. 33
( 39)


street” is that many of these shares end up in the accounts of very unsophisticated in-
vestors, such as “trust accounts” for widows and orphans.

2. Underwriters™ analysts routinely issue “buy” recommendations on their IPOs. This is not
surprising. What is surprising is why this still seems to matter. Why would anyone pay
attention to these obviously con¬‚icted analysts™ opinions?

3. Insiders routinely sell their shares as soon as a pre-agreed lock-up period expires. When
the lockup expiration week comes around, the IPO stock price predictably goes down by
about 2%. Who would want to hold IPO shares through this lockup expiration?

4. IPOs either happen in droves or do not happen at all. When the overall stock market and
the ¬rm™s industry have recently performed well, IPOs are pouring in. Professionals call
this an “open IPO window.” When the opposite occurs, the window is closed and there are
zero IPOs. IPOs are not just reduced in price or scale, but they are typically withdrawn
wholesale. Why?

5. It is not surprising that the average IPO pays 7% in underwriting commission”the max-
imum allowed by the National Association of Securities Dealers (NASD)”though many
issuers ¬nd some backdoor mechanisms to raise the underwriter commissions further.
But it is surprising that virtually every IPO pays 7% commission. In such a competitive
market, why do underwriters not compete more ¬ercely on the commission front?

These are all interesting questions for future research.
¬le=capdynamics.tex: RP
Section 25·3. The Capital Issuing Process.

25·3.E. Raising Funds Through Other Claims and Means

Debt and equity are not the only claims that corporations can issue to raise funds, but they Think of hybrids as a mix
of the debt and equity.
are the broadest categories and best studied. Investment banks regularly help ¬rms to issue
all sorts of debt-equity hybrids, and for the most part, you can think of many hybrids as com-
binations somewhere along a continuum. For example, a bond may be straight, it may have
a conversion feature only at a very high ¬rm value (in which case it is almost like a straight
bond), or it may have a conversion feature at a very low ¬rm value (in which case it is almost
like equity). The aforementioned Ecaps on Page 646 is a good example.
Firms can obtain ¬nancing not only from public markets (investment banking), but also from There are alternative
money providers.
plain old commercial banks”and most large publicly traded corporations do. (Small ¬rms rely
on banks almost exclusively as their loan providers.) But insurance companies, pension funds,
mutual funds, foundations, venture capital funds, private equity funds, and even a multitude
of government support programs have also jumped into the fray, and may help provide certain
companies with needed capital.
Firms can also obtain funds by the issuing of hedging contracts (which may promise future There are ways to avoid
¬nancial markets for
delivery of a good in exchange for cash today), leasing instead of buying, securitization (in
raising money
which the ¬rm sells o¬ assets such as its accounts receivables instead of retaining its assets), altogether.
An often important method to obtain (or grant) ¬nancing is trade credit, in which the seller of Working Capital is often
a surprisingly powerful
a good allows the buyer to delay payment. The customer may even raise capital unilaterally
method to raise funds.
simply by not paying bills on time. But small and shaky ¬rms are not alone. Even large ¬rms
may earn an important competitive advantage through better working capital management.
For example, Walmart has often been accused of squeezing its suppliers (i.e., by not paying
them for a very long time). It can a¬ord to do so because its suppliers dare not risk losing
Walmart™s large market distribution. From 2000 to 2005, the very large British retailer Tesco
increased its accounts payables by £2.2 billion while its inventory stock increased by only £700
million”prompting the British O¬ce of Fair Trading to open an investigation whether this is
due to undue and unfair pressure on suppliers or merely an e¬ciency gain in working capital
management, though one does not exclude the other. Amazon actually has negative working
capital”it ¬rst receives customer payments on the web before it obtains the goods, thereby
having capital with which it can either run its business and/or earn a ¬nancial rate of return.
These are all plausible and common methods to ¬nance operations”whether they are wise or
not depends on the situation and the ¬rm.

25·3.F. The In¬‚uence of Stock Returns

We started this chapter with the observation that stock returns have an in¬‚uence on capital Capital structure can
come about “passively.”
structure, just as active issuing activity does. We could call this the “direct” e¬ect of stock
returns. A ¬rm that is ¬nanced by $1 billion in debt and $1 billion in equity and that loses one
quarter of its value ($500 million) will experience a debt-equity ratio increase from 1:1 to 2:1.
(If so desired, managers can counteract this e¬ect by issuing more equity and retiring some
But stock returns and value changes can have a second conduit by which they can in¬‚uence Speculation? How could
managers know better?
capital structure. Indeed, the mechanism is entirely di¬erent. Although it is tied directly to
past stock returns, it is really about how managers respond through issuing to market returns.
There is some evidence that CFOs believe and act as if they can predict (“time”) the ¬nancial Managers seem to
believe they know when
markets. This is not too surprising. Most managers™ sense of their ¬rm value is based on the
prices are high or low.
corporate internals, not on how the ¬nancial markets have moved recently. If the ¬nancial
markets have moved up, managers™ internal beliefs do not catch up immediately, so they now
believe that they can raise equity relatively cheaply at high market valuations. They feel that
their stock is relatively more overpriced. Note that this mechanism suggests exactly the oppo-
site behavior to what would be required for the ¬rm to return to its original debt-equity ratio. If
the ¬rm wanted to keep a particular debt-equity ratio, it would have to repurchase equity that
¬le=capdynamics.tex: LP
656 Chapter 25. The Dynamics of Capital Structure and Firm Size.

has gone up and issue more equity that has gone down. If the ¬rm wanted to time the market,
it would repurchase equity that has gone down and issue equity that has gone up. Managers
similarly seem to try to time interest rates and the yield curve. If interest rates are higher
(lower) than they were in the past, companies tend to avoid (seek out) bonds. If the yield curve
is steep by historical standards, corporations tend to borrow more at short-term interest rates
than issue long-term bonds. In an e¬cient ¬nancial market, there is no bene¬t to attempts at
market timing, but also no cost to doing so. You can look at this attempt at market timing as
just another investment, which is a fairly harmless attempt by managers to make pro¬table
However, what is surprising is not the fact that managers have tried to time ¬nancial markets,
Weird”it seems to have
worked! but the empirical evidence that this has actually turned out to be pro¬table! Even stranger,
managers have been good not only in predicting their own stock price level, but in predicting
the overall stock market level”an incredibly di¬cult feat. (In fact, why be a corporate manager
if you have this ability?) There is academic controversy as to whether this success has been the
result of coincidence or real timing ability. For example, one counter-argument is that seeming
timing ability is merely survivorship bias: ¬rms that failed in their timing disproportionally
disappeared. It could also just be that when the ¬nancial markets go up, more and more
¬rms raise external funds, and this stops when ¬nancial markets go down. Thus, even though
managers cannot predict the ¬nancial markets, when economists look at when ¬rms raised
funds, they will ¬nd that they did so before the market went down. Either of these two theories
could explain seeming market timing ability when there is none.
Solve Now!
Q 25.8 What is the ¬nancing pecking order?

Q 25.9 What is the ¬nancing pyramid? Is it a good description of empirical reality?

Q 25.10 Does the pecking order imply a ¬nancing pyramid?

Q 25.11 How does a coercive bond exchange o¬er work?

Q 25.12 How does a coercive seasoned equity rights o¬ering work?

Q 25.13 Assume a rights o¬ering for a ¬rm that is worth $500 million and o¬ers its shareholders
to buy one extra share for each share that they already own. The “discount” price for the new
shares is 1/5 the price of the current shares. Assume that half the investors do not participate.
What is the loss to non-participating investors (shares) and the gain to participating investors

Q 25.14 What are the some of the main empirical regularities about IPOs?

Q 25.15 Why are IPOs underpriced?
¬le=capdynamics.tex: RP
Section 25·4. Summary.

25·4. Summary

This chapter covered the following major points:

1. Both capital scale and capital structure dynamics are in¬‚uenced by factors under man-
agement™s immediate control (such as debt issuing or share repurchasing) and factors
beyond management™s immediate control (such as value changes, aka stock returns).

2. A CFO should consider a holistic view of capital policy. Many activities in¬‚uence both the
¬rm scale and debt-equity ratio.

3. Appropriate cash management should be a primary concern in many ¬rms.

4. Many ¬rms follow a “pecking order” ¬nancing scheme, in which they ¬rst use retained
earnings, then progressively less senior debt, and ¬nally equity (as a last resort).

5. Debt o¬erings come in many varieties, and though we surgically dissect their features,
the actual debt o¬erings are often complex packages.

6. Seasoned equity o¬erings are rare. They can be standard, shelf-registered, or rights o¬er-
ings. (Secondary shares are more insider sales than corporate capital structure events.)

7. Initial public o¬erings appear within certain industries and at certain times, i.e., in waves.
The average one-day IPO underpricing is about 10%, but IPOs begin to underperform the
market beginning about 6 months after the o¬ering for about three to ¬ve years.

8. Debt and equity are not the only venues to raise ¬nancing. There are other methods, e.g.,
stretching out the payment of bills.

9. There is empirical evidence that many managers try to time the ¬nancial markets. Remark-
ably, this has often turned out to be pro¬table, although we do not yet fully understand
why this is so.
¬le=capdynamics.tex: LP
658 Chapter 25. The Dynamics of Capital Structure and Firm Size.

Solutions and Exercises

1. Yes and no. Firms that operate need to incur debt, so in this sense the answer is no. However, ¬rms could
change their operations or re¬nance their debt by raising equity.
2. See Table 25.1.

3. To have a 5:1 debt-equity ratio with $600 million in overall value, the ¬rm needs to have $500 million in debt
and $100 million in equity. One way to accomplish this is to issue $250 million in debt and repurchase $100
million in equity.
4. Before the market reacts, the ¬rm will have $400 million in debt, and $200 million in equity. The market
believes these transactions will create $15 million in equity. If all accrued to shareholders, there would be
$215 million in equity for $615 million in value, which would be a 65% debt-asset ratio. If none accrued to
shareholders, the equivalent debt-asset ratio would be 67.5%. The debt equity ratio would be 1.86.
5. Can you invest money better than your shareholders can on their own? Do your shareholders understand
and agree that you are acting in their interest?
6. The ¬rm can match assets and liabilities, obtain a credit line, invest in more liquid assets, and avoid debt.
Doing so is costly”in a public company, too much cash also tempts managers to waste assets and not work
as hard.
7. No. The market pressures forcing poorly ¬nanced companies back to optimal behavior are weak.

8. Managers prefer issuing safer securities ¬rst, before proceeding to less safe alternative.
9. Companies are ¬nanced predominantly by safer securities. Equity is a small part of the pyramid at the top.
The traditional view of the ¬nancing pyramid does not apply to many successful companies, because the
equity will have grown in size.
10. No! Equity can change in value (and debt can accumulate during operations). Many ¬rms follow a ¬nancing
pecking order, but their capital structures do not look like a pyramid.
11. It gives existing bondholders the right to exchange their bond for a more senior bond with lower face value.
Creditors who do not participate are e¬ectively expropriated.
12. It gives existing shareholders the right to purchase more shares at a given price. Investors who do not
participate are e¬ectively expropriated.
13. Assume that the shares are $10 each. You can then purchase shares for $2 each. Of 50 million shares, 25
million will participate. So, you will raise an extra $50 million. Thus, total corporate assets will be $550
million. There are now 75 million shares in total. Therefore, each share will be worth $7.33. Participating
investors will own two shares worth $14.67, for which they will have paid $12. This represents a 22% gain.
Non-participating investors will own one share each, for which they will have paid $10. This represents a
26.7% loss.
14. On average, they appreciate by 10% from the o¬er price to the ¬rst after-market price, and then lose about
5-10% per annum over the three years. (Other regularities are described above.)
15. There are a number of explanations”such as the winner™s curse, payment to investors for revealing infor-
mation, the intent to leave goodwill for future o¬erings, highly elastic cascade-related after-market demand,
and agency con¬‚icts between the ¬rm and the underwriter.

(All answers should be treated as suspect. They have only been sketched, and not been checked.)
¬le=capdynamics.tex: RP
Section A. Appendix: Standard&Poor™s 04/24/2005 Bond Report on IBM™s 2032 5.875% Coupon Bond.


A. Appendix: Standard&Poor™s 04/24/2005 Bond Report on
IBM™s 2032 5.875% Coupon Bond

This needs to be cleared for copyright.
¬le=capdynamics.tex: LP
660 Chapter 25. The Dynamics of Capital Structure and Firm Size.
How Have Firms™ Capital Structures Evolved?

Nothing yet!
last ¬le change: Mar 2, 2006 (10:57h)

last major edit: Apr 2005

The goal of this chapter is to describe, in broad strokes, how publicly traded corporations in
the United States have ¬nanced themselves over the last few decades. (There are very little data
and research on how private ¬rms are ¬nanced.)
You should realize that this chapter is operating at the cutting edge of research. There are
di¬erent interpretations of the data, so it is unavoidable that what you are reading is an inter-
pretation of the evidence. My goal is to give you a taste of what we know”and what we do not

¬le=capempirical.tex: LP
662 Chapter 26. How Have Firms™ Capital Structures Evolved?.

26·1. Mechanisms vs. Rationales

We are now interested in exploring empirically how the debt-equity ratios of publicly traded
What determines the
ways by which capital companies have evolved. To do so, we can examine several layers of causality”one can always
structure can change?
drill deeper and deeper. (Eventually, if you dig deep enough, you will ¬nd yourself in the world
Layers of Causality.
of philosophy and theology.) Think about an analogy”say, you want to know why a car is
fast. The ¬rst layer of causality may be that its speed is due to lots of power, low weight, and
low wind resistance. But why is there a lot of power? This question brings you to a deeper
layer of causality with questions such as how many cylinder and intake valves your engines has.
You can then drill down into yet another layer of causality. Why is this particular number of
cylinders/valves more powerful? Yet another deeper layer of causality emerges with questions
such as why and how gasoline combusts.
We are now going to explore the dynamics of debt-equity ratio changes on two levels. We can
Two layers of causality:
mechanisms and forces. call our ¬rst, somewhat shallow layer the “mechanistic layer”: how important are the various
mechanisms through which debt-equity ratios can evolve? These mechanisms are basically the
cells you have already seen in Table 25.1, such as debt and equity issuing and repurchasing. The
second, deeper layer is more causal and explores the variables, characteristics, and economic
forces that induce ¬rms to engage these mechanisms in the ¬rst place. There is one factor,
which could be classi¬ed either in the ¬rst or second level”the role of stock value changes: you
can think of it either as a mechanism that shifts capital structure around, or as an economic
force, partly within and partly outside the domain of the mechanisms that managers can use.

26·2. The Relative Importance of Capital Structure Mecha-

Let us begin with the big picture mechanisms. We have just looked in the last chapter at the
Stock returns and
long-term debt issuing various mechanisms that in¬‚uence capital structure. Now we ask about the relative importance
are the most important
of each of these mechanisms. That is, has the typical company™s debt-equity ratio been driven
factors changing
more by the ¬rm™s value or by the CFO™s net issuing activities (which include issuing, repur-
debt-equity ratios.
chasing and dividends)? This question can be phrased as “If you knew in advance how much
every ¬rm would issue over the next x years, what fraction of the change in capital structure
could you explain?” Table 26.1 answers this question for ¬ve-year horizons. The rest of this
section discusses the meaning of this table.

Table 26.1. Relative Importance of Factors Determining Capital Structure Changes Over Five

All Net Issuing (and Dividend Activity) 69%
All Net Issuing (without Dividend Activity) 66%
· · · All Net Debt Issuing Activity 40%
· · · · · · Convertible Debt only 4%
· · · · · · Short Term Debt only 14%
· · · · · · Long Term Debt only 32%
· · · All Net Equity Issuing Activity 16%

Direct E¬ect of Stock Returns on Existing Capital Structure 40%

Note: These values measure how much of the change in capital structure from today to ¬ve years from now you
could explain if you had perfect foreknowledge of each component. Net issuing means issues net of retirements.
The samples were all publicly traded U.S. stocks from 1964 to 2003. (The numbers need not add up to 100%, because
one component can have information about the other components.) The equity is measured by its market value.
Source: Welch, 2004.
¬le=capempirical.tex: RP
Section 26·2. The Relative Importance of Capital Structure Mechanisms.

26·2.A. Net Issuing Activity

The ¬rst row of Table 26.1 shows that CFOs were by no means inactive in the capital markets.
If you had perfectly known how ¬rms had issued and retired debt and equity and paid in and
paid out funds, you could have explained 69% of ¬rms™ total capital structure changes over a
¬ve-year horizon. The remaining 31% are necessarily corporate value changes that have not
been directly in¬‚uenced by managerial issuing and repurchasing. Omitting dividends drops
the explanatory power from 69% to 66%, so dividends can explain only a meager 3% of capital
structure”as far as debt-equity ratio dynamics in publicly traded corporations are concerned,
dividends are a sideshow.

Net Debt Issuing
Row 2 in Table 26.1 tells us that 40% of all capital structure changes over ¬ve years were due to Net Debt Issuing.
¬rms™ net debt issuing activity. The next three rows can tell us that long-term debt alone can
account for 32% of changes in debt-equity ratios, that short-term debt has been somewhat less
important, and that convertible debt has been fairly unimportant. It would be interesting to
break these debt issuing activities into their components”issuing and repurchasing”and to
break the repurchasing in turn into sinking fund payments, interest payments, and principal
repayments, so that we could understand better what part of the mechanism really drives
capital structure. Remarkably, despite the obvious importance of debt issuing activity, we
really do not know this decomposition.

Net and Pure Equity Issuing
The next row in Table 26.1 shows that net equity issuing can explain about 16% of changes Net Equity Issuing.
in ¬rms™ debt-equity ratio, and therefore is less important than net debt issuing as a determi-
nant of capital structure. Nevertheless, equity issues are more glamorous, so economists have
studied them in more detail.

Table 26.2. Typical Equity Share Activity Among S&P100 Stocks, 1999-2001

Total Seasoned Equity O¬ering Activity + 3.77%
· · · M&A Related, +3.68%
· · · Not M&A Related, +0.09%
Executive Compensation + 1.05%
Convertible Debt + 0.14%
Warrant Exercise + 0.05%
Share Repurchases “ 1.44%
= Changes in Equity Outstanding = +3.57%

Note: Categories describe equity issued in conjunction with an activity. Equity share activity is measured per annum
and as a fraction of total assets. For scale, changes in total liabilities were about 10.07% of assets, and changes in
retained earnings were 1.37% of assets.
Source: Fama and French, 2004.

Table 26.2 decomposes equity issuing into its components, though only for the very largest Net Equity Issuing.
publicly traded ¬rms. (Unfortunately, we do not have knowledge of a similar decomposition
for smaller ¬rms.) The table dispels the popular myth that most shares occur through plain
seasoned equity o¬erings. Instead, from 1999 to 2001, equity shares appeared most commonly
through equity o¬erings in connection with corporate acquisitions. (We also know that ¬rms
commonly issue not only equity but also debt to ¬nance acquisitions, so we cannot conclude
that ¬rms™ debt-equity ratios declines during acquisitions.) Outside an acquisition, seasoned
equity o¬erings are exceedingly rare. We also saw these patterns in IBM™s case in Section 24·1”
IBM did not issue equity, repurchased some shares into its treasury, and then used equity shares
¬le=capempirical.tex: LP
664 Chapter 26. How Have Firms™ Capital Structures Evolved?.

from its treasury in its acquisition of PwCC partners and in its funding of employee stock option
Moreover, other evidence similarly suggests that, even including M&A activity, public equity
More evidence from
elsewhere suggests SEOs o¬erings are rare. The 10,000 or so ¬rms trading on the NYSE and NASDAQ conducted only
are rare in smaller ¬rms,
about 12,000 equity o¬erings from 1990 to 2000, of which about half were initial public o¬er-
ings and about half were seasoned equity o¬erings. With only 300 SEOs in an average year, the
typical public ¬rm would issue more equity only about every 20 years.

26·2.B. Firm Value Changes

The ¬nal row in Table 26.1 shows the direct e¬ect of stock returns on capital structure. Re-
Value changes are
proxied by stock returns. call that this is the debt-equity ratio change that a company experiences when it increases or
decreases in value” a $200 million ¬rm with $100 million in debt and $100 million in equity,
which doubles in value from $200 million to $400 million, will drop its 1:1 debt-equity ratio to
a 1:3 debt-equity ratio. As mentioned earlier, corporate stock returns can be viewed both as
a mechanism (itself in¬‚uenced by deeper forces) and as an external force that tugs on ¬rms™
debt-equity ratio.
Table 26.1 shows that if you had known perfectly how stock returns would turn out over the
Value changes can
account for a little less next ¬ve years, you could have explained 40% of ¬rms™ total capital structure changes. (Note
than half of capital
how all issuing was able to explain 69%, so a good part of variation must have been explainable
structure changes.
by either.) The fact that stock returns are a major factor should not come as a big surprise
to you. If your recall our IBM example from Section 24·1, it was changes in the stock price
that ¬rst reduced IBM™s equity value by one-third from 2001 to 2002, and primarily caused its
debt-equity ratio to increase from 0.31 to 0.55.
Importantly, you can think of these stock returns as the “relevant” changes that were not
Apparently, managers
did not fully rebalance. undone by managers. If ¬rms had undone the value change and rebalanced through issuing and
repurchasing, then knowing the stock returns would not have helped in explaining changes in
capital structure. Our empirical evidence therefore suggests that even over a ¬ve-year horizon,
¬rms do not fully rebalance their capital structure.
You may wonder whether some part of this 40% could also pick up if managers typically tried to
Trust me: market timing
is only secondary. time the market, and issued more equity as the stock price goes up. Other empirical evidence
suggests that market timing is not a strong force. The reason is that, in response to stock price
increases, ¬rms issue not only equity, but also debt, and tend to pay out more in dividends.
Therefore, the timing e¬ect on net debt-equity ratios is fairly modest. The 40% that we see is
almost entirely the direct value e¬ect of stock returns on debt-equity ratios.
Explaining 40% of something that is as variable and ¬rm-speci¬c as corporate debt-equity ratio
In perspective: observed
capital structure today changes are is quite robust”even though our explanatory variable is conceptually on a fairly
is strongly related to
shallow level of causality. Consequently, if you want to know why some ¬rms have high debt-
past corporate
equity ratios today and why other ¬rms have low ones, a big part of your explanation has to be
not that the former issued a lot of debt and the latter issued a lot of equity, but that the former
had experienced negative stock returns and the latter had experienced positive stock returns.
Managers also typically do not pay out large value gains, or raise more funds in response to
Scale is also not
deliberate. large value losses. Therefore, like debt-equity ratios, ¬rm scale has a large external component,
too”¬rms that are large today may not be large primarily because they raised a lot of funds,
but also because they appreciated in value. In sum, few ¬rms seem to deliberately choose their
target scale and target debt-equity ratio, and then act to retain these targets.
This relationship between stock returns and capital structure would suggest a natural debt-
A lifecycle? Zero debt for
large ¬rms? Maybe not. equity lifecycle for ¬rms. Most ¬rms should start out being highly levered”the owner must bor-
row to ¬nance the ¬rm. Eventually, as the ¬rm survives and accumulates equity, its scale should
increase and its debt-equity ratio should decline. In Table 24.6, many of the non-¬nancial giant
companies indeed seem to have very low debt ratios, often in the single digits. However, we also
know from the previous chapter that large ¬rms continue to have debt-equity ratios around
40%, nothing close to zero. Thus, over the very long run”longer than ¬ve years”some other
factors (e.g., M&A activity and ¬nancial distress) must have caused today™s ¬rms to retain high
¬le=capempirical.tex: RP
Section 26·2. The Relative Importance of Capital Structure Mechanisms.

leverage ratios. The relative importance of the mechanisms that have brought us the diversity
of ¬rms™ capital structures today are still not fully understood.
What can CFOs learn from the fact that stock returns are one major determinant of most ¬rms™ What can a CFO learn
that is of immediate
capital structures? Let™s ask a few pointed questions:

• Is the empirically observed failure to rebalance size and debt-equity ratio evidence of poor
managerial behavior? Absolutely not. It might well be that the optimal ¬rm size increases
and the optimal debt-equity ratio decreases as the ¬rm™s underlying business becomes
more valuable. In this case, managers should be happy with their capital structures. Or it
might be that such rearrangements are fairly expensive, relative to the costs. In this case,
managers may be unhappy with their capital structures, but it would not be pro¬table to
do something about it.

• Could the failure to rebalance be evidence of poor managerial behavior? Yes, it could
be”but it does not need to be! We just don™t know yet one way or the other. In some
¬rms, the evidence that managers are mis-capitalized is fairly suggestive. In other ¬rms,
we are not so sure. There is lively academic controversy surrounding this question.

• Does this failure to rebalance mean that if you run a corporation, you should not worry
about capital structure or appropriate corporate scale? Absolutely not. Even if many other
managers are passive and/or do not do the right thing, you still can! Your managerial
choices should remain intelligent and dynamic.

• Does this mean that we cannot use the capital structures of other companies to judge
what capital structure our own ¬rm has? Probably yes. Their capital structures are less
indicative of deliberate design than they are of their historical performance.

Digging Deeper: Stock returns are a good proxy for the value changes we discussed in Section 25·3.F. Theo-
retically, however, stock returns could miss some of the change in the underlying asset values, if these changes
bene¬ted or hurt debt holders by making debt repayment more or less likely. However, unless the ¬rm is in”or
close to”¬nancial distress, almost all of a ¬rm™s own value change goes to equity owners. In the extreme, risk-free
debt would not be a¬ected at all by ¬rm value changes, and stock returns would be exactly equivalent to the value
change. In any case, we do not mean that debt value changes cannot occur, just that they tend to be so much
smaller that our proxy of stock returns will capture most of how ¬rms di¬er from one another in terms of value
changes at any given point in time. Aside, we do not have good market value data for corporate debt, so we
could not really measure the whole change in value even if we wanted to.

Solve Now!
Q 26.1 What are the most important ¬nancial mechanisms in¬‚uencing capital structure changes
over ¬ve-year horizons?

Q 26.2 How important is non M&A related seasoned equity issuing activity among Fortune-100

Q 26.3 If ¬rms often do not readjust their capital structure, does this mean that capital structure
theories are irrelevant?
¬le=capempirical.tex: LP
666 Chapter 26. How Have Firms™ Capital Structures Evolved?.

26·3. Deeper Causality ” Capital Structure In¬‚uences

You now know how ¬rms change their capital structure”which mechanisms they are using. You
What determines are the
forces operating on the can also think of these mechanisms as “channels” through which other forces can operate”
forces that are one layer deeper in terms of causality. There are at least ¬ve possible mecha-
nisms (channels), so we can ask:

1. What makes ¬rms issue debt?
2. What makes ¬rms retire debt?
3. What makes ¬rms issue equity?
4. What makes ¬rms retire equity (or pay dividends)?
5. What makes ¬rms experience good/bad corporate value performance? (As noted earlier,
you might classify value changes as deeper than a managerial mechanism, though.)

(We are ignoring such channels as pension and operating liabilities.) These questions are get-
ting at the deeper issue of why capital structure is what it is. We know how important the
mechanisms changing debt-equity ratios are, but we do not yet know why ¬rms use them. We
also know that if we had a choice, we would determine ¬rst what drives debt net issuing (espe-
cially long-term debt), then what drives net equity issuing and net short-term debt issuing, and
only ¬nally what drives convertible debt issuing and dividends”in that order.
If a variable strongly in¬‚uences one channel, this in¬‚uence will likely, but not necessarily, per-
The ¬ve channels to
work with. colate into an in¬‚uence on the overall capital structure. For example, if solar ¬‚ares were to
make ¬rms issue debt, then we would also expect solar ¬‚ares to increase ¬rms™ debt-equity
ratios. However, this is not a necessary outcome. If solar ¬‚ares had a strong positive in¬‚uence
on debt-equity ratios through one channel and a strong negative in¬‚uence through another, so-
lar ¬‚ares could end up having no in¬‚uence on overall capital structure. Moveover, you learned
earlier that it is possible for a variable to explain a lot of equity issuing and yet have no in-
¬‚uence on typical debt-equity ratios”if the ¬rms that are subject to this variable are already
100% equity-¬nanced, the ¬rm will still remain all equity. The opposite can also be the case.
Some variable could have only a weak in¬‚uence through every single channel and we would be
tempted to discard it as too weak, but if it worked for all ¬ve channels, it could end up having
a strong in¬‚uence on the ¬rms™ overall debt-equity ratios.

26·3.A. A Large-Scale Empirical Study

A recent large-scale empirical study by Hovakimian, Opler, and Titman (2001) explores how
The Land of Oz.
di¬erent variables exert in¬‚uences on the ¬rst four channels over one-year horizons. The
study shows that our knowledge of the empirical determinants of capital structure is weak.
The authors document that there are a multitude of variables that seem to play statistically
signi¬cant roles”but all of these variables together can explain only a few percentage points of
the total variation in capital structures across ¬rms. For the most part, there is no smoking gun.
The determinants of capital structure remain rather murky. We have only a glimpse of what is
going on.
Here is what the study found:
Big Findings.

The Debt Issuing Channel Firms issue more long-term debt if they have high market/book
Asset Performance and
Industry Benchmark. ratios, if they had good recent stock market performance, and if they have much of their
existing debt coming due soon. Firms issue more short-term debt if they have poor recent
asset performance and if they have less short-term debt than their industry peers. In both
cases, though, the relationship is very weak”these causes could explain only 2 to 3% of
its cross-sectional variation (called R 2 )”a minuscule proportion. In sum, we just do not
know yet what makes ¬rms issue debt.
¬le=capempirical.tex: RP
Section 26·3. Deeper Causality ” Capital Structure In¬‚uences.

The Debt Retirement Channel For the second channel (debt retirement), the authors found Industry Benchmark,
Recent Performance.
that ¬rms reduce their debt if they are above their industry peers in terms of their debt
ratios, and if they have had good recent stock market but bad accounting performance. In-
terestingly, managers™ debt retirement actions are thus the opposite of what it would have
taken to rebalance to the previous debt-equity ratio. We must again ask how important
these causes are, and here we get a much better 12% in explanatory power (R 2 ).

The Equity Net Issuing Channel The third and fourth channels are where most of the aca- Recent Performance.
demic research has focused”especially insofar as seasoned equity o¬erings and divi-
dends are concerned. There are three good reasons for this: ¬rst, we have robust theories
here, speci¬cally the pecking order theory, which seems to be reasonably consistent with
some of the evidence; second, the announcement of market-related equity issuing and
dividend activity plays a prominent role in the ¬nancial press; and third, we have a lot
of publicly available data here. Nevertheless, dozens of studies have informed us that
equity issuing and retiring activity also remains a mystery.
The evidence seems to suggest that ¬rms ¬rst and foremost do not like to issue equity,
consistent with the pecking order theory and any other theory that imputes high costs
to equity issuing activity. When ¬rms do announce that they will issue equity, it is on
average greeted with a negative return on its outstanding stock”the subject of our next
chapter. On balance, ¬rms tend to issue equity (rather than debt) if they have had worse
accounting performance and better stock market performance. (Although ¬rms also tend
to issue debt in response to positive stock returns, their tendency to issue equity seems
stronger”evidence that managers try to “time” the stock market.) Especially ¬rms with
more tax obligations tend to prefer issuing debt over equity.
Altogether, the authors could explain 3% of the variation in ¬rm™s equity repurchasing
activity, and 15% of ¬rm™s equity issuing activity.

Putting this (and other) evidence together, here is my overall impression of what factors play The most important
factors explaining capital
important roles in in¬‚uencing capital structure outcomes, in rough order of their importance:
structure choice, IMHO.

Direct Stock Performance In¬‚uence If you classify stock returns as a cause rather than a mech-
anism, then it is by far the most important variable in non-¬nancial ¬rms. Because ¬rms
do not counteract stock returns, ¬rms with good stock price performance tend to end up
with lower debt ratios, while ¬rms with poor stock price performance tend to end up with
higher debt ratios. (You may want to dig deeper and ask what causes stock performance,
but this would again be a di¬cult predictive exercise.)

Equity Issuance Avoidance Firms seem to want to avoid issuing equity. A seasoned equity
o¬ering is a rarity, and even more so outside an M&A transaction. Given that the costs
of an equity issue are high (including the often negative market reaction), this is not
surprising behavior.

Peer Similarity Firms seem not only to end up with capital structures similar to those of their
industry peers due to their commonality in industry stock returns, but also actively seem
to like being similar. They often issue or retire debt or equity to come closer to their
peers. Some industries (R&D heavy with few tangible assets) have avoided debt ¬nancing
altogether. (You may want to ask what determines peers™ ratios, and why ¬rms want to
be similar to their peers, but this is an even deeper level of causality”mostly beyond our
current knowledge.)

Corporate Income Taxes Firms with high corporate income tax rates tend to actively issue
debt and retire equity, i.e., increase their debt ratios.
Nevertheless, high corporate tax ¬rms usually have low debt ratios. How can this be?
The reason is that good performance translates not only into high pro¬ts and therefore
high corporate taxes, but also into positive stock price performance. The latter directly
reduces the ¬rm™s debt ratio. Although the end e¬ect can be complex, on average, net
issuing activity is usually not enough to undo the direct stock return e¬ect.
¬le=capempirical.tex: LP
668 Chapter 26. How Have Firms™ Capital Structures Evolved?.

Accounting Performance Firms prefer net debt issuing over net equity issuing if they have
better accounting pro¬tability and more tangible assets (which can be easily collateral-
ized). But as with taxes, good accounting pro¬tability correlates strongly with higher
stock prices, which in turn correlates strongly with lower debt ratios.

M&A Activity Much debt and much equity are issued in connection with M&A activity, although
proportionally more debt is issued than equity. M&A Activity may be the most important
reason why most well-performing non-¬nancial ¬rms do not end up with practically no
debt. However, because ¬rms usually start acquiring ¬rms after good stock price perfor-
mance, the overall capital structure e¬ect can be complex. Good operating performance
can lower the debt ratio through the value increase, but then increase the debt ratio
through acquisitions.

Financial Distress Firms that are in dire straits (not the MTV band) have no choice but to retire
some debt and issue equity. This seems to be an unusually solid net issuing in¬‚uence,
but only for ¬rms close to the verge of bankruptcy.

Active Market Timing Firms that experience stock price increases tend to issue more securities”
through both debt and equity, so the capital structure consequence is not too strong.
Moreover, such ¬rms also tend to pay out more in dividends, so even the net equity issu-
ing e¬ect is not yet clear. Nevertheless, when surveyed, CFOs claim that they do watch
their stock market value, and respond to it”perhaps even try to time it. In any case, ac-
tive market timing is the newest and thus the most interesting factor to explore”as more
research comes forth, we may see its importance rise.

Uncertainty Firms with more volatile underlying assets tend to have less debt in their capital

26·3.B. Theory vs. Empirics

The above variables are interesting, but they are not exactly what the theories were asking
What we can learn from
what we have. for. For example, an interest coverage ratio is often used as a proxy to measure the proximity
to ¬nancial distress”but it is not exactly ¬nancial distress. Some ¬rms have low interest
payments relative to earnings, and are in distress; other ¬rms have high interest payments
relative to earnings, and are ¬nancially sound. Yet ultimately, we study such speci¬c variables
only because they are relatively easy to measure empirically. We would have preferred direct
measures of our theories of capital structure, but such measures are usually not as easily
available. Most of the time, our variables are a compromise between empirical availability and
theoretical construct, and we then try to interpret our empirical ¬ndings through the lenses
of our theories. From our proxies, we can draw two basic conclusions about the theories:
First, it appears that agency concerns, pecking order concerns, ¬nancial distress (in very few
companies) and corporate taxes all matter, at least a little”or they matter in di¬erent ways
through di¬erent channels. Second, there are some other proxies that matter, but we just do
not know why they matter. For example, we do not know why ¬rms do not counteract market
in¬‚uences more strongly, and why they seem to “like” capital structures similar to those of
their industry peers.
In sum, we have learned that we do not yet fully understand the factors that are driving ¬rms
Why so weak?
to actively change their capital structures. It seems to be a complex process, possibly with a
lot of idiosyncratic behavior. Our variables are statistically signi¬cant, but they leave a lot to
be explained. You can read the situation in a number of ways:

1. Our variables may not matter much, because they are poor proxies for our theoretical con-
structs (e.g., for tax savings or bankruptcy costs). With more research, we may eventually
¬nd better proxies that will help us understand capital structure empirically better.

2. There are other theories and factors that we do not yet know which may be more important
than those we have now.
¬le=capempirical.tex: RP
Section 26·3. Deeper Causality ” Capital Structure In¬‚uences.

3. Our variables may not matter much, because capital structure choice is practically irrele-
vant. Whatever managers may be acting on”whether based on, say, book market ratios
or their horoscopes”may have only minimal value consequences. You could think of this
as an empirical validation of Modigliani-Miller.

4. Managers may just act poorly and erratically, and there is nothing outsiders can do to
correct it.

It is probably a little of each. Right now, capital structure is an especially fertile area for behav-
ioral ¬nance, because idiosyncratic managerial behavior seems important and because there is
no easy way for ¬nancial markets to arbitrage misbehavior. So, empirical capital structure re-
mains an exciting ¬eld of research. We are de¬nitely making progress in learning how managers
behave, but we also have a long way to go.

26·3.C. Evidence on Equity Payouts: Dividends and Equity Repurchasing

Let us now turn towards dividends and share repurchases, both of which increase the debt- Dividends and
repurchases are the
equity ratio. Although they are not too important in terms of changing ¬rms™ debt-equity
payout mechanisms.
ratios, they are the primary mechanisms by which equity shareholders receive a payback on Typical dividend yields
their investment. As of 2000, the typical ¬rm in the S&P500 paid out about 1% to 2% of its and changes.
market value and half of its earnings in dividends each year. In the mid-1990s, about one in
ten quarterly earnings announcements saw a dividend increase, one in one-hundred quarterly
earnings announcements saw a dividend decrease, and all others kept the dividends where they
were previously.
As we discussed in Chapter 23, share repurchases were clearly the tax-smart method to pay Repurchases and
dividends are now
out cash until 2003, when the Bush administration reduced the dividend tax from 35% to 15%,
approximately equally
and largely eliminated double taxation. The empirical evidence suggests that the historical important.
dividend taxation disadvantage had induced many ¬rms to shift away from dividends and
towards share repurchases as their primary means to return money to shareholders. This was
clearly reversed. The 2003 tax reform has indeed changed corporate payout policy. Chetty and
Saez (2004) found that the elimination of the double taxation has induced nearly 150 ¬rms to
initiate dividend payments, adding about $1.5 billion in aggregate, regular dividends. In sum,
dividend payments are no longer a mystery.
So, with the tax disadvantage gone or at least drastically reduced, what are the remaining Remaining Differences:
changes in inside
important institutional di¬erences between dividends and share repurchases? Not many.
ownership, an informal
understanding about
1. Executives and insiders are often not permitted to tender their shares into share repur- persistence of future
payments, and
chase o¬ers, and thus will own relatively more of the company after a repurchase than “behavioral ¬nance” type
after an equivalent dividend payment. In addition, many executive stock option plans are investor preferences.
written on stock prices that are not adjusted for dividends. This means that executives
prefer a share repurchase to a dividend payment. For example, if a manager of a $60 com-
pany has an option that allows her to purchase shares at $50, then the manager would
be reluctant to pay a $20 dividends”after all, the share price would drop to about $40,
making the right to purchase at $50 a lot less valuable.

Anecdote: The Biggest Dividend Payout Ever?
In the most prominent immediate response to the Bush dividend tax cuts of 2003, Microsoft (MSFT) initiated
dividends. Fifteen months later, on July 20, 2004”and 7 minutes after the market had closed”it announced a
$32 billion special dividend, plus a $30 billion share repurchase, plus an increase in ordinary dividends from 16
cents to 32 cents per share (a yield increase from 0.56% to 1.12%). With a market capitalization of about $300
billion (a P/E ratio of about 20 [forward-looking earnings] and 37 [current earnings], and a cash hoard of $56
billion), the total payout represented about 20% of Microsoft™s market value. A few minutes after market opening
on July 22, Microsoft™s outstanding shares had jumped in value by a little over 3%, so for every dollar changing
hands from investors™ company to investors™ pocket, shareholders also felt 15 cents happier! Interestingly, two
days later, Microsoft announced quarterly earnings 82 percent higher than those from its prior year, but still
short of expectations”and shares promptly fell back to where they had been before the payout announcement.
It appears as if the payout announcement was a positive signal, and the failure to meet earnings expectations”
despite huge earnings growth”just about canceled one another. We researchers are all watching for what will
happen in 2008, when the dividend tax cuts are scheduled to expire.
¬le=capempirical.tex: LP
670 Chapter 26. How Have Firms™ Capital Structures Evolved?.

2. Dividends tend to be more regular”because shareholders expect them to continue”while
share repurchases can be done in occasional chunks. An ordinary dividend payment there-
fore informally obliges management to continue, signaling more optimism. Of course,
management can also pay a “special dividend,” which signals its one-time nature to in-

3. There is some evidence that retail investors simply “like” dividends better than share re-
purchases, and institutions like just the opposite”although no one knows why. (This was
a perverse ordering of preferences when dividends were more heavily taxed, and many
institutions were tax-exempt. It should have been the case that institutions liked divi-
dends and retail shareholders liked dividends”a mystery.) However, some institutional
shareholders are obliged by their charters to hold only dividend-paying stocks. This pro-
vision excluded them from holding such stocks as Microsoft, at least prior to 2003 when
Microsoft initiated its dividends.

There were also other long-term trends at work even before 2003. From 1980 to 2000, share
Disappearing and
reappearing dividends. repurchases had increased from 5% to 50% of earnings, and had become just about as important
as dividends. (Many shares are repurchased but not retired; instead, they are immediately given
out again to compensate employees.) In contrast, dividend initiations had declined”a fact that
is sometimes dubbed “disappearing dividends.” Remarkably, this trend reversed in 2000 as
more ¬rms began to initiate dividends again”i.e., prior to the Bush tax cuts! The reason for
this pattern was the age structure of ¬rms in the market. As many of the new tech ¬rms were
maturing, they started paying dividends. (Of course, the Bush dividend tax cuts further aided
this trend.) In addition, an important factor in the decision of ¬rms to initiate dividends seems
to be the relative premium the market places on dividend payers vs. non-dividend payers. In
the tech bubble of the late 1990s, ¬rms paying dividends did not trade at a higher P/E ratio
than ¬rms not paying dividends. Consequently, ¬rms did not see large valuation incentives for
starting dividends.

26·3.D. Forces Acting Through the Equity Payout Channel

As with our capital structure questions, we would love to know what makes companies pay out
Dividends are paid when
¬rms have cash, but cash. Here, our knowledge is relatively good, especially when it comes to dividends. There are
smoothed out.
two strong empirical regularities:

1. Ordinary (not special!) dividend payouts tend to be persistent: ¬rms are very reluctant
to reduce them. As a result, ¬rms also do not raise ordinary dividends too dramatically,
because this could force them to cut them again in the future. Lintner (1956) ¬rst formally
documented that ¬rms only partly adjust dividends towards a target level if they experi-
ence abnormally high or low earnings. This phenomenon is called dividend smoothing.
Firms still very much smooth dividends, though less so than 50 years ago.

2. Firms pay out money when they have it. That is, they pay out earnings when they have
them and when managers are reasonably con¬dent that good earnings will continue.
There is some debate, however, whether dividend changes signal primarily that future
earnings will be high (so that the dividend yield can continue), or whether they respond
more to past earnings.
The positive association between earnings and dividend payments also means that ¬rms
tend to pay out more in dividends when their stock prices have recently risen. (Interest-
ingly, this may not have debt-equity consequences, because the e¬ect through another
channel runs opposite: ¬rms also issue more new stock in response to good stock re-
¬le=capempirical.tex: RP
Section 26·4. Survey Evidence From CFOs.

Solve Now!
Q 26.4 What deeper characteristics help explain corporate debt-equity ratios?

Q 26.5 How good is our empirical knowledge about the deeper determinants of capital structure?

Q 26.6 Firms with larger tax obligations are known to be more inclined to issue debt. Does this
mean that ¬rms with high tax obligations usually have high debt-ratios?

Q 26.7 If our empirical knowledge about the deeper determinants of capital structure is modest,
does this mean that capital structure theories are irrelevant?

Q 26.8 What remaining tax advantage do share repurchases enjoy over dividends?

Q 26.9 What are the other di¬erences between a share repurchase and a dividend payment?

26·4. Survey Evidence From CFOs

There is another way to approach the question of how managers choose capital structures”just Let™s ask the CFOs.
ask them. Of course, we should not blindly believe that just because CFOs publicly proclaim a
motive, that it is their motive. Graham and Harvey (2001) surveyed 392 CFOs to ¬nd out what
they proclaim makes them issue equity or debt, and they found both interesting and some
rather puzzling results that are di¬cult to interpret.
First, the good news: CFOs do care about the tax bene¬ts of corporate debt, at least moderately. CFOs recognizes taxes
and ¬nancial distress
But they seem more concerned about their credit ratings. We know that credit ratings are closely
related to interest coverage ratios (interest payments divided by earnings) and are a good proxy
for possible ¬nancial distress costs. So, managers are cognizant of the basic trade-o¬ between
taxes and ¬nancial distress.
Now for the bad news: CFOs do not recognize
our other suggestions.
They seem to like
1. Many of our other capital structure arguments seem unimportant to managers, from per- ¬nancial ¬‚exibility
sonal income taxes borne by their shareholders, to expropriation concerns by their credi- (more money, more free
cash ¬‚ow!) and less
tors, to strategic product market factor considerations, to deliberate control of free cash dilution.
¬‚ow incentives, to intentional signaling of good or bad news (inside information), to trans-
action cost considerations.
On the one hand, this may not be as bad as it appears. Managers may still care about
these considerations, because their cost of capital itself re¬‚ects these considerations. (For
example, if their investors face higher tax consequences, it increases the ¬rm™ cost of
capital, and managers do care about their cost of capital.) On the other hand, if a ¬rm
does not need to raise money, it is not clear whether managers compute the appropriate
cost of capital and hurdle rates for their projects. If they do not take these factors into
consideration when estimating the cost of capital that the market would be charging, they
could set too high or too low a project hurdle rate.

2. Managers like “¬nancial ¬‚exibility,” which means that they like having cash around and
having untapped debt capacity for possible future activities. Liking this kind of ¬‚exibility
makes perfect sense from the manager™s perspective”but it also hints that free cash ¬‚ow
is a real problem. Managers seem to primarily like this “¬‚exibility” in order to take over
other companies”a move that is often not value-enhancing for their shareholders. With
almost no chance of bankruptcy in many Fortune 500 companies, it is unlikely that fear
of a cash crunch is the driving concern behind the desire for ¬‚exibility.
¬le=capempirical.tex: LP
672 Chapter 26. How Have Firms™ Capital Structures Evolved?.

3. Managers worry about lower earnings-per-share (called earnings dilution) if they issue
more equity. This makes little sense in itself, because the newly raised funds would
presumably also produce earnings.

4. Even managers who claim to target a debt ratio tend not to retire equity if their equity
has recently increased in value, or to issue more equity if their equity has recently fallen.
This makes little sense, because this is exactly what is required to target a debt ratio.

5. Managers believe that they can time the ¬nancial markets.

• About two-thirds of managers feel that the stock market undervalues their ¬rm”a
fact that restrains many from issuing equity. When their stock market values have
recently increased, then managers feel that they have a “window of opportunity” for
an equity issue. In other words, they believe that they can forecast their stock price,
and the stock market™s usual pessimism is appropriately corrected.
• Even more remarkably, CFOs believe that they can time overall market interest rates:
they issue more debt when interest rates fall or have fallen.

Amazingly, although it seems almost absurd to believe that they have this ability, there is
some new and actively debated empirical evidence that managers have indeed collectively
shown some ability to time the market. To explain such corporate issuing activity and its
success, it appears that we have to look more towards the ¬eld of behavioral ¬nance.

In another survey (by Brav et.al. (2004)) prior to the Bush tax cuts, CFOs generally saw the ques-
A survey on payout
policy. tion of dividends vs. repurchases as one of desirable ¬‚exibility”dividends being steady, share
repurchases being paid “as available.” Other answers mirror those in the Graham and Harvey
survey. Here, too, managers pretty much considered personal income taxes on dividends to be
fairly irrelevant both to them and to the preferences of their shareholders. CFOs also believed
that dividends tended to attract more individual retail shareholders than large institutional
tax-exempt investors. If the CFOs are correct, it is investors who are acting irrationally. Once
again, this seems like a fruitful area of future research for behavioral ¬nance.
Solve Now!
Q 26.10 What factors do CFOs claim matters to them?

Q 26.11 Are answers from managers “prescriptive,” i.e., indicative of what corporations should
¬le=capempirical.tex: RP
Section 26·5. Leverage Ratios By Firm Size, Pro¬tability, and Industry.

26·5. Leverage Ratios By Firm Size, Pro¬tability, and Indus-

Do these dynamics cause di¬erent ¬rms to evolve di¬erent leverage patterns over time? For This section has been
added late, and needs to
example, do larger ¬rms tend to end up having higher or lower leverage ratios? This is a more
be edited.
nuanced question than it appears to be at ¬rst sight, because you can use di¬erent de¬nitions
of debt and equity. In broad strokes, you have three choices:

• You can see debt narrowly in terms of the ¬rm™s ¬nancial indebtedness (long term debt
plus debt in current liabilities), or widely in terms of all liabilities (which includes other
liabilities, such as pension and other liabilities, for example; refer back to the previous
chapter for more detail).
In 2003, just about 50%-60% of the typical ¬rm™s total liabilities was ¬nancial indebtedness.
However, this also di¬ered by ¬rm size. Large ¬rms™ total liabilities were mostly ¬nan-
cial claims, while small ¬rms™ liabilities were mostly other liabilities (such as accounts

• You can see equity in terms of market value or in terms of book value. Although I prefer
the former, the latter is also often used in practice.

• You can compute the leverage ratio by dividing only by the value of the ¬nancial claims
(the sum of ¬nancial debt and ¬nancial equity claim) or more broadly by the value of all

This gives you six possible measures, but we shall look only at three:

A ¬nancial leverage ratio, which we de¬ne as long-term debt plus debt in current liabilities,
divided by the ¬rm™s ¬nancial securities™ market value.

A broad market-value based leverage ratio, which we de¬ne as total liabilities, divided by the
sum of total liabilities plus the market value of equity.

A broad book-value based leverage ratio, which we de¬ne as total liabilities divided by the
bookvalue of assets (the sum of total liabilities plus the book value of equity).

Table 26.3 shows the broad patterns of corporate indebtedness ratios in 2003. (When repeated
for the year 2001, which was a recession year, the table looks very similar.) Publicly traded
¬rms in the United States had average debt ratios of about 25%, although the typical ¬rm had
a median debt ratio of a few percentages less (16% and 20%, based on the measure). There is
quite a bit of variation across ¬rms”one standard deviation is either 26% or 20%, indicating
that debt ratios of 0% or 50% are fairly common, too.
Table 26.3 also shows that the answer to the question of which types of ¬rms are more levered Larger ¬rms tend to
have more ¬nancial
depends on the leverage measure you adopt.
leverage but not more
indebtedness. More
pro¬table ¬rms tend to
By Firm Size Larger ¬rms have higher ¬nancial leverage ratios. However, if you look at leverage
be less indebtedness.
more broadly, you see that larger ¬rms do not tend to be more indebted”and, indeed, by
some measures, seem less indebted.

By Pro¬tability More pro¬table ¬rms are neither more nor less ¬nancially leveraged. However,
on a broader indebtedness measure, more pro¬table ¬rms tend to be less indebted.

Table 26.4 splits ¬rms into industries.

By Industry Consumer goods (drugs, soap, perfumes, tobacco), machinery and business equip-
ment makers, and mining and mineral companies tend to have lower debt ratios (market
value based). Utilities, steel, and automobile companies tend to have higher ¬nancial
leverage ratios. Financial services companies are interesting”they tend to have higher
¬le=capempirical.tex: LP
674 Chapter 26. How Have Firms™ Capital Structures Evolved?.

¬nancial debt ratios, but relatively low broader indebtedness ratios. Note also that there
is great variability across ¬rms”the standard deviation is very high. Even within an in-
dustry, some ¬rms can have very high and others very low debt ratios. (Of course, these
industry de¬nitions are still very broad. It is quite possible that many smaller industries
have their own, unique debt ratios. But even if we use much ¬ner industry de¬nitions, we
would tend to ¬nd great heterogeneity.)
Where do these patterns come from? For utilities, high debt ratios are likely driven by
government regulations. For ¬nancial companies, leverage is part of the business. For
transportation companies, such as airlines, it is partly hard times, partly the fact that
airplanes are easy to collateralize. This is very di¬erent from drug companies, which
tend to have low ¬nancial debt ratios, because R&D is di¬cult to collateralize. Drug
development is either hit or miss. In fact, any debt extended to a drug developer may
almost be called debt. If the drug fails, chances are that creditors will receive zero, just
like equity holders.

In sum, it appears that ¬rm size is a good marker for ¬nancial leverage, and pro¬tability is a
good marker for a broader indebtedness ratio.
¬le=capempirical.tex: RP
Section 26·5. Leverage Ratios By Firm Size, Pro¬tability, and Industry.

Table 26.3. Leverage Ratios by Firm Size and Pro¬tability, in 2003

Financial Broad (MV) Broad (BV)
Mean Mdn S.D. Mean Mdn S.D. Mean Mdn S.D. N
All 25% 17% 22% 16% 26% 21% 4,551
26% 20% 20%

“ $100m 22% 9% 29% 22% 35% 26% 1,419
27% 24% 27%
Market Value

$100m “ $500m 18% 5% 20% 13% 23% 19% 1,249
24% 19% 16%

$500m “ $2,500m 19% 14% 18% 13% 22% 19% 1,083
21% 16% 14%

$2,500m “ $10,000m 24% 19% 20% 16% 22% 20% 493
21% 16% 13%

$10,000m “ 21% 17% 19% 15% 24% 23% 307
18% 14% 12%

Interpretation Increasing Decreasing Weak/Decreasing

“ $100m 16% 4% 23% 16% 33% 25% 1,706
24% 23% 26%

$100m “ $500m 22% 12% 20% 14% 23% 20% 1,223
Book Value

25% 19% 14%

$500m “ $2,500m 31% 27% 21% 15% 21% 19% 950
24% 18% 13%

$2,500m “ $10,000m 34% 31% 24% 20% 23% 20% 426
23% 18% 13%

$10,000 “ 37% 36% 26% 22% 22% 21% 292
24% 17% 11%

Interpretation Increasing Weak/Increasing Decreasing/Weak

Negative 24% 10% 26% 18% 30% 23% 1,639
29% 24% 24%

0% ’ 5% 27% 22% 28% 24% 27% 24% 1,153
24% 18% 15%

5% ’ 10% 20% 15% 17% 14% 24% 20% 844
21% 14% 15%

10% ’ 15% 25% 15% 12% 9% 20% 16% 400
25% 10% 13%

15%’ 28% 24% 13% 7% 22% 15% 562
25% 18% 23%

Interpretation None Decreasing Decreasing

The Financial Leverage ratio is long-term debt plus debt in current liabilities divided by the ¬rm™s ¬nancial securities™
value (the market value of equity plus the long-term debt plus debt in current liabilities). The Broad (MV) Leverage
ratio is ¬rms™ total liabilities divided by the market value of equity plus the book value of total liabilities. The Broad
(BV) Leverage ratio is ¬rms™ total liabilities divided by the book value of assets (which are the sum of the book value
of equity plus the book value of total liabilities. Mdn is the median, S.D. is the standard deviation.) The industry
de¬nitions are originally by Fama and French. The original data came from the Compustat ¬nancial data base.

The table shows that larger ¬rms seem to have higher ¬nancial leverage ratios, but (weakly) lower broad leverage
ratios. More pro¬table ¬rms tend to have lower broad leverage ratios.
Table 26.4. Leverage Ratios by Industry, in 2003

Financial Broad (MV) Broad (BV)
Mean Mdn S.D. Mean Mdn S.D. Mean Mdn S.D. N
1 Food 27% 23% 24% 19% 24% 21% 122
23% 19% 21%

2 Mining and Minerals 15% 6% 13% 7% 16% 12% 94
6% 7% 12%

¬le=capempirical.tex: LP
3 Oil and Petroleum Products 26% 23% 19% 13% 19% 15% 196
23% 13% 15%

4 Textiles, Apparel 25% 16% 27% 19% 24% 21% 89
16% 19% 21%

5 Consumer Durables 27% 18% 31% 22% 27% 23% 119
18% 22% 23%

6 Chemicals 29% 24% 26% 21% 21% 19% 93
24% 21% 19%

7 Drugs, Soap, Prfums, Tobacco 11% 5% 11% 8% 22% 18% 221
5% 8% 18%

8 Construction 31% 25% 32% 25% 27% 23% 106
25% 25% 23%

Chapter 26. How Have Firms™ Capital Structures Evolved?.
9 Steel Works Etc 38% 34% 34% 26% 22% 20% 62
34% 26% 20%

10 Fabricated Products 33% 25% 32% 25% 26% 22% 33
25% 25% 22%

11 Machinery and Business Equip. 15% 6% 19% 13% 26% 21% 693
6% 13% 21%

12 Automobiles 29% 21% 33% 28% 29% 26% 65
21% 28% 26%

13 Transportation 35% 30% 29% 22% 25% 22% 169
30% 22% 22%

14 Utilities 47% 47% 29% 25% 17% 15% 133
47% 25% 15%

15 Retail Stores 23% 14% 28% 21% 29% 25% 293
14% 21% 25%

16 Banks, Insurance, Financials 39% 40% 18% 11% 29% 19% 151
40% 11% 19%

17 Other Industries 18% 7% 21% 14% 28% 22% 1,901
7% 14% 22%

“ Industry Unknown 29% 19% 33% 26% 40% 29% 61
19% 26% 29%

See Table 26.3 for a table description. The table shows that ¬nancial leverage ratios are low for mining, consumer goods, and machinery; and ¬nancial leverage ratios are high among
utilities, ¬nancials, and steel. There are no strong industry patterns on broader indebtedness measures.
¬le=capempirical.tex: RP
Section 26·6. Perspective.

26·6. Perspective

It is important that you keep the empirical evidence in proper perspective. We do know that our What we know.
theories can explain at least some of the behavior of corporations. So, we should not dismiss
them as determinants of observed capital structure. There is a good chance that further re¬ning
of our theories and proxies will explain quite a bit more about how ¬rms behave. We also do
know that we do not know why our theories explain relatively little about the di¬erences in
behavior across companies. There is a good chance that there are other systematic factors that
we do not yet fully understand (probably in the domain of behavioral ¬nance). There is also
a good chance that much corporate behavior is just erratic and will never be explained. We
should keep an open mind.
Why torture you in this chapter with something that we do not fully understand? The reason is It is important that you
know what you do not
that capital structure is an important area, and you must know what we do not yet know! As a
manager, you will meet many investment bankers mustering arguments about what other ¬rms
have been doing, and o¬ering advice as to what you should do. As an investment banker, you
should know not only what factors in¬‚uence ¬rms™ capital structures, but also how important
or unimportant individual factors are”and how you can measure them to ¬nd new potential
clients. As a policy maker, you should know how authoritative the capital structure outcomes
and choices of ¬rms really are.
But perhaps most importantly, the empirical evidence here does not suggest that our theories The evidence says
nothing about the
are worthless. For example, does our empirical evidence mean that just because other ¬rms
normative implications
do not exploit the corporate income tax advantage of debt, that you should ignore it, too? of the theories”in fact,
Absolutely not! You can still think about how important a corporate income tax advantage is to it may tell you where
there is money to be
your ¬rm, and what this means for your optimal capital structure. Perhaps more important”if
many ¬rms are ignoring the factors that they should pay attention to, then over time some will
end up with very poor capital structures. In this case, you can think about how you can come
in and change these existing ¬rms to increase their value. You can e¬ect change from many
di¬erent directions. You can work in the ¬rm itself and argue for a capital structure change.
You can become an investment banker and advise clients on better capital structures. Or, you
can even buy some companies. It has been almost twenty years since there was a wave of
“leveraged buyouts,” in which many public ¬rms were taken over and restructured to generate
value”and much, if not most, of the value was created through better capital structures. Maybe
you will start the next wave of takeovers!
¬le=capempirical.tex: LP
678 Chapter 26. How Have Firms™ Capital Structures Evolved?.

26·7. The Capital Market Response to Issue and Dividend An-

Another interesting question that we can ask is how the ¬nancial markets are responding to
On average, equity
issues are bad news, debt news of an impending equity issue, debt issue, or dividend payment. Is it good news or bad
issues are neutral news,
news? There is a web chapter that describes the evidence in great detail. In short, equity issues
and dividend issues are
are bad news (a loss of 1.5% in ¬rm value, which is around 15% of the size of the issue”an e¬ect
good news.
called dilution). Debt issues are just about neutral. Dividend issues are good news. For issues,
the size of the issue or ¬rm matters little. However, bigger dividends and dividends issued by
small ¬rms are greeted with a relatively more favorable response. There is also considerable
heterogeneity in how ¬rms respond. For example, some ¬rms issuing equity are greeted with
a very positive market reaction.
How should we interpret these ¬ndings? Recall that both debt and equity issues increase the
The evidence suggests
that shareholders like size of the ¬rm, but have opposite e¬ects on ¬rms™ debt ratios. Taken together, the evidence
managers to have less
suggests that increases in debt ratios are good news and increases in ¬rm size are bad news.
cash at their discretion.
For debt issues, the two e¬ects roughly cancel each other out; for equity issues, they act in
the same (negative) direction. This indicates that the market believes that, for the average
publicly traded company, tight ¬nances with high debt burdens and little free cash ¬‚ow enhance
corporate e¬ciency, supporting the agency perspective of capital structure. (The evidence is
also consistent with a corporate tax perspective, and an inside information perspective, but not
a ¬nancial distress costs perspective.) If you want to understand these issues better, please
read the relevant web chapter.

26·8. Summary

This chapter covered the following major points:

• We can explore both the mechanisms of capital structure change and the underlying forces
(causes). These causes can themselves work through multiple mechanisms.

• Over a ¬ve-year horizon, the two most important mechanisms a¬ecting capital structure
are stock returns and long-term debt issuing activity. Both can explain about 40% of the
changes in debt-equity ratios.

• Long-term debt can explain about 30% of the changes in debt-equity ratios, short-term
debt and equity issuing can both explain about 15%, and both convertible debt and payout
policy can explain less than 5%.

• Among the Fortune-100 ¬rms, seasoned equity o¬erings are rare, and appear almost
always in the context of acquisitions. (Executive compensation is remarkably high, and
about as important as share repurchasing activity.)

• We know a number of statistically signi¬cant forces (potential causes), but they can explain
only a very small percentage of capital structure dynamics. Among the more important
in¬‚uences are

“ Stock returns.
“ A reluctance to issue equity.
“ A desire to imitate industry peers.
“ Corporate income taxes.
“ Accounting performance, such as pro¬tability.
¬le=capempirical.tex: RP
Section 26·8. Summary.

“ M&A activity.
“ Financial distress.
“ Market timing.
“ Uncertainty.

• Although dividends and repurchases have been fairly modest in acting as capital struc-
ture channels, we understand them relatively well. They are both about equally important.
Their di¬erences mattered more in the past, before the double taxation of dividends was
reduced in 2003. Among the remaining di¬erences are that managers can participate
in receiving cash from dividends, but cannot tender into share repurchases, that man-
agers with unadjusted stock option plans prefer repurchases to dividends, and that some
investors seem to “like” dividends.

• The most important remaining di¬erence between dividends and share repurchases today
is that dividends tend to be more persistent (“stickier”) than share repurchases.

• Firms tend to pay out dividends when they have retained earnings. That is, they do not
typically ¬nance dividends through other capital markets activity, but through operations.

• In surveys, CFOs claim to be very concerned about their credit ratings and ¬nancial ¬‚exi-
bility. Together with often largely untapped debt capacity, these ¬ndings can be evidence
of signi¬cant free cash ¬‚ow problems. CFOs also claim not to care about taxes borne by
their investors or many other factors suggested by the theories, but they do believe that
they can “time” the market.

• Larger ¬rms tend to have higher ¬nancial leverage ratios, but lower total indebtedness
ratios. More pro¬table ¬rms tend to have lower total indebtedness ratios.

• Even if ¬rms do not seem to act according to the theories, the capital structure theories
still o¬er good guidance about how you can add value by doing things di¬erently.

• The ¬nancial markets respond negatively to the announcement of an equity issue, neu-
trally to the announcement of a debt issue, and positively to the announcement of divi-
dends. However, there is considerable heterogeneity across ¬rms in this response.
¬le=capempirical.tex: LP
680 Chapter 26. How Have Firms™ Capital Structures Evolved?.


A. Appendix: A List of Some Recent Empirical Capital-Structure
Related Publications

Unlike many other subjects of our book”where our knowledge has solidi¬ed over several decades”empirical capital
structure remains a largely unresolved but actively researched area. Much of what I know seems to be unusually
¬‚uid, and has only recently appeared. My summary in this chapter is my own subjective reading thereof.
You will eventually ¬nd references to papers, past and current, for all chapters on the book website. However, to
allow you to make up your own mind on this very unsettled area, I will now break the rule that references are not in
the book but only on the Web site. So here is a short list of papers published after the turn of the millennium. These
papers will in turn reference many related, older, but equally (or possibly more) interesting and relevant papers.

• Franklin Allen and Roni Michaely, 2003. “Payout Policy.” North-Holland Handbook of Economics, ed. Con-
stantinides, Harris, and Stulz.
• Malcolm Baker and Je¬rey Wurgler, 2002. “Market Timing and Capital Structure.” The Journal of Finance
57-1, p.1-32.
• Alon Brav, and John R. Graham and Campbell R. Harvey and Roni Michaely, 2005, “Payout Policy in the 21st
century.” Journal of Financial Economics 77-3, 483-527.
• Raj Chetty, and Emmanuel Saez, 2005, “ Dividend Taxes and Corporate Behavior: Evidence from the 2003
Dividend Tax Cut.” Quarterly Journal of Economics 120-3, 791-833.
• Eugene F. Fama and Kenneth French, 2004, “Financing Decisions: Who Issues Equity?” Journal of Financial
Economics 76, 549-582.
• John R. Graham, 2003. “Taxes and Corporate Finance: A Review.” Review of Financial Studies 16, 1074-1129.
• John R. Graham and Campbell R. Harvey, 2001. “The Theory and Practice of Corporate Finance: Evidence
from the Field.” Journal of Financial Economics 60, 187-243.
• Armen Hovakimian, Timothy C. Opler, and Sheridan Titman, 2001. “The Debt-Equity Choice.” Journal of
Financial and Quantitative Analysis 36, 1-24.
• Brandon Julio and David L. Ikenberry, 2004. “Reappearing Dividends.” Working Paper, UIUC.
• Mark T. Leary and Michael R. Roberts, 2004. “Do Firms Rebalance Their Capital Structures?” The Journal of
Finance 60-6, 2575-2619.
• Peter MacKay and Gordon M. Philips, 2004. “How Does Industry A¬ect Firm Financial Structure?” Review of
Financial Studies 18-4, 1433-1466.
• Ivo Welch, 2004. “Capital Structure and Stock Returns.” Journal of Political Economy 112-1, 106-131.
¬le=capempirical.tex: RP


. 33
( 39)