The Effect of Capital Structure when Expected Agency Costs are Extreme
Harvey, C.R. Lins, K.V. Roper, A.H.
Journal of Financial Economics 74 (2004) 3-30
RESEARCH MOTIVATION
The objective of this paper is, using international evidenceto investigate whether debt can mitigate the effects of agency and information problems.
Prior theoretical research has shown that debt can be used to align managers’ interest. More specifically, when a meaningful conflict exists between outside shareholders and management due to the separation of ownership and control, debt helps to discourage overinvestment of free cash flow by self-serving managers (Jensen Meckling [1976], Jensen [1986] Stulz [1990] and else). Moreover, even without conflicting interest, debt gives management the opportunity to signal its willingness to pay out cash flows or be monitored by lenders or both, and thus to show that they do not or will not overinvest (Ross [1977] and else).
Benefits to debt could be greater (1) when management has a large base of assets in place that it can exploit, because assets in place generate cash flow that can lead to either overinvestment or the outright diversion of corporate funds (Jensen [1986] and else) or (2) when management has few future growth opportunities, because when a firm has expected future growth opportunities, debt can limit management’s ability to pursue positive net present value projects, leading to ex-post underinvestment (Myers [1977]).
Harvey et al adopted a refined setting. In particular, they focused on emerging markets and international debt markets. In emerging markets, managers and families routinely employ pyramid ownership structures to give themselves control rights that far exceed their proportional cash flow ownership, and consequently there potentially exist extreme agency costs. Also, in international debt markets, financial disclosure standards are higher, creditor rights are more effective and environment for monitoring is better, with respect to domestic debt markets. Taking together, if debt can mitigate the agency cost, the results are manifested in above mentioned refined setting.
DATA and RESEARCH DESIGN ISSUE
Harvey et al used 1014 exchange- listed non-financial firms in 18 emerging markets, but most of the tests are done on a sample of 252 firms with both a debt issuance and an identified ownership structure. In terms of the debt markets, they examine Foreign Bond Markets, Eurobond Markets, Internationally Syndicated Bank Issues, and Domestic Public and Private Bond Markets separately.
Harvey et al introduced a measure, Cash Flow Rights Leverage, which is the degree of separation between managerial cash flow rights ownership and control. It is defined as the ratio of the management group’s control rights to its cash flow rights.
Since this paper focused on international evidence, the data resources are different from what we conventionally used. In particular, Ownership Data is from one of the co-author (Lins [2003]), monthly stock return is from Datastream, debt issue is from Global New Issue Database (Security Data Corporation) , internationally syndicated bank issue is from Loanware Database (Capital Data), firm financial data is from Worldscope, and stock daily price is from FACTSET (Global Securities Price Database).
RESULTS
1.Summary statistics (Table1) shows that the cross-sectional sample has a mean debt-to-assets ratio of 0.28, broadly consistent with the values for firms in the United States and other developed countries (Rajan and Zingales [1995]); the average short-term debt to total debt ratio is 0.60, consistent with prior research Demirguc-Kunt and Maksimovic [1999] that firms in developing countries have substantially higher short-term debt ratios than firms in developed countries; mean cash flow rights leverage is 2.13, implying that the average management group in these emerging markets is able to turn one cash flow right into more than two control rights; firms with a debt issuance are larger, more levered, and have a lower proportion of short-term debt to total debt.
2.Cross-sectional analysis
The objective of cross sectional analysis is to draw inferences about the relation among firm value, leverage, and the separation of management group control and ownership, while controlling for a number of other factors.
In order to mitigate endogeneity issue, Harvey et al used 3SLS regression approach. Equation 1 is valuation equation: the dependent variable is Tobin’s Q; the independent variables are Debt-to-Assets and Cash Flow Rights Leverage * Debt-to-Assets; the control variables are Size, Capital Expenditure to Assets (proxy for growth opportunities), Country Dummy and Industry Dummy. Equation 2 is leverage equation: the dependent variable is Debt-to-Assets; the independent variables are Tobin’s Q and Cash Flow Rights Leverage; the control variables are Size, Percentage of Tangible Assets, Profitability and Country Dummy. Equation 3 is ownership equation: the dependent variable is Cash Flow Rights Leverage; the independent variables are Tobin’s Q and Debt-to-Assets; the control variables are Size, Percentage of Tangible Assets, risk proxy (Beta and Standard Deviation of Stock Return) and Country Dummy.
The results are presented in table 2. For equation 1, there is a significant negative coefficient on Cash Flow Rights Leverage, suggesting that an increase in the separation between managerial control rights and cash flow rights is negatively related to firm value. This further suggests that a management group could positively impact firm value by selling its interest in an indirect holding and using these proceeds to purchase shares directly. The coefficient on Cash Flow Rights Leverage * Debt-to-Assets is significant and positive, suggesting that debt plays positive role in alleviating agency problems. However, it shows that the positive coefficient on Cash Flow Rights Leverage * Debt-to-Assets is not big enough to offset the negative impact of debt. For Equation 2, coefficients on Size and Cash Flow Rights Leverage are significant and positive, suggesting that these two variables are positively correlated with leverage. It is also found that Tobin’s Q and Asset Tangibility and Profitability are not significant in the regression, suggesting that they are not correlated with leverage (at least in this sample and under this research design). For Equation 3, the coefficients on Debt is significant and positive, and the coefficients on Size is significant and negative; however Tobin’s Q, Asset Tangibility and Stock Ownership Risk lost significance in the regression (once again, at least in this sample and under this research design).
Harvey et al also provided results from OLS regression. The results are not qualitively different from 3SLS regression, consistent with the hypothesis that debt is able to mitigate agency cost. However, this is just a robustness check. As we discussed above, due to the endogeneity problem, OLS model most likely is misspecified.
Further, Harvey et al partitioned the sample into firms with above- and below-median levels of the proxies for assets in place (percentage of tangible assets) and growth opportunities (Tobin’s Q), to test whether the ability of debt to mitigate managerial agency costs depends on how likely a firm is to face overinvestment problems. The results are presented in Table 3 Panel A and B. It shows that significant positive coefficient on the interaction term between leverage and the cash flow rights leverage in equation 1 only exits in high percentage of assets in place subsample, and that significant positive coefficient on the interaction term between leverage and the cash flow rights leverage in equation 1 only exits in few growth opportunities subsample. These results are consistent with the hypothesis that debt is particularly effective at alleviating agency problems when firms are likely to suffer from overinvestment.
Harvey et al also partitioned the sample into firms with above- and below-median levels of short-term debt to total debt to test whether short-term debt is best suited for reducing the effect of potential overinvestment problems. Unreporeted results show that there is weak evidence that short-term debt is able to mitigate agency problems.
3.Event-study analysis
The objective of event-study analysis is to test the relationship between agency costs and the issuance of specific debt contracts.
Harvey et al used Market Model to perform the event study. Event date is the issue date of the debt financing agreement; Market return is Morgan Stanley market capitalization-weighted indices; Beta is calculated using Scholes and William [1977] model over an estimation window beginning 120 trading days before and ending 20 trading days before the issue date; CAR is calculated for a six-day event window that includes one day prior to issue, the issue date, and four subsequent trading days.
One important assumption of this event-study is that the value of debt comes from increased disclosure or monitoring or both, which together serve to reduce agency costs. However, prior research, for example Smith [1986] shows that at least some, if any, of the value enhancement attributed to debt could simply reflect information about exogenous changes in expected cash flows. As a result, we need to interpret this result by caution.
As discussed above, Harvey et al investigate four different debt markets, namely Foreign Bond Markets, Eurobond Markets, Internationally Syndicated Bank Issues, and Domestic Public and Private Bond Markets. These debt markets are different in terms of Disclosure Requirements, Creditor Rights, Enforceability of Creditor Rights, Creditor Base, Covenants and Monitoring (see figure 1 for detailed comparison). These debt markets are also different in terms of how to alleviate agency costs. For example, Yankee Bonds has stringent disclosure requirement and firms issuing debt at this market can send effective signal on their disclosure quality; similarly, Private Debt market, internationally syndicated bank contracts in particular, has effective monitoring, so firms issuing debt at this market can show their willingness to be monitored.
Description statistics is presented in Table 4. It is shown that firms issue debt at international bond market on average are larger, and the debts issued are larger and longer-dated than internationally syndicated bank issues. It is also shown that internationally syndicated bank market nevertheless remains a vital source for external capital for emerging market firms, in terms of the number of issues.
Regression results are presented in Table 5. It is shown that privately placed domestic bond issues lead to an average abnormal return of -1.04%, indicating that domestic bonds do not appear to provide certification benefits. It is also shown that internationally syndicated bank term loans generate average abnormal returns of 0.52% for outside shareholders.
4.Sequencing debt issue
There are two extant hypotheses: Certification Hypothesis — If debt creates value because of increased financial disclosure (being certified), then abnormal returns should be higher for debt contracts with higher levels of disclosure (for example in either the U.S. Yankee bond market or Eurobond market. Recontracting Hypothesis — If debt creates value by allowing managers to demonstrate their commitment to meet debt service requirements and to abide by covenants, then abnormal returns should be higher for debt contracts with higher levels of monitoring (for example in Internationally syndicated bank loan issues).
The Results are presented in Table 6. Firms whose initial international debt offering is in either the Eurobond or Yankee bond market experience a significant positive average CAR of 0.97%. This result may be explained by the certification hypothesis. When a firm’s subsequent international issue is an internationally syndicated term loan, there is a significantly positive average CAR 0.94%, and 58% are positive. As it is also shown, initial international issues of term loans have an average CAR -.67%. Therefore, it shows that subsequent international issues of term loans have a 1.61 percentage point higher average CAR. This result is consistent with the recontracting hypothesis that investors will pay more for firms that show they can comply with bank loan agreements. Privately placed bonds have relatively low out-of-pocket costs and are unlikely to transmit proprietary information. However, firms whose initial international issue is a privately placed bond do not have significant positive average CAR. Across both public and private bonds, subsequent international bond offerings do not have significant positive average abnormal returns. Possible explanation is that the creditor base is too diverse to credibly commit to effective monitoring.
5.Debt value creation and agency problem
As it is shown in above section, the positive returns exist for initial international issues in the public U.S. or Eurobond markets and subsequent international offerings of internationally syndicated term loans. Therefore, these two become the focus in this section.
The results of univariate analysis are presented in Table 7. The samples are partitioned by Above/Below median cash flow rights leverage, Above/Below median percent tangible assets and Above/Below median Tobin’s Q. There is no evidence that the abnormal returns associated with initial public U.S. and Eurobond offerings are correlated with cash flow rights leverage (Panel A). The abnormal returns associated with subsequent offerings of internationally syndicated term loans are positively related to cash flow rights leverage (Panel B) but the result indicates that firms with potentially extreme managerial agency problems benefit most from subsequent international offerings of syndicated term loans. It is also shown that the value created by debt is concentrated in firms with an above-median percentage of tangible assets or below-median Tobin’s Q values (Both in Panel A and in Panel B).
Harvey et al also conducted multivariate analysis (GLS regression). The intuition is straightforward: if shareholders benefit from debt contracts because they lessen the effect of agency problems, then we should expect the CARs of a debt contract to be positively related to the proxies for agency problems; if debt contracts add value primarily because they transmit new information about a firm’s prospective cash flows, then debt contract CARs should not be related to agency cost proxies. The results are presented in Table 8. It is shown that for initial international public U.S. and Eurobond offerings, there is no evidence that abnormal returns are associated with the extent of agency problems (negative in fact – caution!); for subsequent internationally syndicated term loans, the value created by subsequent syndicated term loan issues is positively correlated with cash flow rights leverage, and the value created by debt is also concentrated in firms with an above-median percentage of tangible assets or below-median Tobin’s Q values.
CONCLUSION
- Leverage helps mitigate the loss in firm value attributable to the separation of management control and ownership. This beneficial effect of debt is concentrated in firms that have either a relatively high percentage of assets in place or few growth opportunities.
- Subsequent issues of internationally syndicated term loans earn positive cumulative abnormal returns, these cumulative abnormal returns are positively related to the separation of control and ownership in the management group and this positive relation is concentrated in firms with a high percentage of assets in place or few growth opportunities. These results support the recontracting hypothesis that shareholders value compliance with monitored covenants, particularly when firms are most likely to face extreme agency problems.
- Initial issues of public international bonds lead to significant abnormal returns but these returns are not correlated with management ownership structures. Nevertheless, these gains are consistent with the certification hypothesis that international public markets certify firm quality with their substantial disclosure requirements.
Note: Definition of Pyramid Ownership — own a majority stake (hold more than 50% of the votes) in a company, which in its turn owns large stakes in other companies.