Journal of Banking & Finance 34 (2010) 324–335

Journal of Banking & Finance 34 (2010) 324–335

Contents lists available at ScienceDirect

Journal of Banking & Finance

journal homepage: www.elsevier .com/locate / jbf

Long-term debt and overinvestment agency problem

Ranjan D’Mello a,*, Mercedes Miranda b

a Department of Finance, Wayne State University, 328.8 Prentis, Detroit, MI 48202, United States b Department of Accounting and Finance, The University of Michigan-Dearborn, Dearborn, MI 48126, United States

a r t i c l e i n f o a b s t r a c t

Article history: Received 6 March 2009 Accepted 27 July 2009 Available online 30 July 2009

JEL classification: G30 G31 G32

Keywords: Debt issues Agency costs Overinvestments

0378-4266/$ – see front matter � 2009 Elsevier B.V. A doi:10.1016/j.jbankfin.2009.07.021

• Corresponding author. Tel.: +1 313 577 7828; fax E-mail addresses: rdmello@wayne.edu (R. D’Mello

(M. Miranda). 1 The underlying assumption is that managers’ perqu

We investigate the role of long-term debt in influencing overinvestments by analyzing the pattern of abnormal investments around a new debt offering by unlevered firms. Before being levered when the dis- ciplining role of debt is missing, firms retain excessive amounts of cash. The introduction of debt leads to a dramatic decline in cash ratios and the relation is stronger for firms classified as having poor investment opportunities. For the sub-sample of firms that overinvest in real assets, issuing debt leads to a reduction in abnormal capital expenditures. The decline in overinvestments is explained by debt service obligations that reduce discretionary funds under managerial control. Further, the reduction in overinvestments has a positive impact on equity value. These conclusions hold in other settings where there is a dramatic change in firms’ capital structures providing strong support for the hypothesis that debt reduces overinvestments.

� 2009 Elsevier B.V. All rights reserved.

1. Introduction ies that analyze capital structures find that firms with poor growth 2

2 An exception is Harvey et al. (2004) who examine the effect of debt on agency and information problems in emerging market firms that routinely have pyramid ownership structures and the role of international debt in ameliorating these problems.

The free cash flow theory posits that managers of firms with funds in excess of that required to finance positive NPV projects do not distribute the cash to shareholders. Rather, these firms either maintain surplus amounts of cash (i.e., overinvest in cash) or invest excessively in real assets (i.e., overinvest in capital expen- ditures). These overinvestments, while maximizing managers’ per- sonal utility, reduce firm value.1 Myers (1977), Jensen (1986), and Stulz (1990) among others argue that a potential solution to the overinvestment problem is to issue debt. Debt disciplines managers by forcing them to pay out excess cash thereby reducing the amount of funds under their discretion. In addition, when debt is issued in a swap for equity, the fraction of managerial stock ownership in- creases further aligning managers’ interest with that of shareholders. Issuing debt is therefore hypothesized to mitigate managers’ pro- pensity to overinvest and this benefit of debt is an important factor in determining a firm’s optimum leverage ratio in the tradeoff theory.

While there is a lot of theoretical justification for issuing debt to control abnormal investments, henceforth referred to as the over- investment control hypothesis, there is limited empirical evidence that directly supports the hypothesis. For example, previous stud-

ll rights reserved.

: +1 313 577 5486. ), mmirand@umd.umich.edu

isites increase with firm size.

opportunities have high debt levels with long-term maturity. However, these results do not clearly establish that managers of firms that do not face the monitoring role of debt engage in value dissipating investments and issuing debt reduces or eliminates these overinvestments.

In this paper we present a direct test of the overinvestment hypothesis by examining the pattern of abnormal investments around a debt issue conducted by firms that have been unlevered for some time before the offering.3 This relatively unique event pro- vides an ideal environment to test two major implications of the hypothesis. First, that self-interested managers of unlevered firm are more likely to overinvest because they are not subject to the dis- ciplining role of debt. Second, that the introduction of leverage will lower these overinvestments because debt reduces the amount of discretionary funds under the control of managers.

3 We do not attempt to answer the more general question of why firms conduct a debt IPO, an issue that has been addressed by Datta et al. (2000). A point to be noted is that while Datta et al. investigate reputation building, information asymmetry, and optimum debt mix as possible rationales, they do not examine reduction in agency costs as a motivation, which is the focus of our paper. Also see Krishnaswami and Yaman (2007) and Hale and Santos (2008). Barry et al. (2009) test the market timing hypothesis.

http://dx.doi.org/10.1016/j.jbankfin.2009.07.021
mailto:rdmello@wayne.edu
mailto:mmirand@umd.umich.edu
http://www.sciencedirect.com/science/journal/03784266
http://www.elsevier.com/locate/jbf
Table 1 Descriptive statistics for sample firms.

Variable Mean Median

Book value of asset 153.24 29.60 Market value of asset 476.02 61.30 Market value of equity 432.26 50.85 Sales 217.39 29.50 Market-to-book ratio 2.62 1.63 Debt ratio – year 0 (%) 39.88 19.08 Cash 28.80 5.89 R&D 4.00 0.00 Capital expenditure 16.36 1.41

Descriptive statistics for a sample of 366 debt offerings between 1968 and 2001 by previously unlevered firms. A debt issue is deemed to have been conducted when the difference in long-term debt relative to the previous year exceeds five percent of the pre-issue book value of assets. An unlevered firm is defined to have conducted a debt offering if it issues long-term debt after three years of no debt and maintains a long-term debt to net asset ratio of five percent for three years subsequent to the offering. Net asset ratio is book assets less long-term debt market value of assets is book value of assets plus the difference between the book value and market value of equity. Market value of equity is the year-end shares outstanding times the market price. Market-to-book ratio is market value of assets divided by book value of assets. R&D is assigned a value of 0 if no data is available. Mean values are in the first row and median values are in the second row. All values except for debt ratio are at the year-end prior to the debt issue. Level variables are expressed in 2001 dollars. All data are obtained from COMPUSTAT.

R. D’Mello, M. Miranda / Journal of Banking & Finance 34 (2010) 324–335 325

While we focus on a new debt offering by an unlevered firm, it could be argued that the implications of the overinvestment control hypothesis can also be observed around other leverage increasing events such as a secondary debt issue, debt-for-equity exchange of- fer, or equity repurchase. However, in all of these events the mar- ginal impact of a change in a firm’s debt ratio on its investment policy might not be significant especially if the relation between leverage and reduction in overinvestments is not linear.

Our results for a sample of 366 debt offers by unlevered firms conducted between 1968 and 2001 strongly support the hypothe- sis that debt influences overinvestments. When unencumbered by the constraints of debt, unlevered firms retain excessive amounts of cash. The liquidity ratios of these firms exceed their industry benchmarks by approximately 20 percentage points in the interval immediately preceding a debt offering. Introducing leverage in the capital structure leads to a significant reduction in this form of overinvestment and within three years of the offering the cash ra- tios of the sample firms are lower than their benchmarks. Issuing debt also affects overinvestments in real assets but only for firms that were investing excessively before the offering. For these firms, there is a dramatic decline in abnormal capital expenditures after the introduction of leverage. Consistent with the overinvestment control hypothesis, the decline in these forms of excess invest- ments is related to the fraction of debt issued, which proxies for the degree to which debt disciplines managers.

We document a significant relation between the decline in over- investments and the interest payments associated with the new debt offering. This finding suggests that the debt service obligation disgorges cash and reduces the amount of discretionary funds under managerial control that results in reduced levels of surplus invest- ments. We also investigate the interaction between investment opportunities proxied by market-to-book ratio, overinvestments, and the role of debt in influencing this agency problem. Firms that are classified as being poorly managed with inferior investments opportunities have higher cash ratios when they are unlevered and experience a greater decline in this form of overinvestment after issuing debt relative to other sample firms. This result suggests that while debt reduces surplus investments in general, it plays an espe- cially important role in reducing excess investments in firms that have the greatest overinvestment agency problem.

For completeness, we also explore the relation between changes in overinvestments around the debt issue and corresponding changes in market value of equity. We find that a decline in abnor- mal cash ratio leads to an increase in equity value and this associ- ation is stronger for poorly managed firms. We find a similar relation between equity value and change in excess real asset investments for firms that overinvest when unlevered. These find- ings indicate that the decline in overinvestments has a positive im- pact on equity value especially for firms with high agency costs.

We test the robustness of our findings around an alternative event that also triggers a dramatic shift in a firm’s capital structure, namely elimination of long-term debt. Debt eliminations mirror debt introductions in that after recapitalization managers of firms previously constrained by debt are now free from these restric- tions. Consistent with the overinvestment control hypothesis, we document a rise in abnormal investments after firms eliminate debt. The increase is related to the growth in discretionary funds under managerial control resulting from firms retaining rather than paying to shareholders monies that was previously used to service debt obligations.

4 The definition of a debt issue is similar to Hovakimian et al. (2001) and Leary and Roberts (2005).

5 We reach similar conclusions if firms are unlevered for a minimum of five or seven years before the debt issue.

1. Sample description and characteristics

The initial debt issuing sample is chosen from all firms listed on the Compustat database for the period 1965–2004 that are non-

regulated (other than SIC codes 4900–4999), that do not belong to the financial sector (other than SIC codes 6000–6999), and that have valid industry classification codes (other than SIC codes 9900–9999). A firm is assumed to issue debt if the change in book value of total long-term debt (data 9 + data 34) exceeds five per- cent of the pre-issue book value of assets (data 6 – total long-term debt).4 Because our test requires firms to be unlevered for an ex- tended period before the offering we restrict our sample to firms that issue debt after having no long-term debt for at least three consecu- tive years.5 To ensure that the debt financing is not temporary and that newly issued debt significantly influences overinvestments in the future, we require the firms to maintain debt of at least five per- cent of its net book assets, defined as book assets less long-term debt, for a minimum of three consecutive years immediately follow- ing the new debt offering year. Therefore, our sample only includes unlevered firms that issue debt between 1968 and 2001. We omit annual observations if the firm’s long-term debt exceeds its asset or there is no valid data for sales (data 12) or assets.

We find 366 debt issues conducted by firms that have been unlevered for at least three years immediately preceding the offering. While the frequency of debt issues is relatively stable for most of the period, there is an increase after 1994. The last seven years of the sample period accounts for approximately 40% of the observations.

The descriptive statistics for the sample firms at the year-end preceding the debt issue are presented in Table 1. Given that the sample period encompasses over three decades, we report all level variables in 2001 dollars. The average (median) book value of as- sets is approximately $153.245 ($29.60) million. The market value of assets, defined as book value of assets plus the difference be- tween market and book value of equity (data 25*199 – 60), of the average firm is $476.02 million. Average (median) market equi- ty is $432.26 ($50.85) million and average (median) sales is $217.39 ($29.50) million. The median market-to-book ratio, de- fined as market-to-book value of assets, exceeds 1.6.

326 R. D’Mello, M. Miranda / Journal of Banking & Finance 34 (2010) 324–335

Sample firms issue debt that is approximately 40% of its pre-is- sue book value of assets on average and the median ratio is 19.08%. These ratios suggest that unlevered firms issue a considerable amount of long-term debt and this dramatic change in capital structure is likely to significantly influence the overinvestment agency problem. The median (average) cash level (data 1) is $5.89 ($28.80) million. While the median firm does not invest in Research and Development (R&D), the average R&D expense (data 46) is $4.00 million.6 Finally, issuing firms spend $16.36 million on capital expenditure (data 128) on average.

We examine overinvestments in two types of assets; cash and real assets, which we proxy by capital expenditures. We concen- trate on these two assets because Bates (2005) and Richardson (2006) find that firms with free cash flow retain most of the surplus cash or invest excessively in real assets and that these forms of overinvestments lead to a decline in shareholders’ wealth.7 How- ever, Harford (1999), Opler et al. (1999), and Harford et al. (2008) document that firms with free cash flow also overinvest by under- taking acquisitions. We find that less than 10% of the sample firms report any acquisition activity (data 129) around the debt issue suggesting that this type of investment is relatively infrequent in our sample. Hence, in subsequent analysis we concentrate on investigating the impact of issuing debt on cash and real assets. We define cash ratio as cash and marketable securities to year- end book value of assets less cash. Investment in real assets is cap- tured by the ratio of capital expenditure to the year-end assets less capital expenditure.

Similar to Bates (2005), we determine whether firms systemat- ically overinvest by comparing the cash and capital expenditure ra- tios of sample firms in a given year to those of median industry firms in that year. Industry is defined as all firms in the same three-digit SIC code as the sample firm and does not include any sample firms. The benchmark firms are of comparable size as the sample firms. At the year-end immediately preceding the debt is- sue the average assets and sales of the matching firms are $183.18 million and $189.13 million, respectively (expressed in 2001 dollars). Our results are robust to the criteria used to create the sample of matching firms. For example, we obtain similar re- sults when we restrict matching firms in a given year to those that are in the same industry and are of similar size as the sample firms in that year, defined as having book assets that are between 25% and 200% of our firms, and then choose the median firm. We also reach similar conclusions when the matching firm is in the same industry and is closest in size to the sample firm at the year-end before the debt issue and this firm is used to adjust for industry effects in the interval around the offering.

1. Results

3.1. Pattern of overinvestments around debt offerings

The overinvestment control hypothesis has two major implica- tions. First, in the absence of debt managers are more likely to overinvest in either cash or real assets or both. Second, the intro- duction of debt reduces free cash flow available to managers as well as disgorges existing cash thus reducing these two forms of overinvestments. We examine these implications by analyzing

6 Similar to Opler et al. (1999) and D’Mello et al. (2008), firms that do not report R&D expenditures are assumed to have zero R&D expenditures.

7 Jensen (1986), Morck et al. (1990), and Harford et al. (2008) document that firms where managers’ interests are not perfectly aligned with those of shareholders invest in negative NPV projects that lower shareholders’ wealth. Further, Harford (1999), Opler et al. (1999), Faulkender and Wang (2006), and Dittmar and Mahrt-Smith (2007) find that investors do not value firms with excess cash flows highly. Pinkowitz et al. (2006) report similar findings in an international setting.

industry-adjusted cash and capital expenditure ratios in the se- ven-year period centered on the year-end immediately following the debt issue (year 0).

The results are presented in Table 2. There are substantial dif- ferences between mean and median results suggesting that outli- ers have a considerable influence on the results. Therefore, we concentrate on median results in the discussion of univariate anal- yses and control for extreme observations in subsequent regres- sion tests.

3.1.1. Industry-adjusted cash ratios around debt offerings In the period before issuing debt, the median industry-adjusted

cash ratio (first row) is significantly positive. For example, three years before the debt offering, the cash ratio of the median sample firm exceeds its industry counterpart by approximately 25 per- centage points. The ratio declines as we approach year 0 but even at the year-end immediately preceding the debt issue the abnor- mal cash ratio exceeds 15%. We observe a dramatic decline in industry-adjusted cash ratio after the firms conduct a debt offering and the ratio becomes significantly negative two years subsequent to the debt issue. We present the time-series graph of the cash ra- tio of the median sample firm and the industry match in Panel A of Fig. 1. While the sample firm’s cash ratio drops dramatically after the debt issue that for the industry match remains relatively con- stant over the seven-year interval.

The last three columns of the table present the sample firms’ average industry-adjusted cash ratios in the three-year period be- fore (years �3 to �1) and after (years 1 to 3) the debt issue and the difference between the two ratios. We define a firm’s average cash ratio as the median industry-adjusted cash ratio over the three- year interval. We use median values to control for the impact of outliers on the results. As an alternative measure we calculate the median raw cash ratio for the three-year interval before and after the debt issue for both the sample firm and the median firm in the industry and the difference in the cash ratios between these two firms is defined as the abnormal or industry-adjusted cash ra- tios for each of the two intervals. The results for these two mea- sures are similar and we only present the results of the first measure.

In the pre-issue period, the median overinvestment in cash rel- ative to the industry benchmark is 20.31%. The median (mean) change in industry-adjusted cash ratio from before to after the is- sue is approximately �19.5 (�38.5) percentage points, significant at one percent and subsequent to the issue, the cash ratio of the sample firms is similar to the industry benchmarks.8 These findings suggest that firms maintain excessive amounts of cash when unlevered but these abnormal investments are eliminated after the introduction of debt, consistent with the predictions of the overin- vestment control hypothesis.

There might be alternative explanations for the pattern in abnormal cash ratios around a debt issue that are unrelated to role of leverage on overinvestments. For example, if firms invest the offering proceeds in real assets, there will be an increase in the book value of assets without a corresponding change in cash caus- ing the cash ratio, defined as cash divided by assets less cash, to de- cline. This possibility suggests that the observed change in cash ratios after the offering is mechanical in nature and that debt does not influence the extent to which managers overinvest.

8 We also investigate whether changes in corporate governance methods in the latter part of the sample period affects the results. We split the sample into two groups using 1994 to divide the sample because there is an increase in the number of observations beyond this year. We observe that the median decline in the pre-1995 sub-sample is 18.03% and that for post-1994 sub-sample is 25% suggesting that the pattern of declining cash ratios after the debt issue holds across time and is not affected by changes in corporate governance mechanisms.

Table 2 Industry-adjusted cash and capital expenditure ratios around new long-term debt offerings.

�3 �2 �1 0 1 2 3 Pre-Issue Post-Issue Difference

Cash ratios 55.38* 50.04* 39.58* 17.96* 10.71 10.49 11.57 49.84* 9.65 �38.48* (24.67) (21.47) (15.07) (2.43) (0.79) (�0.35) (�0.18) (20.31) (0.33) (�19.47***) [359] [359] [359] [364] [364] [365] [365] [365] [365] [364]

Capital expenditure ratios 1.83** 3.61** 2.59* 7.75vz 4.78* 2.35* 2.06* 1.26 1.67* 0.36 (�1.03) (�0.74) (�0.44) (1.00) (�0.10) (0.07) (�0.37) (�0.88) (�0.25) (0.55*) [354] [359] [361] [361] [361] [364] [362] [363] [364] [361]

The sample consists of 366 debt offerings between 1968 and 2001 by previously unlevered firms. An unlevered firm is defined to have conducted a debt offering if it issues long-term debt after three years of no debt and maintains a long-term debt to net asset ratio of five percent for three years subsequent to the offering. A debt issue is deemed to have been conducted when the difference in long-term debt relative to the previous year exceeds five percent of the pre-issue book value of assets. Net asset ratio is book assets less debt. Mean and median values are in the first two rows and the number of observations is in the third row. Cash ratio is cash and marketable securities divided by book assets less cash. Capital expenditure ratio is the ratio of capital expenditure to book assets less capital expenditure. Industry adjustment is made by subtracting the ratio of the median firm in the industry, defined as all firms in the same three digit SIC code. Pre-issue (post-issue) is the three-year interval immediately preceding (subsequent) to the debt offering year (year 0). For each firm we calculate the median value over the pre-issue (post-issue) interval and the average of these values is presented in the pre- issue (post-issue) columns. The Difference column contains the difference in a variable from the post-issue interval relative to the pre-issue interval. * Indicates significance at the 10% levels, respectively. All data are obtained from COMPUSTAT. ** Indicates significance at the 5% levels respectively. All data are obtained from COMPUSTAT. *** Indicates significance at the 1% levels, respectively. All data are obtained from COMPUSTAT.

Panel A: Cash Ratio

Panel B: Cash Holdings

Fig. 1. Graphs of cash ratio (%), defined as cash divided by assets less cash (Panel A), and dollar cash holdings (Panel B) for the sample of unlevered firms that subsequently conducted debt offerings between 1968 and 2001 and the median industry matched firm for the seven-year period centered on the debt issue year (year 0). An unlevered firm is defined to have conducted a debt offering if it issues long-term debt after three years of no debt and maintains a long-term debt to net asset ratio of five percent for three years subsequent to the offering. A firm has issued debt if the difference in long-term debt relative to the previous year exceeded five percent of the pre-issue book value of assets and net asset ratio is book assets less debt.

R. D’Mello, M. Miranda / Journal of Banking & Finance 34 (2010) 324–335 327

We investigate this explanation by analyzing the industry-ad- justed percentage change in annual cash levels as well as in cash levels from the three-year interval before to after the issue (results not tabulated).9 We find that at the year-end of the issue, the med-

9 These results and all other findings that are not tabulated are available on request.

ian annual abnormal percentage change in cash is �13.70% and cash balances continue to decline by over seven percent annually for two subsequent years. The percentage change in cash levels from the three-year interval before to that after the issue is �48.54%. We also graph the cash holdings for the median sample firm and its industry match for the seven-year period centered around the debt issue in Panel B of Fig. 1. We observe that the cash level of the median matching firm increases gradually over time while that for the sam- ple firm declines after the security issue. These results suggest that issuing debt reduces the amount of cash that a firm holds.

Another possible explanation for the pattern in cash ratios is change in accessibility to capital markets. Firms that do not have cheap and ready access to debt capital are likely to be unlevered and retain large amounts of cash to finance future investments. When access to capital markets improve, that is, when firms are able to issue debt, the need for excess liquidity declines resulting in a reduction in cash ratios. Therefore, it is not the debt offering per se but rather the access to capital markets that explains the change in cash ratios in Table 2.

We test this argument by comparing the pattern of overinvest- ments between sample firms that have greater access to capital markets and those that find it difficult to raise external financing. Acharya et al. (2007) and Faulkender and Wang (2006) argue that large firms face fewer constraints and have easier access to capital markets than small firms. These studies also argue that firms that do not have rated debt have difficulty raising external funds. If dif- ference in accessing capital market is the explanation for the pat- tern of excess cash ratios around a debt issue, we should observe large firms maintaining less cash when unlevered and experienc- ing a smaller decline in liquidity at the offering than other sample firms. Similarly, firms that do not have rated debt should have a smaller reduction in cash and a larger cash ratio subsequent to the issue relative to other firms.

We define large firms as those that have positive industry-ad- justed assets before the offering, that is, firms that are larger than their industry counterparts, and these firms are expected to have greater access to capital markets than other sample firms. Sample firms whose debt is not rated in the three-year interval subsequent to the offering are determined to have limited access to capital markets.10 We find that for the 115 firms with positive industry-ad- justed assets, the median abnormal cash ratio is 19.10% before the issue and �1.43% after the issue, a decline of 18.61 percentage points that is highly significant (results not tabulated). For firms that are

10 Debt rating information is available on Compustat beginning 1985.

13 Given our findings that there is a decline in abnormal cash holdings and no

328 R. D’Mello, M. Miranda / Journal of Banking & Finance 34 (2010) 324–335

classified as being constrained, the median decline in cash ratios from the interval before the debt issue to that after is 18.85%, signif- icant at 1%.11 Test statistics of differences in mean and median de- clines in liquidity between the constrained and non-constrained sub-samples are insignificant suggesting that both sub-samples experience similar decreases in abnormal cash ratios. Similarly, firms with rated debt have similar industry-adjusted cash ratios before and after the debt issue and comparable changes in these ratios as other sample firms. These findings suggest that access to capital markets is not the primary explanation for the pattern of cash ratios in Table 2.

The dramatic decline in liquidity might also be an artifact of the process of conducting a security issue and that the security issued itself is not important. An alternative explanation is that firms fi- nance investments with the sum of the offering proceeds and accu- mulated cash thus causing the excess cash ratios to decline after the offering. These arguments imply that there should be a similar decline in cash ratios after a firm issues non-debt securities. We investigate this possibility by examining the cash ratios around the sale of preferred and common stock (data 108). As with our ori- ginal sample, we restrict our attention to firms that issue stock of at least five percent of its pre-issue assets, that the offering is the first in three years, and that these firms do not conduct another equity issue for the subsequent three years. For these firms, we do not observe a significant change in industry-adjusted cash ra- tios from before to after the issue. This result suggests that the principal explanation for the pattern in cash ratio is the fact that the firm issues debt and not that it conducts a security offering.12

We also investigate why issuing debt reduces overinvestment in cash. Jensen (1986), Stulz (1990), Hart and Moore (1995), and Zwie- bel (1996) suggest that debt forces managers to service these obli- gations thus reducing the amount of cash under their control. We test this argument by examining the relation between interest pay- ments associated with the debt offering standardized by assets net of cash and the change in cash ratios of the sample firms around the issue. We find the Pearson and Spearman correlation coefficients to be significantly negative. Similarly, when we regress changes in cash ratios on interest payments the slope coefficient is signifi- cantly negative (one percent level). These results imply that new debt service obligations associated with the introduction of lever- age reduces the accumulated surplus cash ratios of sample firms.

Overall, these findings suggest that long-term debt significantly influences overinvestments in liquid assets. When firms are unle- vered, that is, when managers are free from the restrictions of debt, they retain substantially more cash than comparable industry firms. The introduction of debt disgorges excess cash by forcing managers to make interest payments, causing the firm’s ex-post liquidity ratio to be similar to its benchmark.

3.1.2. Industry-adjusted capital expenditure ratios around debt offerings

In the second row of Table 2, we examine the pattern of overin- vestment in real assets proxied by capital expenditures around the debt offering. Before the issue, the annual industry-adjusted capital expenditure ratio is similar to the benchmark in each of the three years. In the pre-issue period, the average capital expenditure of the median sample firm is 0.88 percentage points lower than the corresponding investment by benchmark firms. These findings im-

11 For this sub-sample of firms, the median (mean) abnormal cash ratios in the pre- issue and post-issue periods are 22.30 (52.11)% and 0.52 (7.93)%, respectively.

12 We examine the possibility that the sample firms coincidentally use the excess cash to initiate or increase share repurchases or dividends beginning the offering year. The median abnormal cash payout ratios are zero around the debt issue suggesting that the reduction in cash ratio cannot be explained by firms altering their payout policy.

ply that managers that do not face the disciplining role of debt do not invest in negative NPV projects.13 Thus, an examination of the role of debt in curtailing excess investments will provide limited information relevant to the overinvestment control hypothesis. In fact, the median industry-adjusted change in capital expenditure from before to after the issue is 0.55 percentage points.

While issuing debt does not influence investments in real assets in the full sample, it could play an important role for firms that have the greatest overinvestment problem. That is, debt alleviates this agency cost but only for firms that invest excessively before the issue. We test this argument by examining the pattern of abnormal investments for firms with excessive real asset invest- ments defined as those that have higher capital expenditure ratios relative to their benchmark firms in each of the two years immedi- ately preceding the debt issue.14 If issuing debt controls real asset overinvestments we should observe a dramatic decrease in indus- try-adjusted capital expenditure ratio after the offering for these firms. For all other firms, debt is not expected to impact capital expenditures because these firms are not overinvesting.

The results of our tests are presented in Table 3. There are 114 firms that are classified as overinvesting in real assets before the debt offering. For this sub-sample there is a dramatic decline in excess investments after the introduction of leverage. Abnormal capital expenditure of the median firm, which is 4.69% before the issue, declines by 2.67 percentage points subsequent to the issue. However, the median excess capital expenditure in the three years after the issue is significantly positive, implying that these firms continue overinvesting in real assets. Thus, while issuing debt re- duces abnormal capital expenditure for firms with the highest agency costs, it does not eliminate the problem.

An alternative explanation for the decline in excess investments is that industry-adjusted capital expenditure ratio is mean reverting. That is, firms with abnormally high (i.e., positive indus- try-adjusted) or low (i.e., negative industry-adjusted) capital expenditures in one period move towards the mean, which we de- fine as the benchmark firm’s capital expenditure, in the following period. This implies that the results of Table 3 can be explained by the criterion used to classify firms as overinvesting rather than by the introduction of debt in the capital structure.

We distinguish the mean reversion argument from the overin- vestment control hypothesis by examining the pattern of capital expenditure ratios of non-sample matching industry firms that have similar investments in real assets as the sample firms in the pre-issue period. For each sample firm that is overinvesting in real assets, we find a matching firm in the same industry with median raw capital expenditure ratio in the three-year pre-issue period that is closest to that of the sample firm. For each matching firm we then find the median capital expenditure ratio in the post-issue period and the change in the ratios across the two periods. Finally, we calculate the difference in the change in the capital expenditure ratios between the sample and the matched firm and average the difference across all firms. Observing a greater decline in the sam- ple firms relative to the matching firm is consistent with the hypothesis that issuing debt reduces overinvestments in firms that were investing excessively in real assets.

increase in industry-adjusted capital expenditure the question arises as to what firms do with the issue proceeds. We find that increase in raw capital expenditure in the issue year relative to the previous year accounts for 30% of the proceeds raised and change in working capital net of cash from year �1 to year 0 is about 17% of the debt issue. Interest payment at the year-end of the offer (time 0) is about 5% of the proceeds and this ratio increases to about 15% in year 3.

14 Our results are not sensitive if we classify firms that invest excessively as those that have positive abnormal capital expenditures for all three years or for a minimum of two years in the pre-issue period.

Table 3 Industry-adjusted capital expenditure ratio around new debt offerings for firms that overinvest.

Pre-issue Post-issue Difference

9.82* 4.08* �5.79* (4.69) (2.01) �2.67* 114 113 113

The sample consists of unlevered firms that conduct a long-term debt offering that had positive industry-adjusted capital expenditure ratios in the two years imme- diately preceding the debt offering year. An unlevered firm is defined to have conducted a debt offering if it issues long-term debt after three years of no debt and maintains a long-term debt to net asset ratio of five percent for three years sub- sequent to the offering. A debt issue is deemed to have been conducted when the difference in long-term debt relative to the previous year exceeds five percent of the pre-issue book value of assets. Net asset ratio is book assets less debt. Capital expenditure is standardized by year-end book value of assets less capital expen- diture. Industry-adjusted capital expenditure is obtained by reducing the capital expenditure of the median firm in the industry defined as all firms in the same three-digit SIC code. Pre-issue (Post-issue) is the three-year interval immediately preceding (subsequent) to the debt issue year (year 0). For each firm we calculate the median value over the pre-issue (post-issue) interval and the average across all firms is presented in the pre-issue (post-issue) columns. The Difference column contains the difference in a variable from the post-issue interval relative to the pre- issue interval. *** Indicates significance at the 1% level. All data are obtained from COMPUSTAT.

R. D’Mello, M. Miranda / Journal of Banking & Finance 34 (2010) 324–335 329

We find that the median (mean) difference in capital expendi- ture ratios between the sample and matched firms is 0.01 (0.07) percentage points in the pre-issue period, indicating that both these sub-samples of firms have similar levels of overinvestments. The change in capital expenditure for the median sample firm is approximately 0.9 percentage points more negative than that for a comparable firm and in the post-issue period the raw capital expenditure of the median matching firm is about 0.5 percentage points higher than the sample firm. These results confirm our ear- lier conclusion that the introduction of debt reduces overinvest- ment for firms that ex-ante were investing excessively in real assets.

The overinvestment control hypothesis posits that leverage re- duces the amount of discretionary funds under managerial control thus lowering abnormal capital expenditures. This argument im- plies a negative relation between the interest payments associated with the newly issued debt and the change in this form of overin- vestment. Consistent with this conjecture, we find the correlation coefficients between these two variables to be significantly nega- tive but only for the sub-sample of firms that have positive indus- try-adjusted capital expenditure before the debt offering. For the full sample, the correlation is positive but insignificant.

Overall, the examination of the capital expenditure ratios around new debt issues yields the following conclusions: Firms, on average, do not overinvest in real assets when the disciplining role of debt is missing. For firms that do overinvest, debt reduces but does not eliminate abnormal capital expenditures. The reason for the reduction is the debt service obligations that reduce the amount of unrestricted funds that is available for overinvesting in real assets.

3.2. Future growth opportunities and the overinvestment control hypothesis

We also investigate the impact of future growth opportunities on the role of debt in controlling overinvestments. Previous studies suggest that firms that are poorly managed with non-positive investment opportunities are more likely to overinvest.15 In the

15 See Harford (1999), Opler et al. (1999), Pinkowitz and Williamson (2004), and Faulkender and Wang (2006).

context of this paper, the argument implies that these firms will have higher industry-adjusted cash and capital expenditures ratios before the debt issue and a greater reduction in these forms of overinvest- ments after the introduction of debt relative to other sample firms.

We test this hypothesis by classifying firms into two groups rel- ative to the industry using the market-to-book ratio as a proxy for the quality of investment opportunities. Firms that have negative industry-adjusted market-to-book ratio in the two years immedi- ately before the issue year are classified as being poorly managed firms with negative NPV investment opportunities compared to the industry benchmark (IMBR < 0). We classify all other firms as being well managed and having good investment opportunities (IMBR P 0). To the extent that firms with relatively poor invest- ment opportunities, that is, firms with negative industry-adjusted market-to-book ratio in only the year immediately before the is- sue, are classified as being well managed, this classification scheme biases the results against the overinvestment control hypothesis.

We present the results in Table 4. We classify 154 of the sample firms as being poorly managed with negative NPV projects in the future. For these firms, the median cash ratio before the debt issue is 29.67%. For other firms in the sample, the median cash ratio is 18.06% implying that even firms that are not classified as being poorly managed but that are free from the restrictions of debt maintain abnormal amounts of cash. The difference between the two sub-samples is significant suggesting that poorly managed firms maintain more excess cash and thus have greater agency costs than other firms.

Subsequent to the issue there is a dramatic decline in cash ra- tios for both sub-samples of firms. The median decline for firms classified as being poorly managed is 21.28 percentage points and this is significantly more negative than the 17.88 percentage points decline for the other sub-sample of firms. We also compare the industry-adjusted percentage change in cash levels from before to after the issue for the two sub-samples. We find that the median decline for the sub-sample of poorly managed firms is 55.36% and this is marginally more negative than the 42.62% for the other firms in the sample. In the post-issue period, the cash ratios of both sub-samples of firms are similar to their industry benchmark firms.

These results indicate that while unlevered firms retain exces- sive amounts of cash before issuing debt, the problem is greater for firms that are poorly managed. Introducing leverage leads to a reduction in this overinvestment for both sub-samples but the decline is greater for firms with poor investment opportunities. Thus, while debt reduces this form of agency costs in general, it plays an especially important role for poorly managed firms. This finding supports Harvey, Lins, and Roper’s (2004) conclusion in emerging markets.

In Panel B, we compare the industry-adjust capital expenditure ratios as well as the change in the ratio from before to after the is- sue for the two sub-samples of firms. Before the offering when firms are not subject to the disciplining role of debt, there is no evi- dence that firms with poorly managed invest more on capital expenditures than their industry counterparts or than the sub- sample of firms with good management. In fact, the industry-ad- justed ratio is significantly negative and it is lower than the ratio for the sub-sample of firms that are classified as being well managed.

Subsequent to the offering, the median industry-adjusted ratio for firms with good investment opportunities is positive and signif- icantly higher than for firms with market-to-book ratios less than the benchmark. This result suggests that well-managed firms use the proceeds of the debt issue to finance positive NPV projects. For firms with negative market-to-book ratio, the median indus- try-adjusted change in real asset investments is significantly posi- tive, which in contrary to the prediction of the overinvestment control hypothesis.

Table 4 Industry-adjusted cash and capital expenditure ratios for firms classified by market- to-book ratio.

Pre-issue Post-issue Difference

Panel A: cash ratios IMBR < 0 67.12* 11.07* �52.09***

(29.67) (0.49) (�21.28) [154] [154] [153]

IMBR P 0 37.23* 8.61 �28.61*** (18.06) (�0.26) (�17.88) [211] [211] [211]

t-Test 2.57** 0.46 2.11**

Wilcoxon Z (1.96) (0.34) (1.69)

Panel B: capital expenditure ratios IMBR < 0 0.13 0.84 0.59

(�1.78*) (�0.65) (0.75) [154] [154] [153]

IMBR P 0 2.07* 2.27* 0.20 (�0.09) (0.37***) (0.31) [209] [210] [208]

t-Test 1.61 1.60 0.31 Wilcoxon Z (4.13) (2.83) (1.49)

The sample consists of 366 debt offerings between 1968 and 2001 by previously unlevered firms. An unlevered firm is defined to have conducted a debt offering if it issues long-term debt after three years of no debt and maintains a long-term debt to net asset ratio of five percent for three years subsequent to the offering. A debt issue is deemed to have been conducted when the difference in long-term debt relative to the previous year exceeded five percent of the pre-issue book value of assets. Net asset ratio is book assets less debt. Cash ratio is cash and marketable securities divided by book assets less cash. Capital expenditure ratio is the ratio of capital expenditure to book assets less capital expenditure. Market-to-book ratio (MBR) is defined as book value of assets plus the difference between market and book values of equity divided by book assets. Industry adjustment is made by subtracting the ratio of the median firm in the industry, defined as all firms in the same three digit SIC code. IMBR < 0 contains firms that had negative industry-adjusted MBR in each of the two years immediately preceding the debt offering year. All other firms are classified as IMBR P 0. Pre-issue (post-issue) is the three-year interval immediately preceding (subsequent) to the debt issue year (year 0). For each firm we calculate the median value over the pre-issue (post-issue) interval and the average across all firms is presented in the pre-issue (post-issue) columns. The Difference column contains the difference in a variable from the post-issue interval relative to the pre- issue interval. * Indicates significance at the 10% levels, respectively. All data are obtained from COMPUSTAT. ** Indicates significance at the 5% levels, respectively. All data are obtained from COMPUSTAT. *** Indicates significance at the 1% levels, respectively. All data are obtained from COMPUSTAT.

330 R. D’Mello, M. Miranda / Journal of Banking & Finance 34 (2010) 324–335

3.3. Multivariate regression analysis

We also examine the overinvestment control hypothesis in a multivariate framework. Initially, we test the implication that in the absence of the disciplining role of debt firms overinvest in cash. We do so by estimating the following regression for the sample of debt issuing firms and their corresponding benchmarks:16

CRi ¼ a0 þ b1 � ðDODUMiÞ þ b2 �MBRi þ b3 � lnðASTiÞ þ b4 � CFRi þ b5 �NWCRi þ b6 � CEXPRi þ b7 � STDEVi þ b8

� RDSi þ b9 � DDUMiþ b10 � XN

m¼1 ðTPDUMmÞ þ b11

� XT

q¼1 IDUMq þ eI ð1Þ

where CR is the firm’s cash ratio, DO DUM is a dummy variable that takes on the value of 1 if the previously unlevered firm conducts a debt offering and 0 otherwise, MBR is the market-to-book ratio, Ln(AST) is the natural log of book assets, CFR is the cash flow ratio, NWCR is the net working capital ratio, CEXPR is the capital expendi- ture ratio, STDEV is the standard deviation of cash flow of the med- ian firm in the industry in the previous five years, RDS is research and development expense to sales ratio, DDUM is a dummy variable if the firm pays dividends, TP DUM is a time period dummy for each five-year interval beginning in 1968, N is the number of five-year intervals, and IDUM is a dummy variable for each industry based on the one-digit SIC code where T is the number of industries.17

We introduce a time period dummy because the distribution of debt offerings in our sample is not constant across time suggesting that the interval when a firm issues debt might play an important role and hence must be controlled for. Additionally, Jensen (1986) and Servaes (1994) find that certain industries are more likely to experience the overinvestment problem thus requiring us to con- trol for industry effects. For each firm, the regression variables other than the dummy variables are the median values over the three-year interval immediately preceding the debt offering but we reach similar conclusions if we use year �1 variables.

We present the results in column (1) of Table 5. The coefficient for DO DUM is significantly positive. This result indicates that after controlling for other factors, unlevered firms maintain cash ratios that exceed their benchmark firms by 30.6 percentage points on average in the interval before the debt offering. This result confirms our earlier findings that firms that are not subject to the disciplining role of debt overinvest in cash.

Next, we test the second major implication of the overinvest- ment control hypothesis that the introduction of debt directly reduces abnormal investments. We do so by estimating the regres- sion similar to (1) but all variables other than the dummy variables are expressed as changes from the three-year interval before to after the debt offering and are adjusted for industry effects.18

Therefore, for example, the dependent variable is the abnormal change in the cash ratio defined as the change in cash ratio of sample firms less that of benchmark firms.

If debt directly restricts managers’ propensity to overinvest, there should also be a relation between the fraction of debt issued and the decline in excess cash ratios. We therefore add an indepen-

16 This regression is similar to specification 4 in Table 4 of Opler et al. (1999). We omit the regulation dummy because regulated firms do not form a part of our final sample. We control for outliers by using the Cook’s D test that eliminates data points that have an excessive influence on the results.

17 The last period contains debt offerings conducted between 1998 and 2001, a period of only four years.

18 Measuring change from year �1 to year +1 yields similar conclusions. Because we use industry-adjusted variables DO DUM and STDEV are eliminated from the regression.

dent variable, Debt Issued, defined as the median debt ratio after the offering and this captures the disciplining role of newly issued debt. We also investigate whether the ability of debt to reduce overinvestment is more valuable to poorly managed firms with few valuable investment opportunities. To do so we introduce a dummy variable IMBR DUM, which takes on the value of 1 if the firm’s industry-adjusted market-to-book ratio is negative in the two years immediately before the offering and 0 otherwise and also interact this variable with the fraction of debt issued (IMBR DUM * Debt Issued).

The regression results are presented in columns (2) and (3). The coefficient for Debt Issued is negative and significant indicating that the introduction of leverage directly leads to a reduction in firms’ abnormal cash ratios. The IMBR DUM variable is also signif- icantly negative implying that firms with high agency costs before the offering experience a greater decline in excess cash ratios than other sample firms. Furthermore, the decline in abnormal cash ra- tios for these firms is related to the new debt issued, which sug- gests that debt plays an especially important role in reducing cash overinvestments for firms classified as being poorly managed.

We also test whether our sample firms overinvest in real assets relative to the industry benchmark by estimating the following regression:

Table 5 Regression results of raw cash ratio and industry-adjusted changes in cash ratio around debt introductions.

Cash ratio Industry-adjusted change in cash ratios

(1) (2) (3)

Intercept 0.223 �0.426 �0.471 (0.012) (0.099) (0.080)

DO DUM 0.306 (<0.001)

Debt issued �0.173 �0.132 (0.003) (0.036)

MBR 0.062 (<0.001)

IMBR DUM �0.115 (0.045)

IMBR DUM*Debt Issued �0.190 (0.024)

Ln(Assets) �0.003 �0.006 0.021 (0.708) (0.862) (0.476)

CFR 0.562 �0.425 �0.395 (<0.001) (<0.001) (<0.001)

NWCR �0.298 �0.293 �0.281 (<0.001) (<0.001) (<0.001)

CEXPR �0.655 �0.978 �0.992 (<0.001) (<0.001) (<0.001)

STDEV 0.288 (0.521)

RDSR 0.037 �0.042 �0.038 (0.414) (0.367) (0.412)

DDUM �0.041 0.012 �0.021 (0.114) (0.836) (0.710)

IDUM NR NR NR TP DUM NR NR NR N 620 294 294 Adj. R2 (%) 53.87 28.11 26.30

The sample consists of unlevered firms that subsequently conducted debt offerings between 1968 and 2001 and the matching firms defined as median firms in the same three-digit SIC code. An unlevered firm is defined to have conducted a debt offering if it issues long-term debt after three years of no debt and maintains a long-term debt to net asset ratio of five percent for three years subsequent to the offering. A firm has issued debt if the difference in long-term debt relative to the previous year exceeded five percent of the pre-issue book value of assets and net asset ratio is book assets less debt. Cash ratio, CR, is cash and marketable securities divided by book assets less cash. DO DUM is a dummy variable that takes the value of 1 if the sample firm conducts a debt offering and 0 otherwise, debt issued is the change in the long-term debt to net asset ratio from before to after the offering, MBR is the Market-to-Book ratio, defined as book value of assets plus the difference between market and book value of equity divided by book assets, IMBR is a dummy variable that takes the value 1 if the firm’s MBR is less that the industry benchmark in the two years immediately preceding the debt offering and 0 otherwise, Ln(assets) is the natural log of book value of assets less cash, CFR is the cash flow ratio defined as earnings before depreciation and amortization less interest, taxes, and preferred and common dividends standardized by assets less cash, NWCR is net working capita

The post Journal of Banking & Finance 34 (2010) 324–335 appeared first on graduatepaperhelp.

 

"Looking for a Similar Assignment? Get Expert Help at an Amazing Discount!"