Toward A More Complete Model of Optimal Capital Structure

Journal of Applied Corporate Finance
S P R I N G 20 0 2 V O L U M E 15 . 1
Toward a More Complete Model of Optimal Capital Structure

by Roger Heine and Fredric Harbus,
Deutsche Bank Securities Inc.
17456622, 2002, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1745-6622.2002.tb00339.x by University Of South Africa, Wiley Online Library on [05/06/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TOWARD A MORE by Roger Heine and Fredric Harbus,
Deutsche Bank Securities Inc.*
COMPLETE MODEL
OF OPTIMAL CAPITAL
STRUCTURE
iability management plays an important but often overlooked

role in creating shareholder value. Classic capital structure
L theory tends to focus on the level of debt relative to total firm
value and typically balances the potential value of interest tax
shields against the expected costs associated with financial distress.1 Consistent
with this theory, most corporate finance practitioners understand the trade-off
involved in making effective use of debt capacity while safeguarding the firm’s
ability to execute its business strategy without disruption. But quantifying that
trade-off to arrive at an optimal level of debt can be a complicated and
challenging task. Moreover, there are a host of questions about the structure of
the firm’s debt—for example, whether the debt should be fixed or floating, long-
term or short-term, and denominated in local or a foreign currency—that must be
considered when determining a company’s value-maximizing capital structure.
In an article published in this journal in 1997, Tim Opler, Michael Saron,
and Sheridan Titman presented an economic model that simulates the effect of
different capital structure choices on shareholder value.2 The basic insight of the
model (henceforth referred to as “Opler et al.”) is that while judicious use of debt
can add value by reducing corporate taxes and strengthening management
incentives to increase efficiency, too much debt can result in underinvestment,
a loss of business, and perhaps a costly reorganization. The Opler et al. model
aims to identify the value-maximizing debt-equity ratio (as well as the optimal
percentage of fixed vs. floating)—one that does the best job of balancing the
value of the tax shield from debt against the expected cost of financial distress.
*This paper does not express the views of Deutsche Bank Securities Inc. or any of its affiliates.
1. See, for example, Richard Brealey and Stewart Myers, Principles of Corporate Finance, 6th Edition (McGraw-Hill, 2000)
for a comprehensive review of traditional capital structure theory.
2. Tim Opler, Michael Saron, and Sheridan Titman, “Designing Capital Structure to Create Shareholder Value,” Journal
of Applied Corporate Finance, Vol. 10 No. 1 (Spring 1997), pp. 21-32.
31
STERN STEWART JOURNAL OF APPLIED CORPORATE FINANCE
We have developed a capital structure optimi- customized as needed to address special topics such
zation model (the Deutsche Bank Liability Structure as acquisition funding strategies or the impact of
Model, or “DBLSM”) that expands on the work of employee stock option plans. The model has been
Opler et al. in several ways. We provide a more applied to dozens of companies in a variety of
comprehensive capital structure framework that is nonfinancial industries around the world.
integrated with a detailed set of interest rate and How does the model work? It begins by devel-
foreign exchange simulation models that have many oping a set of simulation paths with both operating
features in common with contemporary value-at-risk cash flow and market rate components. Specifically,
(VaR) techniques. We also model financial distress based on an industry and company-specific statisti-
costs in terms of several new components. In cal analysis, DBLSM simulates thousands of alterna-
DBLSM, the distress costs of too much debt include tive forecasts of monthly operating cash flows (as
the costs associated with missed investment oppor- approximated by EBITDA, or earnings before inter-
tunities when capital expenditures cannot be fully est, taxes, depreciation and amortization), typically
funded, the reduced operating cash flows and higher going out 20 years into the future. Then, using Monte
working capital needs that occur at lower ratings, Carlo simulation methodologies,3 DBLSM projects
and the potential signaling costs of cutting the monthly interest rates across the yield curve as well
dividend. DBLSM also allows for a broader set of as FX rates for all major currencies over the same 20-
capital structure decision variables. For example, year time horizon. The EBITDA, interest rate, and FX
where the model of Opler et al. produces a single projections together define our final simulation
optimal debt ratio, the output of DBLSM includes paths, where each path has a unique set of values for
both a “target” and a “fallback” debt structure. The these variables at each monthly time point. Once the
basic idea is that when a company is doing well and set of paths is established, it becomes the back-
its cash flows are strong, debt is paid down to a level ground environment against which a variety of
consistent with the firm’s target credit rating. But if alternative capital structure assumptions are applied
conditions worsen and cash flows turn down, man- and evaluated to determine which structure maxi-
agement is prepared to allow its leverage ratio to move mizes shareholder value.
to a temporarily higher level (corresponding to a The shareholder value calculation reflects both
lower “fallback” rating) in order to fund capital expected cost and risk components, with risk mea-
expenditures or dividends that might otherwise have sured in a variety of ways. We frequently find that debt
to be cut. Thus, the company’s target credit rating structure decisions, such as the fixed vs. floating mix,
determines its “permanent” debt capacity, while the end up affecting shareholder value as much as the
fallback rating can be thought of as determining the overall level of debt, at least over a broad range of
optimal amount of cost-effective reserve debt capacity investment-grade credit ratings. We also find that,
that the firm can draw upon when needed. especially for companies operating in volatile business
Besides producing target and fallback credit segments, significant value may be added by building
ratings, DBLSM can be used to simulate the effects in the financing flexibility afforded by a lower fallback
of using floating versus fixed-rate debt, varying the rating in bad times. In this sense, a company’s key target
maturity and currency structure of future debt issues, credit ratios like leverage and interest coverage are
designing different dividend and stock repurchase “dynamic”; that is, they are expected to vary over the
policies, and changing the level and composition of firm’s cash flow cycle as its rating fluctuates between
explicit liquidity reserves (as opposed to reserve the permanent and fallback levels.4
debt capacity)—all within the same integrated simu- In addition to this dynamic aspect provided by
lation framework. The standard model can also be the fallback rating concept, the DBLSM framework
3. More specifically, non-arbitrage constrained, stochastic methods. The 4. While credit ratios will correspond with the modeled ratings, we do not
techniques used to generate these simulation paths are similar to methodologies claim that the company’s actual rating moves in lock-step with the modeled rating.
commonly employed to value options, including the estimation of lognormal To better reflect the stickiness of actual credit ratings, however, our rating sub-
volatilites, cross correlations, and mean reversion. However, the models we model incorporates a two-year look-ahead feature: even if credit ratios at a point
employ are not forced to converge to arbitrage-free implied forward levels but can in time in DBLSM would indicate a higher rating, the rating will not be upgraded
instead reflect both near-term forecasts and long-term means based upon statistical unless, based on the average EBITDA trend line, the credit profile two years ahead
observations and judgment. As discussed later in the context of the fixed versus is also improved; similarly, a downgrade will not occur unless the profile is also
floating rate debt analysis, this allows us to build a risk premium into the rate worse two years ahead.
structure.
32
JOURNAL OF APPLIED CORPORATE FINANCE
We generally find that the additional cost of such [liquidity] insurance is fairly
modest in relation to “normal” capital costs, and quite small compared to the total
market capitalization of the firm.
also has a “liquidity insurance” component that MEASUREMENT OF SHAREHOLDER VALUE

distinguishes it from classic capital structure models. UNDER “NORMAL” VOLATILITY
In most approaches, optimal capital structure is
based on the assumption of “normal” operating cash The value of a levered firm is equal to the sum
flows and business conditions that reflect the known of the value of an unlevered firm plus the net benefit
historical ranges of interest rate and foreign ex- of debt. In the DBLSM framework, the focus is on
change rate volatility. But particularly since 1998, the valuation impact of the incremental cash flows
companies have increasingly encountered sudden deriving from debt and its structure, and thus we do
and severe operating problems that have been not need to know the value of either the levered
compounded by a loss of financial flexibility and firm or the unlevered firm to value the impact of the
liquidity. Unable to raise unsecured debt in the level and the structure of any debt financing. In fact,
capital markets at any reasonable cost, companies we can avoid the estimation challenges that would
have watched as the related loss of equity investor otherwise need to be addressed (including determi-
confidence has led to the evaporation of billions of nation of the appropriate market risk premium and
dollars of shareholder value (over and above the company beta, if a CAPM framework is used) to
losses stemming from the operating problems them- derive the correct discount rate for valuing the
selves). A key question, therefore, is how much back- company’s operating or equity cash flows. And
up liquidity must be in place to avoid a funding although the right discount rate to apply to the debt-
shortfall under these circumstances. While we typi- related incremental flows is not precisely certain,
cally run DBLSM first under normal volatility assump- the range of potential error is much smaller than any
tions, a major benefit of DBLSM is its ability to estimate estimate of the equity discount rate. In DBLSM, we
the expected cost of alternative liability structures that discount the cash flows related to debt and its
can provide the liquidity insurance necessary to structure at LIBOR plus the company’s five-year
sustain the firm through periods of stress. credit spread as they evolve in our simulations.
In short, DBLSM designs a company’s capital While the use of the five-year credit spread versus
structure to maximize shareholder value under other maturity credit spreads is a matter for debate,
conditions of normal, ongoing volatility, but also it has no material impact on the relative capital
models the cost of a liquidity insurance policy structure valuations that emerge from our model.
against the possibility of highly improbable events What is important is the incremental present value
capable of shaking creditors’ confidence in the firm. of cash flows as we change the capital structure.
We generally find that the additional cost of such
insurance is fairly modest in relation to “normal” Benefits of Debt
capital costs, and quite small compared to the total
market capitalization of the firm. Tax shields. Finance theorists have long recog-
In the pages that follow, we start by reviewing the nized that the reduction in the corporation’s tax
benefits and costs of debt as modeled in DBLSM. Most liability due to the deductibility of interest expense
of the costs derive from market imperfections, such as is the most important financial benefit of debt in the
the prohibitive costs of issuing equity in a downturn, capital structure decision process.5 To model the tax
or bond credit spreads in excess of expected default shield of debt, we compare the tax liability incurred
losses. Following this discussion of the model’s with debt to the liability incurred if the company had
inputs, we then turn to the output of the model—that no debt. Of course, interest expense can shield
is, the primary “decision” variables, first under “nor- income only to the extent that the company has
mal” volatility conditions and then when taking income to be sheltered. Loss carryforwards, the
account of the additional liquidity requirements just Alternative Minimum Tax, and accelerated tax de-
discussed. The Appendix furnishes additional detail preciation all can reduce the value of the interest tax
about the workings of the DBLSM model. shield. In some cases it is also necessary to factor in
5. See Franco Modigliani and Merton Miller, “Corporate Income Taxes and the 1980); and Michael Bradley, Gregg Jarrell, and Han Kim, “On the Existence of an
Cost of Capital: A Correction,” American Economic Review, Vol. 53, pp. 433-443 Optimal Capital Structure,” Journal of Finance (Vol. 39 No. 3), pp. 857-878 (July
(1963); Merton Miller, “Debt and Taxes,” Journal of Finance, Vol. 32, pp. 261-276 1984). See John Graham, “Estimating the Tax Benefits of Debt,” Journal of Applied
(1977); Harry DeAngelo and Ronald Masulis, “Optimal Capital Structure under Corporate Finance Vol. 14 No. 1 (Spring 2001) pp. 42-54, for a detailed discussion
Corporate Taxation,” Journal of Financial Economics, Vol. 8, pp. 5-29 (March on the ways in which real world constraints impact the value of the tax shield on debt.
33
VOLUME 15 NUMBER 1 SPRING 2002
the tax loss generated by the exercise of employee is dangerous, but the flexibility to fund expenditures
stock options, even though this loss is not recog- temporarily using more debt can be beneficial. That
nized for financial accounting purposes. Prior to is, when companies cannot fully fund value-adding
2001, there were a number of high-tech companies capital projects and (in some cases) maintain the
whose entire taxable income was offset with such tax common stock dividend, shareholder value can be
losses despite having no net debt. lost. If we assume that the cost of issuing new equity
While debt is tax advantaged at the corporate is high, it may be advantageous for the company to
level, the relative advantage of debt over equity is increase debt temporarily to fund the shortfall and
typically reduced once both corporate and investor avoid the opportunity cost of not making the invest-
taxes are considered. The differential tax rates on ment (or maintaining the dividend).
interest income, dividend income, and capital gains To illustrate this point, let’s consider a company
that are specified in each country’s tax code are key for which a mid-BBB rating7 represents the optimal
drivers of the net tax benefit of debt. For example, balance between the present value of any tax shields
with the capital gains rate typically lower than the and the expected distress costs of debt. If there is some
rate on dividend or interest income, the higher the possibility that Capex cannot be fully funded out of
proportion of earnings retained rather than paid out volatile operating cash flow, DBLSM may indicate an
as dividends, the lower is the net benefit of debt. This optimal target rating of BBB+ with a fallback of BBB-.
is true for classic tax systems where interest expense This means that the extra debt capacity (more than
is tax-deductible but dividends are not.6 However, what would otherwise be considered optimal) that is
many countries (including the United Kingdom, built up in good years (consistent with a BBB+ rating)
France, Germany, Australia, Singapore, and New is expected to be drawn upon in bad years (to the
Zealand) have imputation systems that effectively point where the rating is allowed to fall to BBB-). The
eliminate some or all of the double taxation of willingness to tolerate this fallback level is based in
dividends, further reducing or even completing ne- large part on the probability that the firm will generate
gating the tax advantage of debt financing. In addition sufficient cash flow to repay the temporary debt
to the tax system’s structure, the other main factor quickly enough to avoid the higher costs of servicing
driving the effectiveness of the tax shield is the tax debt with lower ratings. In general, the more volatile
profile of the investor base, including the percentage are the company’s operating cash flows, the higher
of nontaxable equityholders and bondholders. the target rating must be to preserve the debt capacity
As an example, consider the Dutch tax system. necessary to finance Capex. At the same time, though,
Until 2001, the generally zero capital gains rate the higher is the rate of mean reversion in operating
resulted in little or no advantage to debt versus cash flows, the lower is the fallback rating that can be
equity for typical dividend payout ratios. But new tax tolerated because of the higher probability that the
laws that took effect in 2001 provide that investors temporary debt will be rapidly repaid.
are now taxed on the value of their investments In our analysis of shortfalls of Capex or divi-
rather than on their investment income—essentially, dends that cannot be covered out of operating cash
a wealth tax. This makes the form of distribution flow, we assume that the company will always
(whether interest, dividends, or capital gains) irrel- choose to issue incremental debt rather than new
evant to investors, and the interest deduction at the equity. Of course, companies can issue equity when
corporate level once again advantageous, increasing their operating cash flows are down, but typically
the attractiveness of debt to Dutch companies. only at a price that represents a sharp discount to a
Temporary use of debt to finance capital ex- likely already depressed stock. According to efficient
penditures or dividends. Debt cuts both ways with markets theory, companies should generally be able
respect to funding capital expenditures (“Capex”) to issue equity at the current fair value without
and dividends: too much debt on a permanent basis affecting the traded value of the company. In reality,
6. In his classic 1977 paper (cited earlier), Merton Miller shows Practitioners commonly assume that the effective tax rate faced by both
that the value of the tax shield per dollar of debt can be calculated as: equity- and debtholders is equal, in which case the value of the tax shield
G =1 - [(1 - t c )(1 - t e )/(1 - t d )], where G is the relative gain on debt per dollar of debt reduces to t c.
vs. equity, t c is the effective corporate tax rate, and t e and t d are the 7. While our rating notation uses S&P designations, our discussion is
effective tax rates on equity distributions and interest, respectively. understood to represent the Moody’s-equivalent rating as well.
34
Too much debt on a permanent basis will cause heavy debt service obligations that
raise the probability of missing Capex and dividends.
however, information is limited and investors are Similarly, we assign a cost to the first time that
suspicious about both managers’ view of the a company must cut its dividend. Academic research
company’s prospects and the intended use of the has clearly shown the negative impact of dividend
proceeds from an equity offering.8 Thus, a secondary cuts on share prices, with typical findings ranging
issuance of equity to cover shortfalls generally meets from a 1% price decline for dividend reductions to a
with an adverse market reaction for several reasons: 7% decline for complete omissions.11 For companies
If the company is short of operating cash to fund with a long track record of stable or increasing
normal Capex or to maintain the dividend, it is likely dividends, we typically assume a penalty towards the
that the stock price is already depressed and the upper end of this range and sometimes even higher
company’s debt spreads have widened. Some portion (what would happen, for example, if GE cut its
of the equity issued in this situation will go to shore dividend?). But for companies with a limited dividend
up the creditors’ position, transferring value from the track record or volatile dividend payments, we typi-
shareholders to the creditors.9 More importantly, cally assume a penalty close to or equal to zero.
while the market is likely to accept the story that high-
flying growth companies need extra equity to expand, Costs of Debt
it tends to be more skeptical of claims by managers of
depressed companies that they will prudently apply Crowding out of Capex and dividends. We
the new money to rebuild the business. In other have already discussed the dual role of debt with
words, agency and information costs are at a peak respect to Capex and dividend funding. Too much
when the company’s financials are weak. debt on a permanent basis will cause heavy debt
Except for companies with prospects of rapid service obligations that raise the probability of
growth, issuance of equity may imply that manage- missing Capex and dividends. These in turn lead to
ment feels the stock is overvalued, or that the the shareholder value penalties just described.
company’s relationship with potential lenders has Interest costs in excess of the floating borrow-
deteriorated.10 In either case, an equity issue will not ing rate of an AAA-rated company.We treat interest
be warmly received by the market. costs in excess of the floating borrowing rate of an
How does our model quantify the expected cost AAA-rated company as a cost to shareholders. AAA
of missed investments or dividends? When cash flow floating borrowing rates are used as our base, or
is inadequate to fund Capex requirements—even “riskless,” rate because U.S. Treasury or other direct
after using the incremental debt capacity provided government borrowing rates, which are generally
by the fallback rating—we assume a shareholder lower than AAA corporate rates, have special liquidity
value loss equal to a percentage penalty factor times and regulatory benefits that go beyond credit risk. The
the unfunded Capex. For example, a typical penalty shareholder cost derives both from the credit spread
factor for companies trading at P/E multiples in the incurred on debt and from fixing interest rates when
upper teens might be 25%, meaning that for every subsequent floating rates turn out to be lower.
$100 of unfunded Capex, shareholders incur a $25 The argument that the credit spread is a cost is
cost at the time of the Capex shortfall. For growth somewhat controversial.12 While corporate finance
companies trading at much higher P/E ratios, the managers instinctively treat the credit spread on debt
penalty would typically be higher. For example, a as a cost, some academics argue that the spread fairly
pharmaceutical company that has finally won FDA compensates investors for their risk of loss from
approval on a blockbuster drug faces enormous bankruptcy net of expected recoveries, and thus does
opportunity costs if it cannot fund the Capex neces- not represent a true cost to shareholders. However,
sary to produce and distribute the drug. we believe that for investment-grade companies, the
8. See Michael Jensen, “Agency Costs of Free Cash Flow, Corporate Finance, and 11. For a review of dividend policy signaling literature, see Chapter 7 of
Takeovers,” American Economic Review, Vol. 76 No. 2, pp. 323-329 (May 1986). Dividend Policy—Its Impact on Firm Value, by Ronald Lease, Kose John, Avner
9. See Stewart Myers, “Determinants of Corporate Borrowing,” Journal of Kalay, Uri Loewenstein, and Oded Sarig (Harvard Business School Press, 2000).
Financial Economics, Vol. 5 No. 2, pp. 147-175 (Nov. 1977). 12. Surprisingly, there seems to be little academic research directly on the
10. See Stewart Myers and Nicholas Majluf, “Corporate Financing and subject of the credit spread cost to shareholders.
Investment Decisions When Firms Have Information that Investors Do Not Have,”
Journal of Financial Economics, Vol. 13 No. 2, pp. 187-221 (June 1984).
35
credit spread has generally been much greater than better. When an A-1/P-1 issuer is downgraded, it
the level implied by actual bankruptcy losses net of must typically pay a very high premium to crowd out
recoveries.13 This result14 could have much to do with other A-2/P-2 issuers so as to attract a sufficiently
the fact that, in recent years, there have been several large portion of the limited A-2/P-2 market. Similarly,
notable cases in which a major company has experi- other major fixed-income investors such as insur-
enced rapid and dramatic credit deterioration when it ance companies, pensions, and bond funds face
came to light that the company was taking on much restrictions on the quantity of lower-rated term debt
more risk or incurring more extensive losses than they can hold. As the issuance of lower-rated paper
had previously been publicized or disclosed in its increases due to downgrades, the credit spreads on
financial statements. Having been blindsided in the all debt at the lower end of the market are bid up.
past by such cases, investors have little way of knowing Based on these considerations of adverse selec-
which apparently strong companies are misleading tion and institutional supply constraints, we model
them. Hence, investors must charge all companies a the entire credit spread above the AAA borrowing
risk premium to cover the expected cost of these losses. rate as a cost to shareholders (while conceding that
This credit premium “surcharge” should be at least some portion of the spread is there to cover
considered a cost to the shareholders of those net default risk). We justify using the entire spread
companies that are not misleading their creditors— with the argument that most investment-grade firms
a category of costs that economists call “adverse are evaluated as ongoing concerns by their share-
selection.” Auto insurance provides a good example holders. In other words, shareholders generally want
of such costs. Insurance companies have limited management to maximize returns contingent upon
ability to identify bad drivers prior to the occurrence the firm surviving catastrophic events, and the cost
of an accident, and many jurisdictions prohibit insur- of protecting against such an improbable risk is
ance companies from charging differential premiums simply a cost of business. Viewed from another
based upon certain types of statistical evidence. perspective, few analysts are likely to add back any
Therefore, the extra insurance premium that good of the credit spread incurred on debt to arrive at a
drivers pay to cover the losses incurred by bad drivers “truer” economic earnings profile for a company.
should be considered a cost to the good drivers. The Business Effects of Credit Downgrades.
Another source of the excess credit spread is a When a company’s credit deteriorates, particularly
limit on the supply of institutional credit. As a below investment grade, the company finds it increas-
company’s credit standing deteriorates, fewer insti- ingly difficult to negotiate favorable payment terms
tutions are ready or even permitted to invest in with its suppliers. As payments are accelerated, the
weaker credits. In an efficient market, all investors payable account declines and must be replaced by
would be able to invest in companies of any credit interest-bearing debt. Interest costs on incremental
quality but would adjust their required return to working capital are the indirect result of lower credit
reflect the risk of the individual company. While we ratings, and DBLSM explicitly models the cost of this
agree that this is largely true of major equity markets, additional interest, net of any tax shield, by increasing
the supply of fixed-income funding is vastly greater working capital requirements as ratings are lowered.
for strong companies than for weak companies. A Besides affecting supplier credit availability,
well-known example is the commercial paper mar- low ratings also clearly affect relationships with
ket, where the supply of credit to A-2/P-2 issuers is customers and employees. For example, manufactur-
only about 5%-10% of the supply to A-1/P-1 issuers, ers of cars or airplanes may lose business to more
largely because money market funds must, by regu- creditworthy competitors if customers feel the com-
lation, restrict most of their investments to A-1/P-1 or pany will not be around to service its product. Natural
13. Using Moody’s data for annual and two-year default probabilities comprehensive analysis of this question using a calibration approach on a
for cohorts from 1997 through 2000 and simplified modeling assumptions, variety of structural models for credit risk, see Jing-zhi Huang (Pennsylva-
we estimated what credit spread would be required to cover the risk of nia State University) and Ming Huang (Stanford University), “How Much of
default (net of recoveries) as a function of credit rating. This hypothetical the Corporate-Treasury Yield Spread is Due to Credit Risk?: A New
spread was then compared to the observed spread. We found that, down Calibration Approach,” working paper (April 2002).
to mid-BBB ratings, the actual spread is significantly greater than the 14. One could argue that the excess credit spread may be at least partly
“default only-required” spread, with the default component accounting for attributable to systematic risk, which could imply that portion of the excess spread
somewhere between only 0-20% of the total observed spread. For a so attributed is not a cost to shareholders.
36
resource exploration and production companies, bankruptcy event. The Enron case provides a good
despite producing a commodity product, may lose illustration. As counterparties and employees de-
access to reserves if a low credit rating prevents fected, immense intangible value, particularly re-
them from entering into long-term contracts. Com- lated to the energy trading business, was lost. This
panies whose principal asset is human capital may is the present value of the expected future cash flows
lose valued employees who depart for more secure that would have materialized absent the bankruptcy.
firms.
To model this loss of business as a function of PRIMARY DECISION VARIABLES AND
credit rating, DBLSM assumes that an increasing TYPICAL FINDINGS
percent of EBITDA is lost as ratings decline, particu-
larly if they drop below investment grade. The loss Now that we have summarized the key input
factors are estimated based on industry analysis and components of DBSLM, let’s turn to the output
discussions with the company. DBLSM then charges recommendations for capital structure that the model
the present value of lost EBITDA in each simulation is likely to generate. The decision variables that can
month to shareholder value. be changed by management to maximize share-
Bankruptcy Costs. The increased working capi- holder value include not only the leverage ratio but
tal requirements and lost EBITDA are financial both the target and fallback credit ratings, the
distress costs that grow as a firm’s credit quality optimal level of cash reserves, dividend and share
deteriorates. Actual bankruptcy costs (which include repurchase policies, as well as several more detailed
the costs of reorganizing a company whether inside aspects of debt structure such as maturity, currency,
or outside Chapter 11) are generally negligible in our and the mix of fixed vs. floating.
modeling unless the descent to bankruptcy from We start our discussion with the optimal target
investment grade is very rapid. and fallback ratings that determine the level of debt,
DBLSM incorporates a relatively simplistic lagged and then consider the three aspects of debt structure
model of the company’s share price15 and assumes just cited. The questions of optimal dividend policy
that all of the equity value is wiped out upon and cash reserve targets are discussed last.
bankruptcy. If the company has already slid to deep
non-investment-grade ratings prior to bankruptcy, Rating Targets
the share price model would typically indicate little
remaining shareholder value and thus minimal In applying our model to non-financial compa-
bankruptcy costs. Material bankruptcy costs arise nies across many industries, we have found optimal
only when the company has a “brittle” credit target ratings that run the gamut from mid-BBB to
standing that is based largely on cash flows and not AAA levels. While specific conclusions depend on
on tangible assets that other creditors would lend many individual factors, including the firm’s tax
against. In the event that cash flows are suddenly position, the principal determinants in a broad
inadequate to service debt and the quantity of debt sense are the size of Capex and dividends relative
already exceeds the limited secured borrowing to EBITDA, the volatility of EBITDA, and the
power of the assets, the company could experience expected costs associated with falling to lower and
a rapid meltdown, resulting in a dramatic loss in especially non-investment-grade ratings. All other
shareholder value. factors equal, higher Capex and dividends and
This is a separate effect from the loss of business higher EBITDA volatility dictate higher target rat-
discussed above. Whereas the latter captures losses ings, and thus lower leverage ratios, while compa-
in each simulated period due to lower ratings, the nies with lower Capex and dividends and lower
bankruptcy cost captures (by assuming the stock EBITDA volatility can support more debt (lower
price falls to zero) the present value of all lost cash target ratings).16 But there are many intermediate
flows that would have occurred subsequent to the cases with high Capex and dividends but low
15. The stock valuation model is Pt = aPt-1 + b (P/E) EPSt-1 where Pt is the 16. Leverage ratios span a range for a given credit rating, and this is reflected
price in month t, EPSt-1 is the EPS in month t-1, (P/E) is an assumed long term in our ratings sub-model as described in Step 7 of the Appendix.
P/E ratio (with volatility applied around an expected value), and a and b are
coefficients. We adjust a and b to produce results reasonably similar to historical
patterns of annual stock price volatility.
37
FIGURE 1 12%
EVOLUTION OF Actual LIBOR
ACTUAL LIBOR VS. LIBOR
ORIGINALLY PREDICTED Forward LIBOR
10% Forward Curve at April 1993
BY FORWARD CURVES1
LIBOR 8%
6%
4%
2% Example: In April 1993, the forward curve “predicted”

that LIBOR in January 1997 would be 6.68%, but the
actual LIBOR in January 1997 was only 5.69%
0%
Jan/01
Jan/91
Jan/11
Jan/05
Jan/95
Jan/02
Jan/92
Jan/03
Jan/12
Jan/93
Jan/96
Jan/09
Jan/89
Jan/90
Jan/99
Jan/00
Jan/06
Jan/07
Jan/08
Jan/97
Jan/98
Jan/94
Jan/04
Jan/10
Source: Deutsche Bank Debt Capital Markets and Bloomberg historical data.
1. As can be seen in the figure, during the periods in the last 13 years (for which data were readily available) when LIBOR
has significantly declined, the decrease has been far sharper than the forward curves would have predicted. But in the periods
when LIBOR increased, although forward rates also “underpredicted” the increase, the underprediction in this case was smaller
and more short-lived than the overprediction in the other direction, making forward LIBOR rates too high on average.
EBITDA volatility, or low Capex and dividends but Debt Characteristics

high EBITDA volatility.
In practice, DBLSM derives a majority of opti- Fixed versus Floating Mix. The optimal mix of
mal target ratings to be centered in the high-BBB to fixed- and floating-rate debt is frequently of great
high-A range. Based on client feedback, we believe interest to CFOs and corporate treasurers. Our
these findings represent stronger rating and hence interest rate simulations are constructed so that, on
lower leverage recommendations than results gener- average, the future evolution of LIBOR is lower than
ally produced by other, frequently weighted average that implied by forward LIBOR rates.17 (For historical
cost of capital-based, analyses. evidence in support of this approach, see Figure 1.)
For the fallback rating, we find that a gap of at We argue that borrowing at floating rates is
least one rating notch between the target and value-neutral, while borrowing at fixed rates exposes
fallback rating generally adds material value, as the company to potential economic losses or gains (if
compared to an inflexible capital structure that subsequent floating rates go below or above the fixed
allows no deviation from the target ratio. For ex- rate).18 One way to see this is that floating debt, prior
ample, if the target rating is A, a “two-notch” fallback to credit spread considerations, maintains par value
to BBB+ may be optimal. Generally, the greater the while fixed rate debt does not. When a company
volatility of EBITDA, the lower is the fallback rating. borrows fixed and rates decline, there is an unques-
Nevertheless, the model typically tries to leave tionable shift in value from shareholders to creditors.
breathing room between the fallback rating and non- Our basic philosophy is that companies should
investment-grade territory to avoid the costs associ- not incur the risk premium associated with fixed-rate
ated with higher credit spreads, lost business, and financing unless such financing is the most cost-
higher working capital needs that occur at the lower, effective means of reducing other substantial risks
particularly non-investment-grade, ratings. facing the firm. One such possible risk is a spike in
17. Our approach of assuming that the yield curve embeds a risk premium is 18. Once again, the focus of our model is non-financial companies. Other
consistent with the liquidity-preference theory of the term structure of rates. companies with significant financial assets would incur risks of rating downgrades
and financial distress should the gap between asset and liability durations become
excessive.
38
For the fallback rating, we find that a gap of at least one rating notch between the
target and fallback rating generally adds material value, as compared to an inflexible
capital structure that allows no deviation from the target ratio.
FIGURE 2 2.0
CHEMICAL COMPANY:
ANNUAL CHANGE IN 1.5
Year-over-year change in EPS

DILUTED EPS (IN $) 1.0
0.5
0.0
-0.5
-1.0
-1.5
-2.0 Actual Change
-2.5 Change with 100% Instead of 30% Assumed Floating
-3.0
1982 1984 1986 1988 1990 1992 1994 1996 1998 2000
Source: Company data and Deutsche Bank analysis.
interest rates that hurts coverage ratios enough to in EPS would have been with 100% floating debt
jeopardize the firm’s credit rating. Lower ratings could compared to the actual level, which often tends to be
in turn cause a real loss in shareholder value due to in the 20% to 30% range. Figure 2 outlines such an
the higher credit spreads, working capital require- analysis for a large chemical company for EPS changes
ments, and loss-of-business effects already discussed. over the 1983-2001 period. As can be seen in the figure,
A second risk that can be managed with fixed- going from an assumed 30% floating to a hypothetical
rate debt is more accounting-oriented—namely, the 100% floating would have slightly reduced the annual
possibility that greater use of floating-rate debt may change in EPS in most years. This reflects the positive
increase the volatility of reported earnings and EPS. correlation between earnings and LIBOR, such that
While many analysts believe that companies with when earnings rise, the increase is slightly offset by
higher EPS volatility trade at lower multiples, we do higher interest costs, and when earnings fall, the
not attempt to assign a price tag to such volatility. In decrease is slightly muted by lower interest costs. The
theory, shareholders should not care about interest standard deviation of the actual annual EPS changes
expense or even EPS volatility so long as the actual over the period is $1.055, but this would have dropped
risks of rating downgrades or financial distress do to $1.019 had debt been 100% floating.
not increase. In fact, we find that for most moderate We have found in the fixed versus floating
to strong investment-grade borrowers, high levels of analysis that a floating percent at or near 100% is often
floating debt do not materially increase either the optimal for many investment-grade industrial firms.
probability of rating downgrades or EPS volatility. The The fact that many investment-grade industrials
typical volatility in interest rates, when applied to the target only 20-30% floating, when they would have
relatively modest quantity of debt in an investment- greater shareholder value at higher percentages,
grade company, tends to be “washed out” by the basic may perhaps be symptomatic of a mistaken focus on
operating cash flow volatility of the company. Also managing interest rate volatility in isolation rather
important in mitigating volatility is the fact that in than on the expected cost of debt and overall cash
many industries, including natural resource explora- flow volatility.
tion and production, paper, chemicals, and other On the other hand, we also see situations where
industrial product companies, there is a positive very low levels of floating-rate debt (high levels of
correlation between the change in cash flow and fixed-rate debt) is optimal. This tends to occur when
changes in LIBOR. And to the extent that such interest coverage is already quite weak and escalat-
companies’ operating cash flows move up and down ing rates can put the company at significant risk of
with changes in interest rates, the use of floating-rate a further downgrade. Also, high levels of floating-
debt actually reduces earnings volatility. rate debt can materially increase cash flow and EPS
To illustrate this point, we sometimes “re-run” volatility when a company’s EBITDA is very stable
the history of a company to estimate what the change or, worse, negatively correlated with interest rates.
39
An analogy with personal finance may be explain the earnings volatility attributable to naked
helpful in showing why the optimal levels of floating foreign currency borrowings. Interestingly, govern-
rate debt is frequently at one or the other extreme. ment or government-controlled entities seem more
Many oil and gas companies send a contract to their willing to take on foreign currency exposure because
customers once a year allowing them to lock in the they are not subject to the same scrutiny.
cost of fuel for the coming year. The offered contract DBLSM captures several effects of foreign cur-
prices often seem high compared to the expected spot rency borrowings:
price of the fuel. But customers who live from The lower or higher average cost of financing
paycheck to paycheck cannot afford to take the risk compared to the company’s home currency borrow-
of a large potential increase in their winter heating bill, ings, depending on the specification of the FX and
and often sign the contract to lock in this price. More interest rate models;
affluent customers, however, have a greater buffer in The lower volatility of shareholder value and EPS
their finances and are not very likely to sign the to the extent that the interest expense on foreign
contract. But it would be unusual for any customer to currency debt offsets the foreign currency earnings
conclude that hedging only, say, half of the future of foreign subsidiaries;
purchases is best. The greater diversification of funding costs. For
Maturity Considerations. DBLSM analyzes the example, it may be less risky to borrow in multiple,
trade-off between the higher credit spreads but not strongly correlated, currencies than in just one
lower rollover risk associated with issuing longer- currency; and
term debt and the lower average cost but higher The impact on ratings and hence borrowing costs
rollover risk of shorter-term debt. A few years ago, of both local and foreign currency debt. For ex-
we would typically have observed a more or less ample, if a company has significant foreign currency
break-even situation in which the higher credit debt employed to hedge net investments and the
spread on longer-term debt was largely offset by the foreign currency appreciates, the higher book value
expected cost of rolling over short-term debt at of the debt, though offset by translation of the net
potentially less favorable rates. That is, shareholder investment, will still cause important credit ratios
value differences were minor across the new-issue such as debt/total capital or funds flow from opera-
maturity spectrum. More recently, though, with a tions/debt to deteriorate.
general increase in corporate debt spreads, we find
on average a net cost to longer-term maturities. Dividend Policy
Hence, under conditions of “normal” volatility, the
company would be better off on average with Although DBLSM is able to incorporate various
shorter-term debt, even after considering the various types of dividend policies, most take the form of a
costs of financial distress captured by our model. minimum payout ratio and a target growth rate in
Nevertheless, an unusual event that is not dividends per share subject to a maximum payout
captured under normal volatility (for example, a ratio. DBLSM assumes that a company will not cut its
sudden legal liability, natural disaster, or unexpected dividend unless paying the dividend would require
reporting problem) can result in an abrupt withdrawal so much debt financing that ratings would fall below
of unsecured credit support from potential lenders and even the fallback credit rating. As discussed earlier,
a sharp increase in outstanding debt trading spreads. a shareholder penalty is imposed the first time a
We address shortly how to adjust capital structure to dividend cut is necessary.
protect against such “event risk” contingencies. The model reflects several implications of divi-
Currency Mix of Debt. Multinational corpora- dend policy:
tions commonly ask about the optimal currency mix The higher the dividend, the lower are stock
of their liabilities. In practice, companies rarely bor- repurchases, since share repurchases are funded out
row in foreign currencies when they cannot make use of any residual cash flow left after all Capex and
of hedge accounting, mainly because it is difficult to dividends are funded.19
19. In addition to dividend policy, DBLSM is able to address various shareholder negative effects for repurchases beyond certain levels where the company would
value aspects of share repurchases. For example, assumptions can be incorporated have to pay a significant premium to buy in the shares. In the interest of brevity, we
regarding positive stock price effects of modest amounts of repurchases, as well as have not focused on our repurchase modeling capabilities in the present paper.
40
Although liquidity management is often thought of as separate from capital
structure, we believe that it is a mistake to try to separate the two topics.
The higher the dividend, the greater is the risk of bered assets. High-tech companies with little in the
a dividend cut in the future, triggering the dividend way of tangible assets, or airlines that rely heavily on
cut penalty. secured financing, would fit this category.
The higher the dividend, the greater the chance for In these circumstances, the benefit provided by
crowding out Capex in a given period and incurring cash reserves to help avoid the penalties associated
the associated penalty. As a secondary effect, future with dividend or Capex shortfalls—or, in the ex-
Capex can also be squeezed out if the firm has treme, to help cover interest and tax obligations—
already reached its fallback rating level of debt to can outweigh the cost.
meet current dividend requirements.
To the extent that higher dividend payouts have OPTIMAL CAPITAL STRUCTURE ADJUSTED
the effect of increasing debt outstanding, credit FOR LIQUIDITY REQUIREMENTS
ratings will be pushed down, with implications for
higher costs of debt, working capital needs, and lost Up to this point, we have discussed optimal
business. capital structure assuming normal, ongoing behavior
of the simulated variables. We now consider how to
Cash Reserves adjust capital structure to provide sufficient contin-
gent liquidity in light of event risk. Although liquidity
Some industries such as autos, airlines, and management is often thought of as separate from
technology maintain sizeable cash liquidity reserves capital structure, we believe that it is a mistake to try
which can be drawn upon in situations when to separate the two topics. Providing for liquidity
operating cash flows are inadequate. In particular, insurance increases the cost of capital and affects the
DBLSM defines the “free” cash reserve as a pool structure of liabilities. In formulating their capital and
available to pay interest expense and taxes if neces- liability strategies, financial managers should attempt
sary, or to fund dividends and Capex in the event to weigh the costs of alternative approaches of
operating cash flows fall short and the firm is already liquidity insurance with the level of protection they
at its fallback rating debt level. Free cash reserves do provide, much as a driver might compare auto
not include any cash that may be “trapped” within insurance policies. DBLSM can quantify the costs of
low-tax foreign subsidiaries as a result of the tax these different solutions—for example, the relative
penalty that the company would incur to repatriate costs of various debt maturity profiles, or of different
earnings back to its home country. levels of cash (and near-cash) liquidity reserves.
DBLSM measures the cost of maintaining a cash Under normal volatility, DBLSM shows that
reserve as the negative “carry” due to the shortfall longer-term debt, even after taking account of its
between the short-term interest rate earned and the lower rollover risk, is more costly than short-term
cost of the implicit term debt financing necessary to debt. Why should companies, particularly high-
maintain the reserve, plus the reduction in the rated companies, not then finance themselves en-
interest expense tax shield resulting from the interest tirely with commercial paper? The answer is that
earned. Managers sometimes overlook the fact that virtually any company is exposed to event risk and
cash reserves produce a negative tax shield interest, must construct an alternative liquidity plan that
that is, income is subject to a corporate layer of contemplates a complete withdrawal of unsecured,
taxation that would be avoided if the shareholders uncommitted credit for a sustained period of time.
held the cash directly. Such a plan should not include the sale of core assets
Notwithstanding these negatives, DBLSM shows because significant shareholder value would likely be
that the cost of a cash liquidity reserve can be justified sacrificed in a sale under distressed conditions. We
in the following situations: use a time frame of two years to give the company
The company’s operating cash flows are highly enough time to organize its recovery. Our approach
volatile and debt credit spreads escalate rapidly at identifies likely cash inflows and outflows and then
lower ratings; finds the liability structure that provides the needed
Working capital requirements increase dramati- “breathing room” at the lowest expected cost.
cally when business conditions deteriorate; The cost of providing adequate back-up liquid-
Unsecured financing capacity is constrained, par- ity can be viewed by the company as a kind of
ticularly in hard times, by a lack of hard, unencum- insurance policy. The event risk contingency has a
41
low probability of occurring but a very high cost if Given the negative signal of actually drawing
uninsured. The cost, in fact, could be the complete upon committed bank lines, why might it still be
collapse of the company. If, however, back-up liquid- desirable to use these lines as back-up instead of the
ity is strong enough when a crisis hits, lenders see that other alternatives? The cost of issuing commercial
the company has the resources to weather the crisis. paper backstopped with bank facilities can still be
When back-up liquidity is not adequate, potential significantly lower—particularly for companies that
lenders lose confidence and withdraw their support bear an A-1/P-1 commercial paper rating or higher.
because they cannot see how the company will Moreover, if the company is downgraded to non-
survive. investment grade, drawing upon the bank lines
Like many other aspects of capital structure, the sends only a modestly negative signal, since the lines
need for back-up liquidity is magnified by market are designed to replace commercial paper when the
imperfections such as asymmetric information, which company can no longer issue it in the market. Finally,
is a major factor when an event risk situation develops. even if the company remains investment grade, the
Potential lenders may not know all the relevant factors lines give it significant negotiating power with its
and are suspicious that management is hiding the real bank group.
story in order to secure cheap financing. Audited But, having said this, it is becoming increasingly
financial statements are generally not available until clear that 364-day bank lines are not providing
well after the event, and are now viewed with adequate backstop liquidity.20 The shorter-term lines
particular suspicion in light of recent SEC investiga- have been popular in the past because banks do not
tions of financial reporting irregularities. Because have to allocate capital to them and charge only
potential lenders do not have full information, they minimal commitment fees. However, these shorter
will withhold additional support unless there is lines do not provide enough time for a company to
tangible back-up liquidity in place that would allow fix a liquidity problem and are perceived by credit
the company to survive the situation. analysts to be inadequate. Shorter lines can be
To bolster back-up liquidity, a company can: justified only to support truly short-dated funding
Extend debt maturities so that debt does not come needs (such as a retail inventory build-up in the fall
due within the target recovery period; season) or to backstop commercial paper that could
Establish asset-backed securities (ABS) programs, be replaced with ABS.21
test them, and then earmark assets that could be Just as important as securing sources of liquid-
utilized in these programs in a liquidity crunch. This ity is the need to accurately identify the possible uses
is the mainstay of the auto companies’ back-up of cash in a crunch, such as debt repayments and
liquidity. Both GMAC and FMCC made heavy use of operating cash. Additional flows would include
ABS markets in the recent downturn to pay down likely cash restructuring costs, product discounting
commercial paper when they lost their A-1/P-1 com- to liquidate inventory, and adverse shifts in working
mercial paper ratings. Using these programs does not capital as production is scaled back.
seem to sound any alarm bells in the market; In the liquidity planning process, it is also
Create a liquidity reserve composed of liquid cash important to avoid “liquidity traps” that have gotten
or near-cash investments. However, in industries that companies in trouble before. For example, it is
customarily maintain cash reserves, the act of drawing tempting to believe that meeting rating agency
them down when times are clearly troubled sends a requirements for issuance of commercial paper is
signal that the company is encountering difficulties; or sufficient for liquidity purposes and that if the
Establish committed bank facilities and then draw agencies can be talked into requiring less back-up,
upon them when required. But drawing on bank lines the company is better off. This may be shortsighted.
as an investment-grade credit sends very negative Similarly, in an event-risk situation, it is also impor-
signals to both the fixed-income and equity markets. tant to keep in mind that serial put bonds and
20. See Pamela Stumpp and Daniel Gates, “Moody’s Approach to Assessing 21. Note that 364-day lines with term loan options can provide true term
the Adequacy of ‘Liquidity Risk Insurance,’” Moody’s Investors Service —Rating liquidity.
Methodology (January 2000), for a discussion of Moody’s growing concern over the
availability of traditional bank commercial paper back-up facilities. Possible
solutions to this problem are the new “contingent capital” financing products—in
effect, sub debt and equity lines of credit—discussed by Chris Culp in this issue.
42
CASE STUDY: OUTPUT RECOMMENDATIONS FOR COMPANY X
Though the unique characteristics of each com- factors that argued strongly for having a significant portion
pany make it inappropriate to apply the results of one of the acquisition debt at ten-year or longer maturities.
study to other firms, a sample of the main recommen- Target Rating: DBLSM indicated that Company X
dations resulting from a recent DBLSM client study for should manage toward a permanent target rating of A-.
Company “X” may be helpful in illustrating the kind of With the fallback rating also set at A-, this rating
output DBLSM produces. Key inputs for the analysis combination added about $600 million of shareholder
were as follows. X is a large U.S. industrial (with a value versus keeping the target (and fallback) rating at
market cap of roughly $15 billion at the time of this the initial BBB level, where there is a substantially
study) that is a leader in its highly cyclical, commodity- higher net interest cost (that is, the present value of the
type industry.* It was modeled with a long-term mean spread over AAA rates less the tax shield benefit). It was
EBITDA/operating assets of about 16%, and an annual projected to take about five years on average before X
standard deviation for EBITDA/operating assets equal substantially attained the A- level through debt paydown
to about 1/3 of the long-term mean. (A ratio of 1/3 out of operating cash flows.
represents a high EBITDA volatility compared to that of Fallback Rating: With the target rating set at A-,
many other industries we have analyzed, which are allowing the fallback rating to drop to BBB- (three
most often in the 1/5 to 1/4 range.) This high EBITDA rating notches below the target) added roughly $200
volatility risk was only somewhat offset by an annual million of additional value, due primarily to avoiding
rate of mean reversion of about 35%, which is mid- the penalties for missed Capex or dividend cuts that
range for mean reversion rates. The correlation be- would otherwise arise due to the highly volatile nature
tween EBITDA and LIBOR was estimated at positive of the modeled EBITDA. The additional debt capacity
30%, indicating a moderate tendency for operating cash at BBB- allowed debt to expand temporarily to cover
flows to keep pace with interest rates. funding needs in periods of cash flow shortfall.
The analysis was performed to include the debt Fixed versus Floating Mix: The model indicated
associated with a then-impending acquisition that that Company X should aim for 100% floating rate debt
more than doubled previously outstanding debt, and in the long term, but should move toward this target
which put the initial credit rating in the mid-BBB range. gradually over the next several years while the rating
Initial floating rate debt was only about 10% of the total. profile strengthened to A- from its current BBB range.
Future credit ratings were projected by a model driven When compared to the initial case of 10% floating, we
mainly by two ratios: the EBIT/interest coverage and estimated that the ultimate 100% floating would add
the funds flow from operations/debt ratios. The pro- about $750 million of shareholder value. This was
jected corporate tax rate was close to 40%, but the debt primarily due to reduced expected net interest costs,
was estimated at only about 70% effective in shielding which in turn improved expected EPS and credit ratios,
income after flowing taxation through both the corpo- and actually reduced risk measures such as the prob-
rate and personal levels. The dividend payout was fixed ability of Capex shortfalls or dividend cut. The reduced
at 40%. Among other questions, we explored how to risk was the result not only of the reduced debt service
optimize the acquisition funding, target and fallback burden with more floating, but also the observed 30%
ratings, fixed versus floating mix, and maturity structure correlation between changes in LIBOR and changes in
of the debt portfolio. EBITDA/operating assets.
Principal recommendations: Maturity Structure of New Issue Term Debt: For
Acquisition Funding: Though short-maturity fund- new term debt issues in the future, alternative maturity
ing had a lower expected cost than longer-term strategies configurations produced only minor shareholder value
by about $40 million, this difference represented only differences. Therefore we recommended that Company
about 0.25% of the firm’s total market cap. The modest cost X should target its new issues across a spectrum of
of term debt was outweighed by a combination of rating maturities to meet investor demand, strengthen liquidity,
agency, term liquidity insurance, and market receptivity and avoid unnecessary refunding concentrations.
* Numbers have been broadly rounded where appropriate to disguise the

identity of the firm.
43
puttable convertible bonds—securities that are not CONCLUSION
classified as “short term” on the balance sheet—will
almost certainly be put back to the issuer for cash if We have presented some of the underpinnings
the issuer’s condition markedly deteriorates. If these of our approach to capital structure as captured in the
securities are not adequately considered in the Deutsche Bank Liability Structure Model. We do not
design of the firm’s alternative liquidity plans, the claim that our model can predict the actual capital
future uncovered puts will add to investor concerns structures of companies already in place. Rather, our
and magnify the company’s difficulties in obtaining basic philosophy is that the process of determining
new loans in an event risk situation. an optimal capital structure should involve an at-
Another common temptation is to view histori- tempt to identify and quantify all of the important
cally maintained back-up liquidity as no longer shareholder value costs and benefits of the capital
necessary if the company has never had to use them. structure decision variables associated with the level
The pressure to scale back costly back-up liquidity and structure of the company’s debt. From the
is greatest when management is having a hard time multiplicity of variables and inputs discussed above,
meeting promised earnings targets. Finally, writing it is clear that the capital structure optimization
puts on the company’s own stock can create signifi- exercise is complex and goes well beyond a simple
cant short-term cash liabilities at precisely the worst target for the ratio of debt to total capital. Finally, it
time—namely, when the stock is sinking. While is critical to adjust capital structure around the
writing puts can have significant benefits, their requirement to provide liquidity insurance to ensure
potential exercise must be considered in liquidity that the company can weather a loss of investor
planning. confidence when it hits difficult times.
ROGER HEINE AND FRED HARBUS
are Managing Director and Director, respectively, of the Liability

Strategies Group of Deutsche Bank Securities Inc.
APPENDIX: MODEL FLOW
The main steps involved in running DBLSM are as investments that could provide sources of liquidity in
follows: the event of a debt service shortfall.
1. Develop projections for the ratio of EBITDA to 3. Based on both company historical data and
the operating asset base. Specifically, fit a mean- available projections, develop assumptions for for-
reversion time-series model to the ratio of EBITDA to ward-looking Capex, dividend payout policy, book
operating assets, where operating assets equals total and tax depreciation (as well as true economic depre-
assets less free cash and goodwill, plus capitalized ciation), tax rates, tax credits and loss carryforwards,
operating leases. Estimate the ratio’s long-term mean, working capital requirements, share count, and other
rate of mean reversion, and annual volatility of the ratio key parameters necessary to project net cash flows and
using a combination of industry historical data and key accounting items, including EPS. Importantly, the
company-specific financials. The resulting model is the excess of estimated Capex over economic depreciation
basis for the baseline EBITDA projections within DBLSM. drives asset growth in the model, which in turn drives
Overlay any special cash flow projections (such as EBITDA growth after application of the EBITDA/
anticipated acquisitions or divestitures) not captured operating assets ratio.
by the EBITDA trend line. Identify starting values for 4. Estimate how EBITDA and working capital
EBITDA and operating assets. requirements vary as a function of credit rating.
2. Develop assumptions about the “financeability” 5. Identify and load into DBLSM the company’s
of hard assets, targeted cash reserves, and strategic current debt portfolio and affiliated swaps.
44
APPENDIX (continued)
6. At the firm’s actual current rating, estimate b. Set the base case levels of percent of floating rate
the cost of term debt as a function of maturity from debt and percent of debt that is short-term (one-month)
one to 30 years, expressed as a spread to the cor- vs. longer term; set the maturity of new term debt for
responding LIBOR swap rate. Then, for both ten-year any long-term debt issued in future years.
debt and one-month debt, estimate the hypothetical c. With these base parameters held fixed, vary
spread that would be expected if ratings varied the target credit rating over the range from AAA to BB,
across the spectrum from AAA to the B range. Input and for each target rating, set the fallback rating to a
the current levels of absolute short-term and long- level ranging from the target rating itself to the three
term swap rates, by currency. Update both yield or four rating notches immediately below the target
curve and FX simulation parameters including rates rating. Identify the combination of target and fallback
of mean reversion, volatilities, and correlations ratings which together maximize shareholder value
between different points on the respective yield (the “optimal rating” combination).
curves and between different currencies. d. At the optimal rating, vary the percent of
7. Based on industry credit analysis, build a floating rate debt from 0-100% in regular increments
ratings model that estimates the firm’s projected credit (typically 10%) to identify the optimal floating percent.
rating as a function of key credit ratios. Generally two Then run this optimal percent against alternative rating
ratios are used, one involving the level of debt such as combinations in the neighborhood of the previously
debt/total capital and one involving interest expense identified optimal rating to test whether the optimal
such as EBITDA/interest, though the specific choice rating has shifted. If so, re-do the floating rate sensitivity
will vary by industry. The ratings model may take the at this “new” optimal rating. This in turn may give rise
form of either a statistical regression or a look-up table to a different floating rate optimum. Iterate this proce-
of ratio thresholds at different rating levels. A particular dure until the shareholder value-maximizing configu-
rating then corresponds to a range of credit ratios ration of both ratings and floating rate percent is found.
between one threshold level and the next. e. Given the optimal rating and floating rate
8. Using the EBITDA, interest rate, and FX configuration, proceed to optimize other variables.
simulation models, generate a multitude of simulation Like the rating and floating percent, dividend policy
paths (typically on the order of 1,000), where each decisions can also have a significant impact, as an
path represents a particular trajectory of monthly increase in dividend requirements will not support as
EBITDA, yield curves, and FX rates, usually extending much debt. On the other hand, the maturity of new term
out 20 years. The paths reflect estimated correlations debt issues typically has a relatively weak impact. Also
between EBITDA and LIBOR, across interest rates at to be varied are the targeted liquidity reserve assump-
different points on the yield curve, and across differ- tions, and, most importantly, for multi-currency analy-
ent FX rates. ses, the currency mix of the debt portfolio. Again, we
9. Once the underlying set of paths is developed, would test the robustness of previously determined
this same set is used repeatedly against alternative optimal parameters, and iterate if necessary.
capital structure assumptions within DBLSM. By vary- Many variations on this sequence are possible. For
ing each key assumption over a range, we are able to example, if the client wishes to focus purely on the
identify the combination of parameters that optimizes floating/fixed question assuming the current rating is
shareholder value as well as accounting results. The “locked in,” we would skip over step 9c. Additionally,
following is a typical sequence of analysis: if desired, basic assumptions for the underlying paths
a. Input the company’s dividend policy, which for EBITDA and/or interest rates can be changed, and
includes targeting both a payout ratio range and a new optima solved for. For example, high growth vs.
targeted annual dividend growth rate subject to a low growth scenarios for EBITDA, or changes in
maximum payout; further, specify any limits on the EBITDA volatility for the same growth rate, could well
maximum percent of outstanding shares that may be lead to different conclusions regarding the optimal
repurchased annually. rating and floating percent.
45
Journal of Applied Corporate Finance (ISSN 1078-1196 [print], ISSN Journal of Applied Corporate Finance is available online through Synergy,
1745-6622 [online]) is published quarterly on behalf of Morgan Stanley by Blackwell’s online journal service which allows you to:
Blackwell Publishing, with offices at 350 Main Street, Malden, MA 02148, • Browse tables of contents and abstracts from over 290 professional,
USA, and PO Box 1354, 9600 Garsington Road, Oxford OX4 2XG, UK. Call science, social science, and medical journals
US: (800) 835-6770, UK: +44 1865 778315; fax US: (781) 388-8232, UK: • Create your own Personal Homepage from which you can access your
+44 1865 471775, or e-mail: subscrip@bos.blackwellpublishing.com. personal subscriptions, set up e-mail table of contents alerts and run
saved searches
Information For Subscribers For new orders, renewals, sample copy re- • Perform detailed searches across our database of titles and save the
quests, claims, changes of address, and all other subscription correspon- search criteria for future use
dence, please contact the Customer Service Department at your nearest • Link to and from bibliographic databases such as ISI.
Blackwell office. Sign up for free today at http://www.blackwell-synergy.com.
Subscription Rates for Volume 17 (four issues) Institutional Premium Disclaimer The Publisher, Morgan Stanley, its affiliates, and the Editor cannot
Rate* The Americas† $330, Rest of World £201; Commercial Company Pre- be held responsible for errors or any consequences arising from the use of
mium Rate, The Americas $440, Rest of World £268; Individual Rate, The information contained in this journal. The views and opinions expressed in this
Americas $95, Rest of World £70, Ð105‡; Students**, The Americas $50, journal do not necessarily represent those of the Publisher, Morgan Stanley,
Rest of World £28, Ð42. its affiliates, and Editor, neither does the publication of advertisements con-
stitute any endorsement by the Publisher, Morgan Stanley, its affiliates, and
*Includes print plus premium online access to the current and all available Editor of the products advertised. No person should purchase or sell any
backfiles. Print and online-only rates are also available (see below). security or asset in reliance on any information in this journal.
†
Customers in Canada should add 7% GST or provide evidence of entitlement Morgan Stanley is a full service financial services company active in the securi-
to exemption ties, investment management and credit services businesses. Morgan Stanley
may have and may seek to have business relationships with any person or
‡
Customers in the UK should add VAT at 5%; customers in the EU should also company named in this journal.
add VAT at 5%, or provide a VAT registration number or evidence of entitle-
ment to exemption Copyright © 2004 Morgan Stanley. All rights reserved. No part of this publi-
cation may be reproduced, stored or transmitted in whole or part in any form
** Students must present a copy of their student ID card to receive this or by any means without the prior permission in writing from the copyright
rate. holder. Authorization to photocopy items for internal or personal use or for the
internal or personal use of specific clients is granted by the copyright holder
For more information about Blackwell Publishing journals, including online ac- for libraries and other users of the Copyright Clearance Center (CCC), 222
cess information, terms and conditions, and other pricing options, please visit Rosewood Drive, Danvers, MA 01923, USA (www.copyright.com), provided
www.blackwellpublishing.com or contact our customer service department, the appropriate fee is paid directly to the CCC. This consent does not extend
tel: (800) 835-6770 or +44 1865 778315 (UK office). to other kinds of copying, such as copying for general distribution for advertis-
ing or promotional purposes, for creating new collective works or for resale.
Back Issues Back issues are available from the publisher at the current single- Institutions with a paid subscription to this journal may make photocopies for
issue rate. teaching purposes and academic course-packs free of charge provided such
copies are not resold. For all other permissions inquiries, including requests
Mailing Journal of Applied Corporate Finance is mailed Standard Rate. Mail- to republish material in another work, please contact the Journals Rights and
ing to rest of world by DHL Smart & Global Mail. Canadian mail is sent by Permissions Coordinator, Blackwell Publishing, 9600 Garsington Road, Oxford
Canadian publications mail agreement number 40573520. Postmaster OX4 2DQ. E-mail: journalsrights@oxon.blackwellpublishing.com.
Send all address changes to Journal of Applied Corporate Finance, Blackwell
Publishing Inc., Journals Subscription Department, 350 Main St., Malden, MA
02148-5020.

Toward A More Complete Model of Optimal Capital Structure

Uploaded by

Copyright:

Available Formats

Toward A More Complete Model of Optimal Capital Structure

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Toward A More Complete Model of Optimal Capital Structure

Uploaded by

Copyright:

Available Formats

Journal of Applied Corporate Finance

Toward a More Complete Model of Optimal Capital Structure

iability management plays an important but often overlooked

also has a “liquidity insurance” component that MEASUREMENT OF SHAREHOLDER VALUE

2% Example: In April 1993, the forward curve “predicted”

EBITDA volatility, or low Capex and dividends but Debt Characteristics

Year-over-year change in EPS

Source: Company data and Deutsche Bank analysis.

* Numbers have been broadly rounded where appropriate to disguise the

ROGER HEINE AND FRED HARBUS

are Managing Director and Director, respectively, of the Liability

APPENDIX: MODEL FLOW

You might also like