). Values are reported in 2004 dollars using the Consumer Price Index to adjust for inflation. The majority of the asset sales, 70%, are outside the main line of business of the selling company (4-digit SIC code of the division sold is not equal the SIC code of the selling company). The remuneration is primary cash; about 81% of the transactions are 100% cash. In 21% percent of the transactions the proceeds are paid out to debtholders, in only 2% of the cases are the proceeds paid to equityholders. The proceeds are either retained by the firm or no use of proceeds is announced in 77% of the transactions. We show later in the paper the firms that do not announce the use of proceeds behave similarly to the retention sub-sample on several dimensions. Panel B provides characteristics of the bonds in the sample. The mean remaining maturity is about 13 years (median is 8.7). Majority of the bonds are investment grade and senior unsecured, 84% and 94% respectively. The bonds with asset sales covenants are very common, only 9% of the bonds do not have an asset sales covenant. 5. Leverage and Excess Bond and Stock Returns To test the predictions of the capital structure model of asset sales, we perform crosssectional analysis of bond and stock returns. We first focus on the effect of leverage and then consider the use of proceeds. The LBBD database contains only monthly data, thus the analysis is limited to examining excess bond and stock returns for the month of the asset sale. The total bond return in LBBD is the change in price plus the accrued interest. Similarly monthly returns for the NAIC bond data are calculated using last available price for a month. 10

13 Excess bond return is calculated as the total return minus a value-weighted bond index return with similar rating and remaining maturity. 11 This approach is used by Warga and Welch (1993) and Billett, King, and Mauer (2004) and shown to be the best specified and most powerful approach by Kahle, Maxwell, and Xu (2007). There are six major rating categories (AAA, AA, A, BBB, BB, B) and three maturity categories. 12 Firms with multiple bonds outstanding are handled in two ways. First, each bond is treated as a separate observation. Since the assumption of independence of the observations may be violated in this case, cluster adjusting or bootstrapping is used to calculate the appropriate statistics. This approach over-weights firms with multiple bonds outstanding (these are typically large, highly rated firms) but allows analysis of excess returns across bond characteristic. Second, equally-weighted portfolios are constructed and the portfolio excess return is examined. This creates only one excess bond return per asset sale transaction. The equity abnormal return is calculated as the monthly holding period return less the monthly return on CRSP value-weighted market portfolio Univariate analysis of excess returns Results for the univariate analysis of excess returns at the individual bond level, bond portfolio level, and for equity are reported in Table 2. The mean excess equity return in the full sample is 0.99% and significant at the 10% level. In case of individual bonds, the standard errors are adjusted for possible heteroskedasticity and serial correlation by clustering at the transaction level (see, for example, Williams (2000)). The magnitude of the equity excess return is consistent with prior research on asset sales. This suggests that on average asset sales create value for 11 The index returns are kindly provided by William Maxwell. 12 AAA bonds are only divided into two duration categories, less than or more than 7 years due to a small number of AAA bonds available. 11

14 equityholders. Mean bond excess return for the full sample (at the individual bond level) is positive but not statistically significant. Mean bond portfolio return is 0.23% and significant at the 10% level. This provides some evidence that assets sales also create value for debtholders. The significance of the bond portfolio returns is likely to be driven by low-graded bonds as they have higher weight in the portfolio sample than in the bond sample. All the median excess returns in the full sample are positive but not significant suggesting that only subset of firms experience increase in value from the asset sales consistent with our model. In case of individual bonds, standard errors are calculated using bootstrapping with 1,000 replications. The capital structure theory predicts that firms with too much leverage create value from asset sales transactions and both equity and bonds display positive excess returns. We use two proxies for high leverage. First, we consider firms with market leverage above the sample median. Some firms may hold high levels of leverage optimally, thus not all firms in the above median leverage sample are necessarily over-levered. Any misclassifications will decrease the power of the test. Additionally, we rely on credit ratings. Firms with non-investment grade rating (junk) are likely to be over-leveraged (see, for example, Fabozzi and Fobozzi (1995)), and the agency cost of debt are likely to be high in these firms. We classify a firm as a junk grade if all of its bonds in our sample are junk grade. Table 2 shows that positive equity and bond excess returns are concentrated in the subsample of high-leverage firms. Both means and medians excess bond returns are positive and statistically significant in the sub-sample of firms with leverage above the median. Mean bond portfolio return, for example, is 50 basis points and significant at 5% level; the median is 14 basis points and significant at 5% level. The magnitude of the returns is highly economically significant. In contrast, both mean and median excess bond returns are negative and generally 12

15 statistically insignificant in the sub-sample of firms with leverage below median. The results also suggest that only non-investment grade firms experience significantly positive mean bond excess returns. Further, difference in means (medians) tests show that the high-leverage samples (using both proxies) has significantly higher mean and median excess bond returns than the low-leverage samples. Firms with high leverage also show positive equity excess returns. For the noninvestment grade sample the mean (median) abnormal equity return is 4.00% (3.49%) and is significant at the 1% (5%) level. The mean excess equity return for the above median leverage sample is 2.07% and is significant at the 5% level. The median is 1.50% but not significant. In contrast, mean and median equity excess returns for low leverage categories (for both proxies) are close to zero and insignificant. Overall, the impact of leverage and credit rating are consistent with the capital structure model of asset sales. Both equityholders and debtholders of firms with above median leverage or junk grade debt experience positive excess returns from asset sales. We do not find evidence of wealth transfer from bondholders to shareholder as the excess returns are highly positively correlated. Also, the evidence that positive excess returns are mainly concentrated in the subsample of firms with high-leverage suggests that neoclassical explanation alone cannot explain value creation from assets sales Multivariate analysis of excess returns This section investigates whether the impact of leverage on bondholder and equityholder abnormal returns remains after controlling for other factors that may potential affect these returns. 13

16 Several regression specifications are used to explain the abnormal bond and stock returns, results are presented in Table 3. John and Ofek (1995) suggest value is created from asset sales through an increase in focus. Similar to John and Ofek we construct a dummy variable that takes the value of 1 if the 4- digit SIC code for the division being sold is the same as the firm s 4-digit SIC code. These transactions are non-focus increasing. Specifications (1) (2) report the effect of junk status and market leverage on bond, bond portfolio and equity excess returns after controlling for focus increasing transactions. These specifications for the bond sample additionally include an asset sale covenant dummy variable, which equals one if no asset sales covenants are included in the bond contract and zero otherwise. All specifications also include time period dummy variables. Consistent with the univariate results, we find that junk status and market leverage have significant positive effect on both bond and stock excess returns. For example, the coefficient on leverage is 2.50 in the bond portfolio sample and significant at 1% level. The coefficient on leverage in the equity sample is 8.26 and significant at 1% level. Interestingly, the constant terms in the bond and bond portfolio samples are negative and significant suggesting that asset sales transactions by low leverage firms may decrease bondholder value. Efficiency based theories of asset sales suggest that firm size and profitability are likely to be related to the benefits of asset sales (for example, see Warusawitharana (2008)). Specification (3) controls for firm characteristics such as firm size, profitability, liquidity and market-to-book. These variables are related to the degree of financial constraint and agency problems and, thus, may impact the excess returns. We report results using market leverage only as junk bond status combines affect of different firm characteristics. Additionally, we include industry dummies, thus, high-leverage firms are identified relative to the industry benchmark. 14

17 The coefficients on leverage for both bond and equity excess returns remains positive and significant after controlling for these additional firm characteristics. For example, the coefficient on leverage is 3.13 in the bond sample and significant at 1% level. The coefficient on leverage in the equity sample is 8.13 and significant at 5% level. Among other variables only impact of firm size is significant firm size is negatively related to excess returns across individual bonds and portfolio sub-samples. Specification (4) in Table 3 additionally controls for the size of the asset sale transaction. We should note, however, that the size of the transaction may be endogenously determined as a function of firm characteristics. We this caveat in mind, we report results after controlling for the transaction size. We find that the main results on leverage are robust to this control. The last specification in Table 3 controls for the effect of managerial ownership. Agency based explanations for assets sales recognize that managers may undertake assets sales to achieve their own objectives at the expense of the stakeholders. Hence, the structure of managerial ownership may affect value creation from asset sales. To control for the managerial ownership structure, we include two variables: one captures managerial equityholdings (Stock), another captures managerial option holdings (Options). These variables are not available for all firms, which results in reduction in the sample size. Previous studies (e.g. Ortiz-Molina (2006)) document non-linear effect of managerial ownership on the value of debt that is why we also include quadratic terms. We continue to find that strong effect of leverage on bondholder and equityholder excess returns remains after controlling for managerial ownership. The results also support the argument that stakeholder are concerned that assets sales are motivated by potential opportunistic actions of the managers as the impact of managerial ownership variables, specifically option holding, on the excess returns is significant. 15

18 Overall, the results of the multivariate analysis suggest that the level of debt is a significant factor in explain excess returns from asset sales consistent with the capital structure theory of the asset sales presented in this paper. 6. Use of Proceeds from Asset Sales The capital structure theory suggests asset sales create value when the firm uses the asset sale to lower leverage. This would require the firm to use the proceeds to retire debt. Thus, payout to debtholders is important in explaining excess returns. We classify the use of proceeds from assets sales into four categories: Debt payout (21% of the transactions), equity payout (2% of the transactions), retention (46% of the transactions), and the last category includes transaction in which the firms do not announce the use of proceeds (31% of the transactions). Lang, Poulsen, and Stultz (1995) and Bates (2006) use similar classifications with one exception they do not analyze transactions when no use of proceeds is announced. Use of proceeds is collected from articles available during the announcement month of the asset sale. These are not legally binding statements and the firm is not obligated to retire debt even if it is announced to do so. Thus, we first investigate if firms that announce the proceeds will be used to retire debt actually retire debt. Table 4 reports changes in debt around the year of the asset sales transaction delineated by the use of proceeds. It is important to note significant differences between debt payout sub-sample and retention or not announced sub-samples. Consistent with the stated use of proceeds, debt payout firms reduce total debt, mainly through a reduction of short-term debt. As a result, the average book leverage declines by 7%. In contrast, firms which announce that the proceeds will 16

19 be retained or firms that do not announce the use of proceeds, on average, increase total debt, mainly through an increase in long-term debt. The book leverage increases in both cases by 2%, though, the means are not statistically significant. Further, the changes in the leverage ratios for these two sub-samples are significantly different from the debt payout sub-sample. The equity payout sub-sample is not significantly different from the debt payout sub-sample in terms of change in debt; however this may be due to the small number of equity payouts resulting in low statistical power for the test. To determine if use of proceeds is related to value creation from asset sales, we first perform univariate analysis and partition the sample based on the use of proceeds. Table 5 presents the results; panel A contains the whole sample. The significant excess returns are concentrated in the debt payout sample. Payout to debt yields significantly positive mean and median bond excess returns. For example, bond portfolios have a mean excess return of 0.77%. Mean and median equity excess returns are also positive and significant, 4.03% and 2.75% respectively. The retention sub-sample, equity payout sub-sample and sub-sample of firms that do not announce a use of proceeds (NA) show no significant mean/median bond or stock excess returns with one exception: median bond excess returns are significantly negative, -0.19%, in the equity payout sub-sample suggesting that bondholders are concerned with the wealth transfer to shareholders. One potential concern with the results is that the use of proceeds is not assigned randomly across firms but rather is endogenously determined as a function of firm characteristics. We investigate whether our findings of positive excess returns associated with the debt payout are robust to this concern in appendix B. We continue to find that positive excess bond and stock 17

20 returns are concentrated in the debt payout sub-sample. Interestingly, these results also indicate that not announcing the use of proceeds results in negative bond and equity excess returns Use of proceeds and leverage We now turn to analyzing the interactions between leverage and the use of proceeds. To this end, we first perform univariate analysis and then run multivariate regressions that include a product of leverage (or junk status) and debt payout. The capital structure model predicts the highly leveraged firms that pay-out proceeds to debtholders experience the highest excess returns. Panels B and C of Table 5 present the results across firms with leverage below and above median. The results suggest that positive excess returns associated with the debt payout are driven by highly leveraged firms. High leverage firms that payout out proceeds from assets sales to debtholders experience positive bond and equity excess returns. Both means and medians are statistically significant. Mean bond portfolio excess return, for example, is 1.08% (median is 0.66). Mean and median equity excess returns are 4.64% and 2.80% respectively. In contrast, debt payout by low leverage firms results in insignificant excess returns. The results also provide some evidence that retention by low-leverage firms may result in negative excess returns: median bond returns are negative and significant, This finding is consistent with the agency based explanations for asset sales. Panel D and E of Table 5 presents the results for the noninvestment grade bond sample and investment grade bond sample respectively. These results confirm that leverage and debt payout are important factors explaining value creation from assets sales. For junk-grade firms, mean (median) bond excess returns are 1.79% (0.85%) for individual bonds, and 2.21% (1.47%) for bond portfolios and 9.54% (8.20%) for equity. All excess returns are statistically significant. In 18

21 contrast, for investment grade firms, only median bond returns are significantly positive. Also the magnitudes of these returns are smaller compared to the junk-grade sub-sample, 0.28% and 0.26%. The results of the univariate analysis suggest that value creation from asset sales is concentrated in highly leveraged firms that use the proceeds to retire debt consistent with capital structure model of asset sales. The results of multivariate analysis are presented in Table 6. As before, we report bond, bond portfolio and equity excess returns. Specifications (1) include the interaction variable junk*debt payout, while specifications (2)-(3) include the interaction variable leverage*debt payout. Consistent with the capital structure model and univariate results, we find that coefficients on the interaction variables are positive and statistically significant in all specifications. For example, the coefficient on junk*debt payout in the bond portfolio sample is 2.29 and significant at 1% level. The coefficient on junk*debt payout in the equity sample is 7.73 and also significant at 1% level. These results further confirm that debt payout by highly leveraged firms increases excess returns generated from asset sales. Interestingly, the coefficient on the debt payout variable becomes insignificant or even negative after the interaction variables are included suggesting that debt payout by low leverage firms may result in lower excess returns from asset sales. 7. Summary This study provides insights on value creation from assets sales by analyzing crosssectional excess returns of both bondholder and shareholder using a comprehensive dataset. The theoretical and empirical analyses suggest that asset sales can create value when highly leveraged 19

22 firms use the proceeds to retire debt. In a sample of large asset sales transactions conducted from , we find that both excess stock and bond returns are concentrated in subsamples of firms with high leverage that use the proceeds to retire debt. These factors remain significant determinates of excess returns for both bonds and stocks after controlling for other factors that may create value such as increase in focus, debt covenant structure, managerial ownership structure and firm characteristics. Overall, the results suggest that a capital structure adjustment is an important determination of excess returns generated by asset sales. 20

23 References: Alexander, G.J., P.G. Benson, and J.K. Kampmeyer, 1984, Investigating the valuation effects of announcement of voluntary corporate selloffs, Journal of Finance 39: Bates, Thomas W., 2006, Asset sales investment opportunities, and the use of proceeds, Journal of Finance 60: Billett, Matthew T., Dolly King and David C. Mauer, 2004, Bondholder wealth effect in mergers and acquisitions: new evidence from the 1980s and 1990s, Journal of Finance 59: Brown, D., C. James, and R. Mooradian, 1994, Asset sales by financially distressed firms, Journal of Corporate Finance 1: Cleary, Sean, 1999, The relationship between firm investment and financial status, Journal of Finance, 54: Datta, S., and M. Iskandar-Datta, 1996, Who gains from corporate asset sales? Journal of Financial Research 19: Datta, S., M. Iskandar-Datta, and K. Raman, 2003, Value creation in corporate asset sales: the role of managerial performance and lender monitoring, Journal of Banking and Finance 27: Eckbo, B. E. and K. Thorburn, 2003, Control benefits and CEO discipline in automatic bankruptcy auctions, Journal of Financial Economics 69: Fabozzi, F., and T. Fabozzi, 1995, The handbook of fixed income securities, Irwin Professional Publishing, New York, New York. Faulkender, Michael, and Rong Wang, 2006, Corporate financial policy and cash holdings, Journal of Finance 61: Fazzari, Steven R., Glenn Hubbard, and Bruce Pettersen, 1988, Financing constraints and corporate investment, Brooking Papers on Economic Activity Fazzari, Steven, R., Glenn Hubbard, and Bruce Petersen, 2000, Investment-cash flow sensitivities are useful: a comment on Kaplan and Zingales, Quarterly Journal of Economics 115: Gamba, Andrea, and Alexander Triantis, 2008, The value of financial flexibility, Journal of Finance 63: Hadlock, C. and J. Pierce, 2008, Does the KZ index provide a useful measure of financial constraints? Unpublished Working Paper, Michigan State University. 21

24 Hart, O., 1991, Theories of optimal capital structure: a principle-agent perspective, Paper prepared for the Brookings Conference on Takeover, LBO s, and Changing, Corporate Forms Hart, O., and J. Moore, 1998, Default and renegotiation: a dynamic model of debt, The Quarterly Journal of Economics 113: Hite, Gailen L., James E. Owens and Ronald C. Rogers, 1987, The Market for interfirm asset sales: partial sell-offs and total liquidations, Journal of Financial Economics 18: Jain, Prem C., 1985, The effect of voluntary sell-off announcements on shareholder wealth, Journal of Finance 40: Jensen, Michael C., and William H. Meckling, 1976, Theory of the firm: managerial behavior, agency costs and ownership structure, Journal of Financial Economics 3: John, Kose and Eli Ofek, 1995, Asset sales and increased focus, Journal of Financial Economics 37: John, Teresa A., 1993, Optimality of spin-offs and allocation of debt, Journal of Financial and Quantitative Analysis 28: Kahle, Kathleen, William F. Maxwell, and Danielle Xu, 2007, Measuring abnormal bond performance, Unpublished Working Paper, University of Arizona. Kaplan, Steven and Michael S. Weisbach, 1992, The success of acquisitions: evidence from divestitures, Journal of Finance 47: Kaplan, Steven N., and Luigi Zingales, 1997, Do Investment-cash flow sensitivities provide useful measures of financing constraints? Quarterly Journal of Economics 112: Kaplan, Steven N., and Luigi Zingales, 2000, Investment-cash flow sensitivities are not valid measures of financial constraints, Quarterly Journal of Economics 115: King, Tao-Hsien Dolly, and David C. Mauer, 2000, Corporate call policy for nonconvertible bonds, Journal of Business 73: Lang, Larry, Annette Poulsen, and René Stultz, 1995, Asset sales, firm performance, and the agency costs of managerial discretion, Journal of Financial Economics 37:3-37. Lucas, R., 1978, On the size distribution of business firms, Bell Journal of Economics 9: Maksimovic, V., and G. Phillips, 2001, The market for corporate assets: who engages in mergers and asset sales and are there efficiency gains? Journal of Finance 56: Maksimovic, V., and G. Phillips, 2002, Do conglomerate firms allocate resources inefficiently across industries? theory and evidence, Journal of Finance 57:

25 Maxwell, William F. and Ramesh P. Rao, 2003, Do spin-offs expropriate wealth from bondholders, Journal of Finance 58: Maxwell, William F. and Clifford P. Stephens, 2003, The wealth effects of repurchases on bondholders, Journal of Finance 58: Ortiz-Molina, H., 2006, Top management incentives and the pricing of corporate public debt, Journal of Financial and Quantitative Analysis 41: Parrino, R. and M. Weisbach, 1999, Measuring investment distortions arising from stockholderbondholder conflicts. Journal of Financial Economics 53: Penas, Maria F. and Haluk Unal, 2002, Gains in bank mergers: evidence for the bond markets, Unpublished Working Paper, University of Maryland. Reisel, N., 2007, On the value of restrictive covenants: empirical investigation of public bond issues, Unpublished Working Paper, Southern Methodist University. Shultz, P., 2001, Corporate bond trading costs: a peak behind the curtain, Journal of Finance 46: Smith, C. and J. Warner, 1979, On financial contracting: an analysis of bond covenants, Journal of Financial Economics 7: Warga, A., and I. Welch, 1993, Bondholder losses in leverage buyouts, Review of Financial Studies 6: Warusawitharana, M., 2008, Corporate asset purchases and sales: theory and evidence, Journal of Financial Economics 87: Williams, R., 2000, A note on robust variance estimation for cluster-correlated data, Biometrics 56: Yang, L., 2008, The real determinants of asset sales, Journal of Finance 63:

26 Appendix A Capital structure and agency cost model of asset sales A.1 The model setup The model involves a firm which has assets in place at date 0. These assets will produce a cashflow in each of three future periods (date 1, date 2, and date 3). There are two possible states of the economy: prosperity or recession. The economy begins in prosperity at date 0. At date 1 the firm receives the prosperity cashflow, Y P. At this time if the economy stays in prosperity (probability δ) the firm will continue to receive the prosperity cashflow at date 2 and date 3. Alternatively, the economy can move to recession at date 1 (probability 1-δ). If this occurs the firm will receive the recession cashflow at date 2, Y R < Y P. At date 2 the economy can either recover (and move back to prosperity), or the recession can continue. The economy recovers with probability δ, and stays in recession with probability (1-δ). If the economy recovers at date 2, then the firm receives the prosperity cash flow, Y P at date 3. If the recession continues, the firm receives a lower recession cashflow at date 3, Y RR < Y R. There are two types of mangers, empire builders and non-empire builders. Manger type cannot be verifiably disclosed to investors. Non-empire building managers make decisions to maximize shareholder value. Empire builders will make decisions to maximize the value of real assets controlled by the firm. At date 0 the initial capital structure is established. The firm will have investment opportunities at date 1 and date 2. It is assumed there will always be a negative NPV investment available. At date 2 a positive NPV investment arises with probability. The date 1 negative NPV project has an initial investment cost of I 1 and produces a cashflow of R 2 at date 2, R 2 < I 1. The date 2 negative NPV project has an initial investment cost 24

27 of I 2 and produces a cashflow of R 3 at date 3, R 3 <I 2. The positive NPV project at date 2 has investment costs I + at date 2 and produces a cashflow of R + at date 3, R + > I +. A.2 Model analysis Following Hart (1991) and Shleifer and Vishny (1992) in the absence of state contingent contracts the firm must have short and long term debt to prelude empire building managers from taking the negative NPV projects. To prevent investment in negative NPV project during prosperity in period 1, debt levels much satisfy the following conditions. Condition 1: I 1 >Y P -D 1 : This condition insures that after paying the short term debt the firm does not have enough cash available to invest without accessing the capital markets. Condition 2: I 1 -(Y P -D 1 )<Y P +R 2 -D 2 Condition 3: I 1 -(Y P -D 1 )<Y P +R 2 -D 2 +Y P -D 3 Condition 2 and 3 ensure that long term senior debt is sufficient to preclude the firm from raising the additional capital needed to invest. Similar conditions are required to prelude value destroying investment at date 2 in prosperity. Condition 4: I 2 >Y P -D 2 +Y P -D 1 : This condition insures that the cash flow generated at date 1 and date 2 less the required debt payments is not sufficient to invest at date 2, and hence the firm must access the capital markets to invest. Condition 5: I 2 -(Y P -D 1 +Y P -D 2 )<Y P +R 3 -D 3 Together, the 5 conditions above establish minimum short and long term debt that must be outstanding at date 0 to insure managers cannot take value destroying investments at date 1 and date 2 in prosperity There is no cost of financial distress in this model. This means the firm will always have at least enough short and long term debt to restrict value destroying investment. If financial distress costs were added, the firm would still choose to have debt to restrict future investment as long as the value gained by precluding negative NPV investments is larger than the expected costs of financial distress. 25

28 A.2.1 Case of recession We now examine the interesting case when the economy is in recession at date 1. Under the following two conditions the firm will be unable to take the positive NPV project at date 2 if the recession continues, because of insufficient free cash flow and inability to raise new capital. Condition 6: I + >(Y P -D 1 )+Y R -D 2 : If condition 6 holds, this means the firm does not have enough free cash flow to take the positive NPV project without accessing the capital market. Condition 7: I + - (Y P -D 1 )+Y R -D 2 > Y RR + R + - D 3 : When condition 7 holds, the firm cannot raise enough capital to take the positive NPV project due to existing senior debt. This leads to the benefit of an asset sale. Upon the realization of recession at date 1, if conditions 6 and 7 hold the firm will under-invest at date 2 if the recession continues. We assume the firm is not in bankruptcy at date 1 in recession. This is to insure these are voluntary asset sale and not initiated due to financial distress. This leads to condition 8: Condition 8: Y R + ( )*Y P + (1- δ)*y RR > D 2 + D 3 Condition 8 insures that at date 1, in recession, the expected future cashflow (or total asset value) is greater than the outstanding debt. When conditions 6-8 hold the firm can increase value through an asset sale. A.2.2 Use of proceeds There are several possible methods by which the firm can alleviate the under-investment problem through an asset sale. For example, if I + < Y R + ( )*Y P + (1- δ)*y RR, the firm could simply sell a fraction of assets, α, such that [Y R + ( )*Y P + (1- δ)*y RR ]*α = I + and retain the cash from the asset sale. 14 In this case at date 2 if the positive NPV project arrives the firm has enough free cashflow to take the project. Although this alleviates the underinvestment problem, this 14 This assumes the assets are perfectly divisible and sold at fair market value. 26