Capital structure, risk and asymmetric information 
Nikolay Halov
NYU Stern School of Business
[email protected]
Florian Heider
European Central Bank
[email protected]
December 1st, 2005
Abstract
This paper argues that firms may not issue debt in order to avoid the adverse selection
cost of debt. Theory suggests that since debt is a concave claim, it may be mispriced
when outside investors are uninformed about firms’ risk. The empirical literature has
however paid little attention the caveat that the “lemons” problem of external
financing first identified by Myers (1984) only leads to debt issuance, i.e. a pecking
order, if debt is risk free or, if it is risky, that it is not mispriced. This paper therefore
examines whether and for what firms the adverse selection cost of debt is more than a
theoretical possibility? And how does this cost relate to other costs of debt such as
bankruptcy? Absent any direct measure of something that is unknown to investors and
thus cannot be in the econometrician’s information set, we present an extensive
collage of strong and robust evidence in a large unbalanced panel of publicly traded
US firms from 1971 to 2001 that firms avoid issuing debt when the outside market is
likely to know little about their risk.

We thank Heitor Almeida, Dan Bergstresser, Kobi Boudoukh, Vidhan Goyal, Roman Inderst,
Alexander Ljungqvist, Eli Ofek, Daniel Wolfenzon, Jeff Wurgler and seminar participants at NYU, the
ECB, the Norwegian School of Management, the ZEW Mannheim, the University of Frankfurt,
University of Vienna, Cambridge University and conference participants at the EFA 2004, the 2005
ESSFM meeting in Gerzensee and the ASSA/AFE 2005 meetings (Philadelphia) for helpful comments.
-0-
This paper argues that firms avoid issuing debt in order to avoid the adverse selection
cost of debt that arises when outside investors are imperfectly informed about risk.
The starting point is Myers’ (1984) classic intuition that firms issue securities that
carry the smallest adverse selection cost, i.e. that are least likely to be mispriced by
imperfectly informed outside investors. In other words, firms issue securities that
avoid being priced at a discount by investors who protect themselves against buying a
“lemon”.1 The intuition is usually thought to imply a pecking order where debt
dominates because it is thought to be robust to this mispricing (see for example the
recent survey by Frank and Goyal (2005)).
Despite its intuitive appeal, the argument that debt is robust to the adverse selection
problem of outside financing has empirically fallen on hard times. Fama and French
(2002) for example find that “the less levered nonpayers [of dividends] are typically
small growth firms” and that “the least-levered nonpayers make large net new issues
of stock […], even though they appear to have low-risk debt capacity. This is not
proper pecking order behavior” (italics added). Not only is debt issued when it should
not and not issued when it should according to an adverse selection logic, but also its
main alternative, the trade-off theory cannot account for the reluctance of certain
firms to issue debt.2
The empirical literature has however paid surprisingly little attention to the important
caveat that “debt issues can create information problems if the odds of default are
significant. […]. Rational investors will take this behaviour into account in pricing the
risky debt issue.” (Brealy and Myers (2000), p.526). Indeed, theory has pointed out
that in order to arrive at the pecking order starting from Myers’ intuition, one needs to
assume either that i) debt is risk free because there is no investment risk (as in Myers
The argument builds on Akerlov (1970)’s “lemons” problem.
See Graham and Harvey (2001), Frank and Goyal (2003) and Leary and Roberts (2004a) for similar
views. Both these workhorses of capital structure theory appear at odds with some of the data.
1
2
-1-
and Majluf (1984)), or ii) that debt is correctly priced because all firms have the same
risk (Daniel and Titman (1995)) or iii) that debt is not mispriced because uninformed
outside investors do not care about risk when making decisions (Nachman and Noe
(1994)).3
The basic intuition is straightforward. Debt is a concave claim that is going to be
mispriced by uninformed investors, i.e. it has an adverse selection cost, if risk matters,
and the mispricing is more severe if outside investors know less about risk. Our paper
therefore tackles the following questions: is the adverse selection cost of debt only a
theoretical possibility and thus negligible in practice? If not, for what kind of firms
does the adverse selection cost of debt matter? How does the cost of potential
mispricing of debt relate to other costs such as bankruptcy or debt overhang? And, as
a corollary, does the standard pecking order hold for firms whose debt is unlikely to
be mispriced? To the best of our knowledge these questions have not been addressed
before in the empirical capital structure literature.
Our empirical strategy is related to the analysis of Helwege and Liang (1996), ShyamSunder and Myers (1999) and Frank and Goyal (2003). Their tests are based on how
firms finance their need for external capital. Using statement of cash-flow data, they
construct a measure of this need, the financing deficit, and analyze to what extent
firms issue debt to finance the deficit. Our innovation is to condition the sensitivity of
debt issuance with respect to the financing deficit on different measures that capture
whether, and to what extent, the outside market does not know the risk of a firm. We
will show that, controlling for other costs of debt, firms issue less debt to finance their
deficit if outside investors know less about risk.
3
Nachman and Noe show that debt is only information insensitive if the distribution of cash-flows
exhibits Conditional First-Order Stochastic Dominance, which means that investors price securities in
their model independent of risk. Note also that the lognormal distribution need not satisfy this condition
which illustrates the difficulty of applying Myers and Majluf (1984) appeal to option pricing theory to
argue for the optimality of debt.
-2-
A major challenge is that any measure of asymmetric information about risk must be
an indirect one since something that is not known to investors cannot be in the
econometrician’s information set. Absent a direct measure, we use a number of
different indirect measures and employ several approaches, using i) firms’ recent asset
volatilities, ii) changes of implied volatilities from option prices and iii) the impact of
credit ratings to establish an extensive collage of robust evidence for an adverse
selection cost of debt.
The idea behind using recent asset volatilities is that an outside investor knows less
about a firm’s investment risk if the firm’s asset value has fluctuated a lot prior to an
issue. A concern of that particular measure is of course that asset volatilities are also
linked to the probability of going bankrupt, and thus to the bankruptcy cost of debt.
We show however that the probability of going bankrupt, as measured by the
modified Z-score (see MacKie-Mason (1990)), first decreases and only later increases
with asset volatility. The firms that are most likely to go bankrupt are both firms with
assets fluctuate a lot and those whose assets fluctuate little. We also follow the
procedures in Frank and Goyal (2003) to control for conventional cross-sectional
determinants of leverage and those in Lemmon and Zender (2002) to control for debt
capacity concerns. In all specifications we find strong and robust evidence in favor of
an adverse selection cost of debt.
In addition to asset fluctuations, we employ two other measures that allow us to
disentangle the adverse selection from the bankruptcy cost of debt. First, using option
pricing data, we calculate the change of firms’ implied volatilities from the time series
of daily option prices just prior to issuing securities. Outside investors know less
about firm risk if option prices indicate that the market’s assessment of future risk has
changed often. At the same time, it seems implausible that a firm does not issue debt
-3-
following a recent change in the market’s assessment of its future volatility – the
change could be a decrease of implied volatility - because of expected bankruptcy
costs.
Second, we show that firms with any credit rating issue mostly debt to finance their
deficit irrespective of whether the outside market is imperfectly informed about risk.
At the same time, there is a strong negative relationship for firms without a rating
between the extent to which they issue debt to finance their deficit and measures of
how little the outside market knows about their risk. Since having a rating does not
affect the probability of default, it is unlikely that firms do not issue debt because of
the bankruptcy cost of debt alone.4 Instead, the evidence is consistent with the idea
that credit ratings bridge the information gap about risk and that having a rating
allows firms to avoid the adverse selection cost of debt.
The overall evidence provides the following answers to our initial questions: i) the
adverse selection cost of debt is not just a theoretical possibility,5 ii) it matters most
for those firms for whom the outside market knows little about their risk, iii) the
adverse selection cost of debt complements other costs of debt, e.g. the cost of
bankruptcy (for firms with few tangible assets or with very low Z-scores) and debt
overhang (for firms with the highest market-to-book ratios), and iv) the standard
pecking order is indeed a special case that applies only when debt is unlikely to be
mispriced.
4
We control for the fact that large firms with more tangible assets have more debt and at the same time
are more likely to have a rating (see Faulkender and Petersen (2005)).
5
Note that Myers’ (1984) adverse selection intuition is orthogonal to “signaling”. He argues that firms
issue information insensitive securities while signaling requires firms to issue securities that have
maximal sensitivity. Firms need to issue “worst case financings” in order to credibly signal their inside
information (see Brennan and Kraus (1987)). While the literature has identified many reasons why
signaling is less plausible for capital structure decisions (theoretically unstable (Nachman and Noe
(1994)), not appealing in the context of financial market behavior (Cadsby et al. (1990)) or not a major
concern for CFOs (Graham and Harvey (2001)), these counter-arguments do not apply to Myers’
(1984) intuition (see also Barclay and Smith (1999)). Generally speaking, signaling generates the
prediction that riskier firms issue more debt-like securities (as this is their worst case financing). This is
the opposite of what we observe empirically.
-4-
Our evidence fills a gap in the capital structure literature. An adverse selection cost of
debt is a previously undocumented channel through which risk affects capital
structure decisions of publicly traded firms. The standard argument of how risk affects
capital structure is based on the trade-off between the tax benefits and the bankruptcy
costs of debt. The trade-off theory itself and related issues such as debt conservatism,
the magnitude of bankruptcy costs and the existence of target leverage ratios are
subject of intense debate in the literature.6 We note here that linking capital structure
to risk has been difficult in the past. The survey by Harris and Raviv (1991) shows
that the evidence is mixed. Rajan and Zingales (1995) exclude measures of risk
arguing that traditional measures of risk such as size or the volatility of earnings are
too imprecise. This papers offers evidence that risk affects capital structure via an
adverse selection channel.
The argument that firms avoid debt if it is likely to be mispriced also seems
reasonable in the light of the evidence gathered from surveys (see for example
Graham and Harvey (2001)). They find that flexibility, credit ratings and volatility are
the top 3 priorities for CFOs when asked about factors that affect their decision to
issue debt. All three are consistent with our argument and the evidence. In contrast,
bankruptcy costs rank second to last in the list of CFOs’ priorities.
Our paper can also help making the aforementioned stylized fact that small young
non-dividend paying firms issue equity, despite not having exhausted their debt
capacity, while large mature dividend-payers issue debt less puzzling. The fact is not
inconsistent with asymmetric information once one recognizes that the standard
6
See Frank and Goyal (2005) for a comprehensive survey and discussion. See Hovakimian et al.
(2001), Mayer and Sussman (2002) and Hovakimian et al. (2004) for the horse race between the
standard pecking order and the trade-off theory as well as efforts to combine them empirically. See
Welch (2003), Flannery and Rangan (2005), Kayhan and Titman (2003) and , Leary and Roberts
(2004b) for the question whether there are “target” levels of leverage as predicted by the trade-off
theory and if yes, what do firms do to reach them?
-5-
pecking order is only a special case of Myers (1984) logic that should not apply
generally. At the same time, trade-off arguments focus only the firm’s view, i.e. its
desire to avoid the cost of bankruptcy. Our paper complements this by adding the
market’s view, i.e. the market’s pricing of debt given that it may know little about the
firm’s risk.
The organization of the paper is as follows. Section 1 develops our empirical strategy.
Section 2 describes the sample and presents some descriptive statistics. Section 3
contains the main empirical results. Their robustness and possible alternative
explanations are analyzed in section 4. Section 5 concludes.
1. Empirical strategy
This section presents and discusses our empirical strategy to test for a possible
adverse selection cost of debt. It builds upon the methodology of Shyam-Sunder and
Myers (1999) and Frank and Goyal (2003) (see also Helwege and Liang (1996)).
Shyam-Sunder and Myers (1999) propose a test based on cash-flows (changes) rather
than stock variables (levels) of how firms finance their need for external capital.
Consider the following accounting identity of cash flows:
DEF  I  DIV  W  C  D  E
A firm’s financing deficit DEF, i.e. the difference between uses of funds (dividends
DIV, investment I and changes in net working capital W) and internal sources of
funds (the internal cash-flow C), must be balanced by external sources of funds, i.e.
either the issuance of debt D or equity E (we follow the definitions of Frank and
Goyal (2003) – see the appendix; see also Helwege and Liang (1996), Shyam-Sunder
and Myers (1999), and Chang and Dasgupta (2003)).
Since Shyam-Sunder and Myers (1999) and Frank and Goyal (2003) assume that the
adverse selection problem of external financing automatically leads to the standard
-6-
(1)
pecking order in which debt dominates equity, they run the following pooled panel
regression
Dit  a  bDEFit  
(2)
and argue that there is support for the standard pecking order if a=0 and b is close to
one.
But once risk is imperfectly known by outside investors, debt will have an adverse
selection cost so that one cannot generally expect (2) to return a coefficient b that is
close one. We therefore employ (2) conditionally by ranking firms into deciles, n=1,
2…10, according to measures that proxy for the role of risk in the asymmetric
information problem (we discuss these measures in the next section), and then run
regression (2) separately in each decile n: 7,8
Dit  a n  bnD DEFit  
Our hypothesis is that the estimated coefficients on the financing deficit can be ranked
monotonically: bˆ1D  bˆ2D    bˆ10D . The market knows less about risk for those firms
that have been ranked into higher deciles. These firms faces a higher adverse selection
cost of debt and avoid issuing mispriced deb to finance the deficit. As a corollary we
expect that bˆ1D is close to one since the standard pecking order is a special case that
applies only when investors are well informed about risk.9
7
We also ran the parametric alternative with an interaction term of the deficit with proxies for the role
of risk in the asymmetric information problem. The advantage of the semi-parametric version (3) is that
one does not have to assume that the impact of not knowing risk is linear.
8
Chirinko and Singha (2000) point out that running regression (2) on the entire sample when there is a
significant amount of equity financing biases the coefficient b towards zero. This observation
reinforces our argument that one should run the conditional version (3).
9
In addition to (3), we also test the extent to which equity is issued to finance the deficit in each decile.
Since (1) is an accounting identity, checking that the estimated coefficients on the deficit from debt and
equity issuance add up to one in each decile is a useful test of the accuracy of the cash-flow data.
-7-
(3)
1.1 Proxies for asymmetric information about risk
The hypothesis is that asymmetric information about risk drives up the cost of issuing
debt due to a lemons discount. Any measure of asymmetric information must however
be indirect since something that is not known to the market cannot be in the
econometrician’s information set. Our approach therefore is to use several different,
indirect measures in order to capture the extent to which outside investors are
imperfectly informed about firms’ risk. We use i) firms’ recent asset volatilities and
their changes, ii) times series of implied volatilities from option prices and iii) the
impact of credit ratings.
Recent asset volatility is a readily applicable measure that we expect to be correlated
with asymmetric information about risk. We use firms’ recent volatility of assets to
group them into deciles and argue that the outside capital market knows less about the
risk of future investments for firms that have been ranked into higher deciles. In other
words, we argue that when raising external financing, firms whose asset values have
fluctuated a lot prior to the issue face a higher adverse selection cost of debt than
firms whose asset values have been stable.
Of course, using recent asset volatility to group firms into deciles is going to be an
imperfect measure. The immediate concern is that it is also picking up the probability
of going bankrupt. We address this concern in several ways. First, our results continue
to hold if we use changes of asset volatilities instead of levels. Thus, a firm could be
sorted into a high decile because it experienced a decrease of its asset volatility.
Second, we show that firms in higher risk deciles do not necessarily have a higher
probability of bankruptcy as measured by the modified Z-score (MacKie-Mason
(1990)). Third, we control for factors that affect the expected cost of bankruptcy such
as profitability, size, tangibility of assets and the value of growth options. Finally, and
-8-
most importantly, we employ alternative measures to capture the degree to which the
outside market knows less about risk.
Using option prices, it is unlikely that we inadvertently measure the probability of
bankruptcy. We calculate the time series standard deviation of a firm’s volatility
implied from option prices using data from Optionmetrics’ IvyDB prior to issuing
securities (the details of this calculation are found in the appendix). Our results hold if
we use this alternative measure to rank firm into deciles. The advantage of this
measure is that it captures changes in the market’s assessment of a firm’s future risk.
Controlling for factors such as tangibility and profitability, it seems unlikely that a
firm’s decision not to issue debt following a change (i.e. even a decrease) in the
implied volatility of its options can be explained by the bankruptcy cost of debt. The
disadvantage of this measure is that option price data is only available for a smaller
number of firms from 1996 to 2001.
We also show that ranking firms into risk deciles has no impact on how the deficit is
financed for firms that have any credit rating, but it has a strong impact on firms that
do not have one. The former issue mostly debt to finance their deficit irrespective of
their risk group, while the latter issue less debt in higher risk deciles. This difference
is inconsistent with the idea that firms in higher risk deciles issue less debt only to
avoid bankruptcy. We argue instead that credit ratings bridge the information gap
about risk and therefore prevent the mispricing of debt.
1.2 Recent asset volatility measures
We construct two measures of asset volatility. The first one consists of unlevering the
volatility of equity. Unlevering is needed since the volatility of equity mechanically
-9-
increases with leverage ceteris paribus.10 We compute the standard deviation of the
daily return on the market value of a firm. The market value of assets is defined as in
Fama and French (2002) (see also our appendix).11 If there are less than 90 days of
stock price data, the firm/year observation is deleted from the sample. The second
measure recognizes that equity is a call option on the value of firm assets with the
exercise price being the value of the debt (Merton (1974)). 12 The exact procedure is
explained in the appendix.
The Spearman rank correlation between the two measures of asset risk in our sample
is 0.95. Given that both measures give virtually identical rankings, we only report
results using the simpler first measure.
To ensure that there is no contemporaneous interplay between the issue decision and
asset volatility, we use last year’s asset volatility. Using longer lags would weaken the
link between the role of risk in the adverse selection problem and the current capital
structure decision.13
1.3 Other determinants of leverage
A pure adverse selection model of firms’ capital structure decisions is based on
information frictions at the moment when firms contact the external capital market. It
uses a different set of variables than conventional, mostly cross-sectional empirical
research on the level of debt that is usually rooted in the trade-off theory (see also
Frank and Goyal (2003)). The basic trade-off theory states that the level of leverage is
10
. We also use unadjusted equity volatility and the results are, as expected, stronger. This rules out that
the unlevering procedure drives our results.
11
We also try the definition of Baker and Wurgler (2002), which excludes convertible debt, and also
try using just total liabilities. The results are not affected.
12
An advantage of the Merton method is that we can use the CRSP return series that is adjusted for
stock splits and dividends.
13
There is an issue concerning the overlap or gap between the calendar year used for stock price data
and the fiscal year used for financial data. This overlap or gap exists for 48% of all firms. We check the
robustness of our results by using only firms whose fiscal year is the calendar year. The results are
unchanged.
- 10 -
determined by trading off the tax benefit of debt against the expected cost of financial
distress (see for example Bradley et al. (1984)). Hence, firms with a high present
value of tax benefits and/or a low present value of expected distress costs should have
higher levels of debt. Rajan and Zingales (1995) narrow the list of conventional
determinants down to four main variables: profits, size, tangibility of assets and the
market-to-book ratio.
More tangible assets support debt because it means that firms can collateralize the
debt which reduces bankruptcy costs. The market-to-book ratio is usually seen as a
proxy for growth opportunities that should be negatively related to leverage. The
argument is that leverage exposes firms to the “debt overhang” problem (Myers 1977)
and that the future value of the firm is lost in bankruptcy. A recent alternative
explanation for a negative relationship is market timing. Firms with a high market-tobook ratio are overvalued and hence issue equity to take advantage of it (Baker and
Wurgler (2002)). Sales are usually positively associated with leverage. There is no
clear theoretical foundation but one normally argues that larger firms have a higher
reputation or are less likely to go bankrupt so they can borrow more.
Profits show up regularly as a negative determinant of leverage. Traditionally this has
been seen as a challenge for conventional trade-off models of leverage. They predict
that more profitable firms should issue more debt since more profitable firms have a
smaller risk of bankruptcy and have more taxable income to shield (see Titman and
Wessels (1988) and Fama and French (2002)).14
14
Recent dynamic trade-off theories can predict a negative relationship between profitability and
leverage (for example Strebulaev (2004) and Hennessy and Whited (2004)).
- 11 -
In order to control for conventional cross-sectional determinants of leverage, we
follow the methodology of Frank and Goyal (2003) and add first-differences of the
conventional determinants to (3).15 The set of decile regressions (3) then becomes:
Dit  a n  bn DEFit  bn
 bn
LOGSALES
TANG
TANGit  bn
MTB
MTBit  bn
PROF
PROFit
LOGSALES it  
Our hypothesis is that the monotonic ranking of the estimated coefficients on the
financing deficit across risk deciles, bˆ1D  bˆ2D    bˆ10D , remains once we add the
conventional determinants of leverage.
1.4 Debt capacity considerations
Lemmon and Zender (2002) find that the same factors that determine the level of debt
in cross-sections, also influence the sensitivity of debt issuance with respect to the
financing deficit. It is not surprising that the standard cross-sectional determinants of
leverage, e.g. firm size, age, tangibility and the market-to-book ratio, also affect the
cost of issuing debt, i.e. a firm’ s debt capacity.
We control for other factors besides adverse selection limiting firms’ debt capacity by
first sorting firms into market-to-book, size, age or tangibility groups j and then
ranking them into asset volatility groups i. Then we run regression (2) in each group
ij. For example, if we sort firms into quintiles, we run 5*5=25 separate regressions
D
and obtain 25 coefficients on the financing deficit b̂ij , i,j=1,…,5. The use of double
sorting allows for a non-linear dependence of the sensitivity of debt issuance with
respect to the financing deficit on these other debt capacity factors. Our hypothesis is
15
We confirm the standard signs and statistical significance on the conventional variables in a version
of (4) without the financing deficit that is applied to the entire sample (i.e. all deciles together). We also
added lagged leverage as an explanatory variable and find that is has a negative coefficient. This
confirms the findings of Flannery and Rangan (2003) and Kayhan and Titman (2003) that there is mean
reversion of leverage.
- 12 -
(4)
that no matter what control group j firms have been sorted into, they all have a
sensitivity of debt issuance to the financing deficit that decreases in higher risk
deciles, i.e. bˆ1Dj    bˆ5Dj for all j=1,…,5.
2. Data
We study a large, unbalanced panel of all firms from the merged CRSP-Compustat
(CCM) database from 1971 to 2001.16 We make the following standard adjustments.
We exclude financial firms (SIC codes 6000-6999), regulated utilities (SIC codes
4900-4999), and firms involved in major mergers and acquisitions (Compustat
footnote code AB). Furthermore, we exclude firm/year observations that report cash
flows data using format code (item 318) 4 or 6 (both undefined by Compustat) and 5
(for the Canadian file) or if the format code is missing.
To be able to link Compustat reliably to CRSP data we use only records with link type
‘LC', 'LN', 'LO', 'LS', 'LU' or ’LX’. A small number of CRSP securities that link into
more than one Compustat firm have also been deleted.
In order to remove outliers and misrecorded data, we remove observations for certain
variables that have missing values or are in the extreme 0.5 % left or right tail of the
distribution (see the appendix for the list of variables that have been treated this way).
To ensure that the sample does not contain equity issues due to IPOs, we exclude
observations for the year in which a firm’s stock price becomes first available in the
CRSP database. The maximum number of observations in our sample then is 103,351
firm-years.
16
Our sample only starts in 1971 since we require cash flow data.
- 13 -
Table 1 shows balance sheets, cash flows and other firm characteristics at the
beginning and at the end of our sample period, 1971 and 2001, as well as for two
intermediate dates, 1980 and 1990.
Table 1: Balance sheets, cash flows and other firm characteristics over time
Panel A presents average balance sheets and panel B shows the average of the cash
flows. Confirming the finding of Frank and Goyal (2003, 2005), the table shows that
equity plays an important role in financing the deficit.
Note also the difference between the mean and the median of net debt and equity
issues. The median is zero for both. A typical firm appears to stay out of the market
for external finance most of the time, but if it does seek external finance, the
magnitude of the market intervention is large relative to firm size.
3. Analysis
We argued that the standard pecking order should not be a good description of debt
issuance for all firms in the sample because it should only work well for those firms
that have the smallest adverse selection cost of debt. Running regression (3) on the
full sample (pooled OLS standard error in brackets) confirms this:
Dˆ it  0.004  0.375DEFit
(0.000) (0.002)
(R 2  0.36)
The coefficient on the financing deficit is only half of the 0.75 (R2 of 0.68) reported
by Shyam-Sunder and Myers (1999) on the much smaller sample of 157 firms in
Compustat with continuous reporting from 1971 to 1989. Thus, we confirm the result
- 14 -
of Frank and Goyal (2003) that the support for the standard pecking order in ShyamSunder and Myers does not carry over to a broader sample of firms.17
Our interpretation of this finding is however very different. Frank and Goyal interpret
it as evidence against the standard pecking order, and thus also as evidence against the
argument that capital structure decisions are affected by adverse selection costs. We
agree that the standard pecking order does not work for a broad sample of firms. But
since we pointed out that the pecking order is only a special case, one should not
expect it to hold for all firms in the first place. Evidence against the pecking order
does not imply that adverse selection does not matter when issuing securities.
In order to examine differential financing policies of firms with different levels of
adverse selection costs of debt, we rank firms each year into deciles according to our
proxy for the extent to which the outside market does not know firms’ risk. Table 2
shows balance sheets, cash flows and other descriptive statistics across deciles using
recent asset volatilities to rank firms.
Table 2: Balance sheets, cash flows and other descriptive statistics across deciles
Firms in higher deciles have more cash on their balance sheet whereas differences in
tangibles (i.e. net property, plant and equipment) and intangibles are small (panel A).
As far as liabilities are concerned, firms in higher deciles have roughly the same
amount of short-term as and less long-term debt than firms in lower deciles.
Comparing cash flows across deciles reveals a hump shaped pattern for dividends and
internal cash flows (panel B). We also find that the median internal cash flow in the
highest decile is larger than in the lowest decile (not shown in the table).
17
Our coefficient is only slightly larger than the 0.28 (R2 of 0.14) reported by Frank and Goyal using
an unbalanced panel from 1971-89.The difference seems to come from the different time period and the
fact that our requirement about the availability of stock price data eliminates a number of small firms
from the sample.
- 15 -
The average financing deficit of firms in higher deciles increases, but the median
financing deficit remains close to zero except for the three highest deciles. Average
net debt and equity issues both increase for firms in higher deciles and the increase is
more pronounced for equity. The medians however are mostly zero. This again
indicates that a typical firm is reluctant to contact the external capital market, but if it
does raise external capital, the size of the issue is large.
Firms in higher deciles are younger, smaller and have higher market-to-book ratios
(panel C). Note however that profitability and modified Altman’s Z-scores (see
MacKie-Mason (1990)) first increase and then decrease across risk deciles. Firms in
higher asset volatility deciles are therefore not less profitable or more likely to go
bankrupt than firms in lower deciles.
Table 2 also shows that there is more dispersion of asset volatilities within higher
deciles and that firms in higher deciles have a larger changes of their implied
volatilities. Together with the pattern on profitability and the modified Z-score, we do
therefore not expect to inadvertently pick up the probability of default when ranking
firms into risk deciles.
The central result
Table 3 contains the central result of our paper. It shows the results from running
regression (3) in each decile.18
18
The table reports OLS standard errors. We also computed White standard errors that correct for
heteroscedasticity. The corrected errors are about three to four times larger, which does not affect our
conclusions.
- 16 -
Table 3: Financing the deficit across deciles
The table shows support for our hypothesis. Firms from higher deciles issue
monotonically less debt to finance their deficit. In the lowest decile, a one standard
deviation change of the financing deficit from its mean produces roughly a one
standard deviation change of net debt issues. In the highest decile, a one standard
deviation change from the mean deficit increases net debt issues by about a third of a
standard deviation.
To illustrate the result, we plot the coefficients on the financing deficit and the
associated R2 from Table 3 in Figure 1.
Figure 1: Financing the deficit across deciles
Note that the estimated intercept is close to zero across all deciles. This suggests that
there is no factor that is common to all firms in a decile throughout the sample period
that could affect the pattern of net debt issues. Furthermore, the estimated coefficients
on the deficit from the net debt and the net equity regression add up to one across
deciles. This indicates that we are not missing cash-flows.
Note also that the standard pecking order now works very well in the lowest decile.
The coefficient on the financing deficit in the lowest decile is 0.87 (R2= 0.85), which
is larger than the 0.75 obtained by Shyam-Sunder and Myers (1999) on a small
subsample of 157 firms with continuous reporting. This supports the argument that the
standard pecking order is indeed a special case.
In addition to the regression result, Table 4 shows the proportion of companies that
either issue debt, equity or do nothing in each decile.19
19
Issuing debt or equity is defined as a change in D or E that exceeds 1% of book assets.
- 17 -
Table 4: Issue decisions across deciles
The proportion of debt issues decreases across deciles while the proportion of equity
issues increases, which lends further support for our hypothesis that firms in higher
deciles issue less debt.
Credit ratings
If firms do not issue debt to avoid its adverse selection cost, and if the adverse
selection cost of debt is caused by being imperfectly informed about risk, then we
should find that
i) firms for whom the market is well informed about their risk should issue a
lot of debt irrespective of the decile they have been ranked in, and
conversely that
ii) firms for whom the market is ill informed about their risk should issue
monotonically less debt if they have been ranked in higher deciles,
We use the presence/absence of an S&P credit rating to distinguish between firms in
i) and ii). The argument is that rating agencies bridge the information gap for risk that
exists between firms and the outside capital market. Table 5 and Figure 2 show
support for both i) and ii). Firms with an S&P rating have a sensitivity of debt
issuance on the financing deficit of 0.77 or larger for any decile up to the nineth one.
In contrast, firms without a credit rating exhibit a steady decrease of their debt
sensitivity across deciles from 0.87 in the lowest to 0.15 to the highest decile.
- 18 -
Table 5: Financing the deficit of rated and unrated firms across deciles
Figure 2: Financing the deficit of rated and unrated firms across deciles
Faulkender and Petersen (2004) show that larger firms with more tangible assets are
more likely to have a credit rating. Large firms and firms that have a lot of tangible
assets are also likely to use more debt financing. To ensure that the result in figure 2 is
not driven by the fact that rated firms in higher risk deciles are larger or have more
tangible assets than unrated firms in the same risk decile, we split the entire sample
into firms with and without a credit rating and run the following regression on both
subsamples:
Dit  a  bDEFit  b RISK DEFit * LNRISK it  b SIZE DEFit * LNSIZEit
 bTANG DEFit * TANGit   it
The regression allows the coefficient on financing deficit to depend on i) recent asset
volatility, ii) size and iii) the tangibility of assets.20 The results are shown in Table 6:
Table 6: The impact of risk on the deficit coefficient for rated and unrated firms
The coefficient on DEF*LNRISK is negative and it is more than twice as large for
unrated than for rated firms. Thus, controlling for size and tangibility, we find that
firms with any rating issue a lot of debt to finance their deficit, and that the decision to
issue debt is much less affected by past variations of asset values than for firms that
do not possess a rating. Since having a rating does not affect the probability of default,
20
We also controlled for size and tangibility non-parametrically by sorting firms into 25 size-tangibility
groups, then within each of these 25 groups running regression (3) for rated and unrated firms across
risk deciles (i.e. 25*2*10=500 regressions) and finally averaging the 25 coefficients on the deficit to
obtain a version of figure 2 that controls for size and tangibility. The result is very similar to figure 2.
- 19 -
(5)
it is difficult to see how these results could be explained by the bankruptcy cost of
debt alone.21
4. Robustness
The pooled panel regression (3) is the simplest possible tests of our hypothesis. We
now perform a series of robustness checks to see whether the simple model is
misspecified and whether alternative theories of the issuing decision can explain our
results.
In order to address the potential problem of cross-sectional correlation in a pooled
panel regression, we follow Fama and French (2002) and use the Fama-McBeth
procedure (Fama and McBeth (1973)). The procedure consists of running a crosssectional regression for each year, reporting the average of the cross-sectional
coefficient estimates and using the time-series standard deviations of the crosssectional estimates to calculate standard errors.22 In addition, we also estimate our
base model (3) using both firm and year fixed effects to control for time and firm
invariant unobservable factors affecting debt and equity issuance.
The results of performing each procedure are shown in Table 7.
21
We present more robustness checks using credit ratings and proxies for the bankruptcy cost of debt
below in section 4.
22
We also analyze the autocorrelation in the time series of the cross-sectional estimates. The first-order
autocorrelation is sometimes as large as 0.8. Sometimes it is statistically insignificant from zero. We
address the issue by fitting an AR(1) process to the time series of cross-section coefficients on the
financing deficit and then inflate the standard errors using the information on the auto-correlation. The
result is an increase of the standard errors by a factor 3 to 4.
- 20 -
Table 7: Financing the deficit across deciles: Fama-McBeth and fixed effects
procedures
Table 7 confirms that the evidence is very robust to alternative specifications. The
sensitivity of debt issuance to the financing deficit decreases monotonically across
deciles, from 0.87-0.89 for the lowest decile to 0.16-0.31 in the highest decile.
In order to control for conventional cross-sectional determinants of leverage, we run
regression (4), i.e. net debt issues on first-differences of the conventional determinants
of leverage and the deficit, in each decile.
Table 8: Regression of net debt issues on conventional variables and the
financing deficit across deciles
Table 8 shows that controlling for conventional cross-sectional determinants of
leverage variables does not change our estimates of firms’ sensitivity of debt issuance
to the financing deficit across deciles.
Next we allow for the possibility that there are other factors that affect the sensitivity
of debt issuance with respect to the financing deficit, i.e. we control for debt capacity
concerns. The results from applying the double sorting procedure described in section
1.4, shown below in Tables 9 to 12 and Figures 3 to 6, confirm that firms in higher
asset volatility deciles generally issue less debt to finance their deficit, even after
controlling for other debt capacity concerns.
- 21 -
Table 9: Financing the deficit across size and asset volatility quintiles
Figure 3: Financing the deficit across size and asset volatility quintiles
The negative relationship across risk groups is found in all size groups and it is
stronger for smaller firms except the very smallest. The largest and the smallest firms
stand out as they generally use more debt and more equity financing respectively.
Table 10 and Figure 4 show that irrespective of their age, firms from higher risk
deciles have lower sensitivity of debt issuance to the financing deficit the results of
controlling for age. The oldest firms appear to issue relatively more debt than others.
Table 10: Financing the deficit across age and risk quintile
Figure 4: Financing the deficit across age and risk quintile
Table 11 and Figure 5 present the results for tangibility. Again, the monotonic
negative pattern of the coefficient on the financing deficit across asset volatility
groups is present in each tangibility group. Note also that this negative relationship is
stronger for firms with fewer tangible assets.
Table 11: Financing the deficit across tangibility and asset volatility quintiles
Figure 5: Financing the deficit across tangibility and asset volatility quintiles
Finally, Table 12 and Figure 6 confirm that the negative monotone negative pattern of
the coefficient on the financing deficit across asset volatility groups also holds in all
market-to-book quintiles. Firms with the highest market-to-book ratios stand out since
- 22 -
they issue much less debt than other firms. This could be an indication that these firms
try to avoid a debt overhang problem that would limit their future growth potential.23
Table 12: Financing the deficit across market-to-book and asset volatility
quintiles
Figure 6: Financing the deficit across market-to-book and asset volatility
quintiles
To sum up, controlling for debt capacity concerns, we find robust evidence that firms
ranked into higher risk deciles issue less debt to finance their deficit. In addition, our
semi-parametric procedure allows us to identify certain groups of firms that stand out.
The largest and the oldest firms use relatively more debt, while the smallest firms and
those with the highest market-to-book ratios use relatively more equity.
Overall, the evidence suggests that there is an adverse selection cost of debt.
However, other factors that make debt costly such as bankruptcy, debt overhang or a
lack of reputation seem to play a role, in particular for certain subgroups of firms, e.g.
those with the highest market-to-book ratios.
Since the adverse selection cost of debt is caused by being imperfectly informed about
risk, we argued that the standard pecking order should be seen as a special case that
applies only for those firms with the lowest adverse selection cost of debt. Tables 9 to
12 and Figures 3 to 6 indeed show that firms in lowest decile have very high debt to
deficit sensitivities (except for the smallest firms or those with the highest market-tobook ratios). For example, controlling for size, the coefficient on the financing deficit
is between 0.83 and 0.87 in the lowest asset volatility group, except for the lowest size
quintile where the coefficient drops to 0.50. This does not support the argument that
23
Another possibility could be that the most overvalued firms time the equity market (see Baker and
Wurgler (2002)).
- 23 -
the standard pecking order only works for larger firms (e.g. Frank and Goyal (2003,
table 6), provided that one controls for fluctuations in asset volatilities and that one
excludes the smallest firms.
A further alternative to examine the role of debt capacity concerns for our results is to
examine the subsample of firms with investment grade debt, i.e. firms that should
have relatively large debt capacities.24 Table 13 shows that regression (3) continues to
produce the strong negative relationship between asset fluctuations and the sensitivity
of debt issuance to the financing deficit for firms that are unlikely to default. This is
further evidence that the bankruptcy cost of debt by itself cannot be responsible for
firms not issuing debt.
Table 13: Financing the deficit across deciles for firms with investment grade
debt capacity
Next, we explore in more detail the relationship between the adverse selection and the
bankruptcy cost of debt. Although both are linked to risk, the two costs are distinct.25
The expected cost of bankruptcy depends on the probability of default and the cost of
bankruptcy to the firm, while the adverse selection cost depends on the degree to
which the market is imperfectly informed about the future pay-off of holding a
concave claim. To price debt the market must not only consider the downside risk but
also the upside potential since holding debt means foregoing the latter.
To disentangle an adverse selection cost of debt from the cost of going bankrupt, we
again use the presence (absence) of an S&P credit rating to identify firms for whom
investors are well (ill) informed about risk. Then we perform a double-sorting,
24
We say that a firm has investment grade debt capacity if its unlevered Z-score is larger than 1.67.
This cut-off corresponds to the median Z-score of those firms that do have an available S&P rating of
BBB.
25
We already saw in Table 2 that the Z-score is not collinear with changes of asset volatilities or
changes of implied volatilities.
- 24 -
ranking firms first by their modified Z-score in groups j and then by their asset
volatility in groups i, for both the subsample of firms with a rating, k=1, and those
without one, k=2. Using for example quintiles, we thus run regression (3) on
5*5*2=50 different subsamples and compare the coefficients bijk.
The hypothesis is that one should find different comparative statics of the sensitivity
of debt issuance to the financing deficit with respect to the risk decile for firms with a
rating compared to those without one. More precisely, we expect to expect a negative
pattern for all Z-score groups only for unrated firms, i.e. bˆ1Djk    bˆ5Djk for all
j=1,…,5 when k=2 but not when k=1.
The results are shown in Table 14-15 and Figure 7-8 show support for the hypothesis.
Table 14: Financing the deficit across Z-score and asset volatility quintiles –
unrated firms
Figure 7: Financing the deficit across Z-score and asset volatility quintiles –
unrated firms
Controlling for the probability of going bankrupt, unrated firms in higher asset
volatility groups generally issue less debt to finance their deficit. Note that firms with
the lowest Z-score stand out since they use relatively less debt than other firms. This
is similar to our finding that the smallest firms and those with the higher market-tobook ratio also appear reluctant to issue debt.
In contrast, none of the Z-score quintiles of rated firms exhibits the monotonic
negative pattern of the regression coefficient b across risk groups.
- 25 -
Table 15: Financing the deficit across Z-score and asset volatility quintiles –
rated firms
Figure 8: Financing the deficit across Z-score and asset volatility quintiles –
rated firms
Finally, we examine whether the sample period matters for our results. Table 16 and
Figure 9 show the results of running (3) across risk deciles in each decade separately.
Table 16: Financing the deficit across deciles in the 70s, 80s and 90s
Figure 9: Financing the deficit across deciles in the 70s, 80s and 90s
The monotone negative pattern of the coefficient on the financing deficit across risk
deciles is present in all decades. It grows stronger as we move from the 70s to 80s,
and from the 80s to the 90s.
5. Conclusion
One of the most important questions in corporate finance is what securities do firms
issue to finance their investments. Myers (1984) provides a particularly influential
treatment of this, potentially complex, issue. He argues that issuing securities is
subject to an adverse selection or “lemons” problem since rational investors anticipate
that they know less than the firm about the investments they are being asked to
finance. To protect themselves, rational investors may price securities at a discount.
Firms can avoid the costly discount by using securities that are robust to the lemons
problem.
Myers’ intuition is usually taken to imply a pecking order where debt is issued prior to
equity. Yet, theory has repeatedly pointed out that the pecking order is a special case
that applies only if risk play no role (see for example Nachman and Noe (1994),
- 26 -
Daniel and Titman (1995) or Stein (2003)). The intuition is that debt is a concave
claim that is mispriced by investors who are uninformed about risk.
26
Since the
pecking order is a special case that applies only when debt has no adverse selection
cost, it is then not surprising that one cannot find general robust support for the
pecking order (as in Fama and French (2002) or Frank and Goyal (2003)).
This paper therefore asks: is the adverse selection cost of debt empirically relevant?
And for what kind of firms does it matter? And how does this cost to other costs of
debt such as bankruptcy or a debt overhang?
Using a large unbalanced panel of publicly traded US firms from 1971 to 2001 we
find very robust and economically significant evidence that i) firms appear to avoid
issuing debt if the outside market knows little about firms’ risk, ii) the adverse
selection cost of debt is irrelevant for firms that have any rating, and vice versa,
suggesting that ratings appear to bridge the information gap between firms and
outside investors about risk, and iii) the adverse selection cost complements other
costs of debt, for example firms with the highest market-to-book ratios or with the
highest probability of default issue considerable less debt.
And interesting perspective on our results, and one to warrants further research in our
view, is that the adverse selection cost of debt relates to the supply of debt by the
outside market while the bankruptcy cost relates to firms demand of debt. It is would
be interesting to integrate demand and supply into a comprehensive model of firms’
capital structure decisions.
Our paper also relates to recent efforts to develop and analyze dynamic capital
structure models (Fischer et al. (1989), Strebulaev (2004) and Hennessy and Whited
(2004)). The literature on dynamic capital structure decisions usually assumes
26
Outside the capital structure literature, it is standard to consider the adverse selection cost of debt.
For example, Stiglitz and Weiss (1981) build their credit rationing argument on it.
- 27 -
exogenous frictions when firms issue securities. Our evidence suggests that
information asymmetries could be an important determinant of these frictions.
- 28 -
References
Akerlof, G., 1970. The market for lemons. Quarterly Journal of Economics, 84, 488500.
Baker, M. and J. Wurgler, 2002. Market timing and capital structure, Journal of
Finance 57, 1-32.
Barclay, M.J. and C.W. Smith, 1999. The capital structure puzzle: Another look at the
evidence. Journal of Applied Corporate Finance, 12, 8-20.
Bradley, M. G.A. Jarrell and E.H. Kim, 1984. On the existence of an optimal capital
structure: theory and evidence. Journal of Finance, 39, 857-877.
Brealey, R. and S. Myers, 2000. Principles of Corporate Finance, 6th edition.
McGraw-Hill.
Brennan, M. and A. Kraus, 1987, Efficient financing under asymmetric information,
Journal of Finance, 42, 1225-1243.
Cadsby, C.B., Frank, M. and V. Maksimovic, 1990. Pooling, separating and semiseparating equilibria in financial markets: Some experimental evidence. Review of
Financial Studies. 3, 315-342
Chang, X. and S. Dasgupta, 2003. Financing the deficit: Debt capacity, information
asymmetry, and the debt-equity choice, Working paper, Hong Kong University of
Science and Technology.
- 29 -
Chirinko, R.S. and A.R. Singha, 2000. Testing static tradeoff against pecking order
models of capital structure: a critical comment. Journal of Financial Economics,
58, 417-425.
Daniel, K. and S. Titman, 1995. Financing investment under asymmetric information.
In Jarrow, R., Maksimovic, V. and Ziemba, W. (eds), Handbook in Operations
Research and Management Science. Finance (Vol. 9), ch. 23. Elsevier.
Fama, E.F., and K.R. French, 2002. Testing tradeoff and pecking order predictions
about dividends and debt. Review of Financial Studies, 15, 1-33.
Faulkender, M. and M. Petersen, 2005. Does the source of capital affect capital
structure?, Review of Financial Studies, forthcoming.
Fischer, E.O., R. Heinkel and J. Zechner, 1989. Dynamic capital structure choice:
Theory and tests, Journal of Finance, 44, 19-40.
Flannery, M.J. and K. Rangan, 2003. Partial adjustment towards target capital
structures, Journal of Financial Economics, forthcoming.
Frank, M.Z. and V.K. Goyal, 2003. Testing pecking order theory of capital structure.
Journal of Financial Economics, 67, 217-248.
Frank, M.Z. and V.K. Goyal, 2005.Trade-off and Pecking Order theories of debt.
Unpublished working paper, University of British Columbia and Hong Kong
University of Science and Technology (to appear in E. Eckbo (ed), Handbook of
Corporate Finance: Empirical Corporate Finance, Elsevier, 2005.)
Graham, J.R., and C.R. Harvey, 2001. The theory and practice of corporate finance:
Evidence from the field. Journal of Financial Economics, 60, 187-243.
- 30 -
Harris, M. and A. Raviv, 1991. The theory of capital structure. Journal of Finance,
46, 297-356.
Helwege, J. and N. Liang 1996. Is there a pecking order? Evidence from a panel of
IPO firms. Journal of Financial Economics, 40, 429-458.
Hennessy, C.A. and T.M. Whited, 2004. Debt dynamics, Unpublished working paper,
University of California at Berkeley and University of Wisconsin, Madison.
Hovakimian, A, T. Opler and S. Titman, 2001. The debt-equity choice. Journal of
Financial and Quantitative Analysis, 36, 1-24
Hovakimian, A., G. Hovakimian and H. Theranian, 2004. Determinanats of target
capital structure: the case of dual debt and equity issues. Journal of Financial
Economics, 71.
Kayhan, A. and S. Titman, 2003. Firms’ histories and their capital structures,
Unpublished working paper, University of Texas (Austin).
Leary, M.T and M.R. Roberts, 2004a. Financial slack and tests of the Pecking Order’s
financing hierarchy, Unpublished working paper, Duke University.
Leary, M.T and M.R. Roberts, 2004b. Do firms rebalance their capital structures.
Journal of Finance, forthcoming.
Lemmon, M.L. and J. Zender, 2002. Debt capacity and test of capital structure
theories, Unpublished working paper. University of Colorado and University of
Utah.
- 31 -
MacKie-Mason, J.K., 1990. Do firms care who provides their financing? Journal of
Finance, 45, 1471-1495.
Mayer, C. and O. Sussman, 2002. A new test of capital structure. Unpublished
working paper. University of Oxford.
Merton, R. C., 1974. On the pricing of corporate debt: The risk structure of interest
rates. Journal of Finance 29, 449-470.
Myers, S.C., 1977. Determinants of corporate borrowing, Journal of Financial
Economics, 5, 147-175.
Myers, S.C., 1984. The capital structure puzzle. Journal of Finance, 39, 575-592.
Myers, S.C. and N.S. Majluf, 1984. Corporate financing and investment decisions
when firms have information that investors do not have. Journal of Financial
Economics, 13, 187-221.
Nachman, D. and T.H. Noe, 1994. Optimal design of securities under asymmetric
information. Review of Financial Studies, 7, 1-44.
Rajan, R.G. and L. Zingales, 1995. What do we know about capital structure? Some
evidence from international data. Journal of Finance, 50, 1421-1460.
Shyam-Sunder, L. and S.C. Myers, 1999. Testing static tradeoff against pecking order
models of capital structure, . Journal of Financial Economics, 51, 219-244.
Stein, J.C. 2003. Agency, Information and Corporate Investment. In: Handbook of the
Economics of Finance, edited by G. Constantinides, Milt Harris and Rene Stulz,
Elsevier, 111-165.
- 32 -
Stiglitz, J. and Weiss, 1981. Credit Rationing in markets with imperfect information.
American Economic Review, 71, 393-410.
Strebulaev, I.A., 2004. Do tests of capital structure theory mean what they say?,
Unpublished working paper, London Business School.
Titman, S. and R. Wessels, 1988. The determinants of capital structure choice.
Journal of Finance, 43, 1-19.
Welch, I, 2004, Stock returns and capital structure. Journal of Political Economy,
112, 106-131.
- 33 -
Appendix
Using a Merton model to compute asset volatility
From Ito’s lemma, we have
 E  V
Vt Et
Et Vt
where  E is the instantaneous variance of the rate of return on equity (the standard
deviation of daily stock returns from CRSP),  V is the instantaneous variance of the
rate of return on the firm (to be solved for), Vt is the market value of the firm and Et is
the market value of equity (both calculated as described below). The derivative of the
market value of equity with respect to the market value of the firm in the Merton
model is:
 ln(Vt / Bt )  (rf  12  V2 )T 
Et


Vt
V T


where  is the cumulative distribution function of the standardized normal
distribution N(0,1), T is the time to maturity of the debt (we try both 10 and 20 years)
and rf is the risk free rate (from Kenneth French’s website).
Variable definitions
Investments: For firms reporting under formats 1 to 3, it equals Compustat item #128
+ #113 + #129 + #219 - #107 - #109. For firms reporting under format 7, investments
equal #128 + #113 + #129 - #107 - #109 - #309 - #310.
Change in net working capital: For firms reporting under format 1, it equals
Compustat item #274 - #236 - #301. For firms reporting under format 2and 3, it
equals #274 + #236 - #301, and for firms reporting under format 7, it equals - #302 #303 - #304 - #305 - #307 + #274 - #312 - #301.
- 34 -
Internal cash flows: For firms reporting under formats 1 to 3, it equals Compustat
item #123 + #124 + #125 + #126 + #106 + #213 + #217 + #218. For firms reporting
under format 7, internal cash flows equal #123 + #124 + #125 + #126 + #106 + #213
+ #217 + #314.
Market value of a firm: Book value of debt = #181 + #10 (or #56 or #130 depending
on availability and in that order) + market value of equity = number of common
shares outstanding times the closing share price (from CRSP)
Variables that are trimmed
In order to remove outliers and misrecorded data, observations that are in the extreme
0.5 % left or right tail of the distribution or have missing values are removed. This
trimming has been applied to the following variables: current assets (Compustat item
#4), current liabilities (#5), cash dividends (#127), investments (defined above),
internal cash flows (defined above), change in net working capital (defined above),
financial deficit, net debt issued (#111-#114), net equity issued (#108-#115), all as a
percentage of total assets, as well as tangibility (#8/#6), market-to-book ratio,
profitability (#13/#6), and log(sales) (natural logarithm of #12).
Calculating the variation in firm specific implied volatility from option prices
We use end-of-day option prices, option open interest and implied volatility estimates
from the Ivy DB database provided by OptionMetrics. The sample is from January
1996 to December 2001. To filter out misrecorded data and very illiquid contracts, we
exclude days/contracts that have zero open interest or have a bid-ask spread larger
than 50% of the option price at midpoint. The sample includes at-the-money call
options with maturity closest to but higher than 182 days. Ideally, we would use
longer maturity contracts, e.g. LEAPS, but these contracts are quite illiquid so that our
- 35 -
sample would be considerably reduced. The sample has 13,418,700 day-firm implied
volatility estimates. We then calculated the standard deviation of the implied volatility
per firm over the previous calendar year. If there are less than 90 trading days the
firm-year is excluded from the sample.
- 36 -
Table 1
Balance sheets, cash flows and other descriptive statistics over time
The table reports average balance sheets for the sample. Financial firms, utilities and companies that could not be
matched properly with CRSP are excluded. Unless labeled as median, each item in Panel A and Panel B is
calculated as a percentage of the book value of total assets and then averaged across all firms of our sample in that
year. Definitions of variables follow Frank and Goyal (2003) and Fama and French (2002). See text and appendix
for details.
Year
Number of observations
1971
1518
Panel A: Balance sheet items
Assets:
+Cash (#162)
0.040
+Short term investments (#193)
0.035
+Receivables-total (#2)
0.194
+Inventories (#3)
0.247
+Current assets-other (#68)
0.014
0.539
+Current assets-total (#4)
+Net property plant and equipment (#8)
0.356
+Investments and advances - equity method (#31)
0.020
+Investments and advances - other (#32)
0.025
+Intangibles (#33)
0.036
+Assets - other (#69)
0.024
1.000
=Total assets (#6)
Liabilities
+Debt in current liabilities (#34)
0.068
+Account payable (#70)
0.090
+Income taxes payable (#71)
0.020
+Current liabilities - other (#72)
0.061
0.239
=Current liabilities - total (#5)
+Long-term debt - total (#9)
0.199
+Liabilities - other (#75)
0.012
+Deferred taxes and ITC (#35)
0.020
+Minority interest (#38)
0.005
0.476
=Liabilities - total (#181)
+Preferred stock - carrying value (#130)
0.011
+Common equity - total (#60)
0.513
=Stockholders' equity - total (#216)=(#130)+(#60)
0.524
1.000
=Total liabilities and stockholders' equity
Panel B: Corporate cash flows
+Cash Dividends (#127)
0.018
+Change in net working capital
0.022
-Internal cash flow
0.099
+Investments
0.082
=Financial deficit (Mean)
0.023
Financial deficit (Median)
0.001
Net debt issues (#111-#114) Mean
0.012
Net debt issues (Median)
0.000
Net equity issues (#108-#115) (Mean)
0.011
Net equity issues (Median)
0.000
Panel C: Other descriptive statistics
Age (years since first appearance in CRSP)
7
Market value of assets (in millions of dollars)
503.233
Book value of assets (#6) (in millions of dollars)
436.892
Tangibility (#8/#6)
0.356
Log sales (log(#12))
4.73
Market-to-book ratio
1.52
Profitability=Operating income(#13) / Assets(/#6)
0.128
- 37 -
1980
2925
1990
3481
2001
3810
0.030
0.045
0.217
0.245
0.020
0.575
0.349
0.014
0.026
0.020
0.023
1.000
0.085
0.031
0.205
0.186
0.029
0.544
0.320
0.010
0.025
0.049
0.054
1.000
0.127
0.056
0.154
0.126
0.037
0.501
0.276
0.010
0.020
0.128
0.064
1.000
0.066
0.114
0.018
0.087
0.286
0.200
0.015
0.026
0.003
0.529
0.009
0.461
0.471
1.000
0.094
0.111
0.008
0.097
0.312
0.192
0.034
0.020
0.006
0.564
0.015
0.422
0.437
1.000
0.063
0.086
0.006
0.118
0.274
0.184
0.045
0.016
0.005
0.524
0.021
0.456
0.476
1.000
0.015
0.024
0.106
0.102
0.034
0.003
0.017
0.000
0.017
0.000
0.009
-0.011
0.044
0.071
0.025
-0.001
0.004
-0.001
0.021
0.000
0.005
-0.022
0.000
0.058
0.041
0.002
0.001
0.000
0.040
0.001
11
464.232
514.434
0.349
4.74
1.40
0.144
12
966.102
858.079
0.320
4.45
1.54
0.065
13
2943.950
1550.136
0.276
5.25
1.90
0.014
Table2
Balance sheets, cash flows and other descriptive statistics across deciles
The table reports average balance sheets, cash flow items and other descriptive statistics for each asset volatility decile. Firms are ranked in deciles according to the daily standard
deviation of the return on market value of assets (book value of debt + market value of equity) in the previous calendar year. Rank 10 firms have highest standard deviation. Unless
labeled as median, each item is calculated as a percentage of the book value of total assets and then averaged across all firms in a decile. Definitions of variables follow Frank and
Goyal (2003) and Fama and French (2002). See text and appendix for details. Z-score equals 3.3*(#170, pretax income)+(#12, sales)+1.4*(#36, retained earnings)+1.2*[(#4,
current assets)-(#5, current liabilities)]/(#6, assets) (see MacKie-Mason (1990)).
Decile
Number of observations
1 (Low)
10348
2
10331
3
10340
4
10332
5
10336
6
10335
7
10338
8
10334
9
10337
10 (High)
10320
Panel A: Balance sheet items
Assets:
+Cash (#162)
+Short term investments (#193)
+Receivables-total (#2)
+Inventories (#3)
+Current assets-other (#68)
+Current assets-total (#4)
+Net property plant and equipment (#8)
+Investments and advances - equity method (#31)
+Investments and advances - other (#32)
+Intangibles (#33)
+Assets - other (#69)
=Total assets (#6)
Liabilities
+Debt in current liabilities (#34)
+Account payable (#70)
+Income taxes payable (#71)
+Current liabilities - other (#72)
=Current liabilities - total (#5)
+Long-term debt - total (#9)
+Liabilities - other (#75)
+Defered taxes and ITC (#35)
+Minority interest (#38)
=Liabilities - total (#181)
+Prefered stock - carrying value (#130)
+Common equity - total (#60)
=Stockholders' equity - total (#216)=(#130)+(#60)
=Total liabilities and stockholders' equity
0.039
0.024
0.182
0.191
0.025
0.474
0.369
0.020
0.040
0.052
0.050
1.000
0.034
0.021
0.189
0.205
0.027
0.483
0.367
0.017
0.024
0.061
0.048
1.000
0.039
0.024
0.195
0.210
0.028
0.505
0.356
0.016
0.021
0.059
0.045
1.000
0.045
0.027
0.197
0.208
0.028
0.515
0.351
0.014
0.021
0.057
0.044
1.000
0.051
0.034
0.201
0.205
0.028
0.532
0.340
0.012
0.021
0.055
0.043
1.000
0.064
0.042
0.204
0.202
0.028
0.551
0.322
0.012
0.022
0.054
0.042
1.000
0.076
0.055
0.210
0.195
0.029
0.578
0.301
0.010
0.022
0.050
0.041
1.000
0.091
0.070
0.208
0.188
0.028
0.602
0.281
0.009
0.024
0.043
0.043
1.000
0.107
0.082
0.203
0.176
0.028
0.614
0.267
0.008
0.026
0.043
0.044
1.000
0.124
0.082
0.184
0.157
0.026
0.592
0.268
0.011
0.030
0.050
0.051
1.000
0.098
0.119
0.010
0.090
0.321
0.304
0.068
0.024
0.010
0.724
0.017
0.259
0.276
1.000
0.077
0.108
0.011
0.092
0.290
0.270
0.047
0.030
0.006
0.641
0.013
0.345
0.359
1.000
0.069
0.104
0.012
0.093
0.280
0.239
0.037
0.029
0.005
0.588
0.011
0.401
0.412
1.000
0.065
0.102
0.013
0.093
0.274
0.217
0.029
0.027
0.005
0.550
0.008
0.441
0.450
1.000
0.065
0.101
0.013
0.095
0.275
0.193
0.024
0.025
0.004
0.519
0.009
0.471
0.481
1.000
0.064
0.102
0.013
0.095
0.275
0.172
0.020
0.021
0.004
0.492
0.010
0.498
0.508
1.000
0.067
0.100
0.013
0.096
0.277
0.151
0.017
0.018
0.003
0.466
0.011
0.524
0.534
1.000
0.065
0.096
0.012
0.095
0.269
0.131
0.015
0.016
0.003
0.433
0.012
0.554
0.567
1.000
0.064
0.098
0.011
0.097
0.270
0.112
0.014
0.013
0.003
0.411
0.016
0.573
0.589
1.000
0.074
0.106
0.008
0.096
0.285
0.098
0.014
0.008
0.004
0.410
0.017
0.574
0.591
1.000
- 38 -
Decile
Number of observations
1 (Low)
10348
2
10331
0.009
0.066
0.004
0.075
0.005
-0.001
0.000
-0.002
0.005
0.000
0.012
0.074
0.008
0.087
0.008
-0.002
0.005
-0.001
0.002
0.000
13.7
15.3
Market value of assets (in millions of dollars)
2287.082
2206.197
1896.059
1523.325
Book value of assets (#6) (in millions of dollars)
Tangibility (#8/#6)
Log sales (log(#12))
2468.440
0.369
6.096
1726.506
0.367
5.988
1273.871
0.356
5.797
Market-to-book ratio
Profitability=Operating income(#13)/Assets(/#6)
Median modified Z-score
Median asset STD in t-1
STD of asset STD in t-1
STD of implied volatility from option prices
1.127
0.103
1.797
0.004
0.002
0.053
1.160
0.119
2.126
0.008
0.002
0.048
1.256
0.127
2.291
0.010
0.003
0.052
+Cash Dividends (#127)
+Investments
+Change in working capital
-Internal cash flow
=Financial deficit (Mean)
Financial deficit (Median)
Net debt issues (#111-#114) (Mean)
Net debt issues - Median
Net equity issues (#108-#115) - Mean
Net equity issues - Median
Age (years since first appearance in CRSP)
- 39 -
3
10340
4
10332
5
10336
6
10335
7
10338
8
10334
9
10337
10 (High)
10320
0.013
0.094
0.015
0.099
0.022
0.000
0.013
0.000
0.010
0.000
0.011
0.097
0.019
0.095
0.030
0.001
0.014
0.000
0.017
0.000
0.009
0.098
0.020
0.085
0.042
0.003
0.014
0.000
0.028
0.001
0.007
0.096
0.019
0.063
0.060
0.006
0.016
0.000
0.044
0.002
0.006
0.093
0.004
0.019
0.085
0.010
0.017
0.000
0.068
0.003
0.004
0.086
-0.035
-0.070
0.125
0.014
0.018
0.000
0.107
0.004
Panel C: Other descriptive statistics
14.7
13.5
12.1
10.7
9.4
8.3
7.2
6.6
1307.745
877.455
588.056
400.242
210.674
144.904
883.251
0.351
5.466
636.009
0.340
5.130
430.103
0.322
4.726
257.375
0.301
4.305
176.327
0.281
3.812
90.361
0.267
3.181
62.547
0.268
2.169
1.343
0.132
2.369
0.013
0.003
0.058
1.447
0.132
2.402
0.015
0.005
0.060
1.582
0.127
2.374
0.018
0.006
0.070
1.750
0.112
2.278
0.022
0.008
0.072
1.964
0.083
2.109
0.027
0.010
0.084
2.213
0.027
1.712
0.034
0.012
0.094
2.694
-0.088
0.658
0.052
0.126
0.117
Panel B: Corporate cash flows
0.013
0.014
0.081
0.088
0.011
0.013
0.093
0.097
0.012
0.017
-0.001
0.000
0.009
0.011
-0.001
0.000
0.003
0.005
0.000
0.000
Table 3
Financing the deficit across deciles
Pooled panel OLS regressions of net debt issues D and net equity issues ∆E on the financing deficit DEF are estimated for each decile n=1,…10: Dit  a  bnD DEFit   it ,
Eit  a  bnE DEFit   it . Ranking based on the daily standard deviation of the return on market value of assets during the previous calendar year. Firms with rank 10 have highest
standard deviation. Standard errors are reported below the coefficients, in italics. All coefficients on financial deficit are significant at the 1 % level.
1 (Low)
-0.004
0.000
2
-0.001
0.000
Panel A: Dependent variable - Net debt issued
3
4
5
6
-0.001
-0.001
-0.003
-0.004
0.000
0.000
0.001
0.001
Financial deficit
0.868
0.004
0.822
0.004
0.807
0.004
0.764
0.005
0.708
0.005
0.570
0.005
0.457
0.005
0.326
0.005
0.230
0.004
0.147
0.004
Adjusted R squared
0.849
0.802
0.787
0.728
0.665
0.542
0.419
0.293
0.209
0.129
1 (Low)
0.004
0.000
2
0.001
0.000
Panel B: Dependent variable - Net equity issued
3
4
5
6
0.001
0.001
0.003
0.004
0.000
0.000
0.001
0.001
7
0.005
0.001
8
0.003
0.001
9
0.003
0.001
10 (High)
0.001
0.001
Financial deficit
0.126
0.004
0.175
0.004
0.192
0.004
0.235
0.005
0.291
0.005
0.430
0.005
0.542
0.005
0.673
0.005
0.770
0.004
0.853
0.004
Adjusted R squared
0.109
0.157
0.173
0.203
0.251
0.402
0.504
0.638
0.747
0.832
Decile
Intercept
Decile
Intercept
- 40 -
7
-0.005
0.001
8
-0.003
0.001
9
-0.002
0.001
10 (High)
-0.001
0.001
Figure 1
Financing the deficit across deciles
Pooled panel OLS regressions of net debt issues D and net equity issues ∆E on the financing deficit DEF are estimated for each decile n=1,…10: Dit  a  bnD DEFit   it ,
Eit  a  bnE DEFit   it .The figure plots coefficients on financial deficit and adjusted R-squared for each decile.
1
0.9
0.8
0.7
Net debt issued: coefficient on
financial deficit
0.6
Net debt issued: adj. R-squared
0.5
Net equity issued: coefficient on
financial deficit
0.4
Net equity issued: adj. Rsquared
0.3
0.2
0.1
0
1
2
3
4
5
6
7
Asset volatility decile
- 41 -
8
9
10
Table 4
Issue decisions across deciles
The table reports data on net issues of debt/equity larger than 1% of total assets (significant outside financing). The table reports proportion of firms in each decile (in %) that
follows a particular financing pattern. Ranking based on the daily standard deviation of market value of assets during the previous calendar year.
Do nothing
Issue debt only
Issue equity only
1 (Low)
20.6
27.39
2.01
2
19.06
26.57
1.67
3
20.55
25.86
2.27
- 42 -
4
20.95
25.54
3.28
5
22.09
25.22
4.2
6
23.48
23.01
6.46
7
22.76
19.7
10.62
8
24.09
18.3
13.84
9
23.63
16.24
16.88
10 (High)
26.11
13.75
19.15
Table 5
Financing the deficit for rated and unrated firms across deciles
Firms are split into 2 subsamples depending on availability of S&P issuer credit rating data. Pooled panel OLS regressions of net debt issues D on the financing deficit DEF are
estimated for each decile in each subsample: Dit  a  bnD DEFit   it . Ranking is done for the whole sample based on the daily standard deviation of the return on market value of
firms assets during the previous calendar year. Firms with rank 10 have highest standard deviation. Standard errors are reported below the coefficients, in italics. All coefficients
on financial deficit are significant at the 1 % level.
Firms with S&P issuer credit rating data
Decile
2
3
-0.004
0.001
0.004
0.001
0.006
0.001
Financial deficit
0.877
0.006
0.838
0.007
Adj. R squared
0.873
Number of Observations
2822
Intercept
1 (Low)
4
5
6
7
8
9
10 (High)
0.005
0.001
0.003
0.002
0.001
0.002
-0.001
0.004
-0.007
0.007
-0.009
0.015
-0.014
0.016
0.807
0.008
0.794
0.010
0.735
0.013
0.761
0.014
0.800
0.018
0.774
0.030
0.695
0.049
0.577
0.057
0.823
0.802
0.787
0.738
0.786
0.812
0.719
0.652
0.617
2943
2380
1783
1186
789
446
253
106
65
Firms without S&P issuer credit rating data
Decile
2
3
4
5
6
7
8
9
-0.005
0.000
-0.003
0.000
-0.003
0.000
-0.003
0.001
-0.004
0.001
-0.004
0.001
-0.005
0.001
-0.004
0.001
-0.003
0.001
-0.001
0.001
Financial deficit
0.865
0.004
0.813
0.005
0.804
0.005
0.751
0.005
0.700
0.005
0.544
0.005
0.427
0.005
0.306
0.005
0.221
0.004
0.145
0.004
Adj. R squared
0.841
0.794
0.781
0.709
0.648
0.510
0.387
0.273
0.201
0.126
Number of Observations
7526
7387
7960
8548
9150
9545
9892
10081
10228
10253
Intercept
1 (Low)
- 43 -
10 (High)
Figure 2
Financing the deficit for rated and unrated firms across deciles
Firms are split into 2 subsamples depending on availability of S&P issuer credit rating data. Pooled panel OLS regressions of net debt issues D on the financing deficit DEF are
estimated for each decile in each subsample: Dit  a  bnD DEFit   it . Ranking is done for the whole sample based on the daily standard deviation of the return on market value of
firms assets during the previous calendar year. Firms with rank 10 have highest standard deviation. The figure plots coefficients on financial deficit for each group and for each
decile.
1.0000
0.9000
0.8000
0.7000
0.6000
rated
0.5000
unrated
0.4000
0.3000
0.2000
0.1000
0.0000
1
2
3
4
5
6
Decile
- 44 -
7
8
9
10
Table 6
Financing the deficit for rated and unrated firms across deciles
Firms are split into 2 subsamples depending on availability of S&P issuer credit rating data. The following pooled panel OLS are estimated for both subsamples:
Dit  a  bDEFit  b RISK DEFit * LNRISK it  b SIZE DEFit * LNSIZEit  bTANG DEFit * TANGit   it . ΔD is net debt issuance, DEF is the financing deficit, LNRISK is the natural
logarithm of last year’s asset volatility, LNSIZE is the natural logarithm of the book value of assets and TANG is tangible assets divided by the book value of assets. Standard
errors are reported below the coefficients, in italics.
Decile
Rated firms
Unrated firms
Intercept
0.002
0.000
0.005
0.000
DEF
0.606
0.023
-0.703
0.006
DEF*LNRISK
-0.089
0.004
-0.247
0.002
DEF*LNSIZE
-0.031
0.003
0.031
0.000
DEF*TANG
0.047
0.014
0.420
0.006
Adj. R squared
0.807
0.5203
Number of Observations
12772
90567
- 45 -
Table 7
Financing the deficit across deciles: Fama-McBeth procedure and fixed effects
Firms are ranked into deciles according to daily standard deviation of the return on market value of assets in the previous calendar year. The regression Dit  a  bnD DEFit   it ,
is estimated for each decile/year combination. The table reports in panel A, for each decile, time-series means of cross sectional regression intercepts, slopes and the t-statistic
using the time-series standard errors (in italics). Panel B and panel C report the coefficient on the financing deficit, and the t-statistic in italics, using fixed year and fixed firm
effects respectively. All coefficients on financial deficit are significant at the 1 % level.
Decile
Intercept
Financial deficit
Decile
Financial deficit
Decile
Financial deficit
Panel A: Fama-McBeth procedure
4
5
6
1 (Low)
2
3
-0.004
-0.002
-0.001
-0.002
-0.003
-6.575
-1.576
-1.129
-1.870
-3.539
0.872
0.838
0.821
0.792
72.570
56.658
51.779
53.221
1 (Low)
2
7
8
9
10 (High)
-0.004
-0.005
-0.005
-0.005
-0.004
-4.089
-5.214
-5.407
-4.827
-3.277
0.759
0.668
0.590
0.522
0.423
0.307
32.862
19.494
14.940
10.611
8.537
6.962
6
7
8
9
10 (High)
Panel B: Year fixed effect
4
5
3
0.867
0.819
0.805
0.764
0.708
0.571
0.464
0.338
0.242
0.157
243.10
205.03
195.63
166.61
143.28
110.54
87.36
67.40
54.38
40.74
1 (Low)
2
3
7
8
9
10 (High)
Panel C: Firm fixed effect
4
5
6
0.885
0.859
0.840
0.805
0.795
0.670
0.533
0.380
0.274
0.178
218.99
183.90
164.54
134.47
130.56
92.39
69.87
48.57
41.02
31.96
- 46 -
Table 8
Regression of net debt issues on conventional variables and financing deficit across deciles.
The regression Dit  an  bn DEFit  bn TANG TANGit  bn MTB MTBit  bn PROF PROFit  bn LOGSALES LOGSALESit   is estimated for each decile. ∆D is net debt issued. Tangibility
is defined as property, plant & equipment over total assets. Market-to-book is defined as in Fama and French (2002). LogSales is the natural logarithm of net sales. Profitability is
operating income before depreciation over total value of assets. Firms are ranked into deciles according to daily standard deviation of the return on market value of assets in the
previous calendar year. OLS standard errors reported below the coefficients.
Decile
Intercept
1 (Low)
-0.004
0.000
2
0.000
0.000
3
0.000
0.000
4
-0.001
0.000
5
-0.004
0.001
6
-0.005
0.001
7
-0.006
0.001
8
-0.006
0.001
9
-0.005
0.001
10 (High)
-0.004
0.001
∆ Tangibility
0.006
0.006
0.037
0.007
0.024
0.008
0.037
0.009
0.056
0.010
0.068
0.010
0.129
0.011
0.101
0.011
0.142
0.011
0.102
0.010
∆ Market-to-Book
-0.012
0.001
-0.014
0.001
-0.012
0.001
-0.012
0.001
-0.009
0.001
-0.006
0.001
-0.003
0.001
-0.003
0.001
-0.001
0.000
-0.002
0.000
∆ Logsales
0.000
0.001
-0.001
0.002
-0.004
0.002
-0.001
0.002
0.004
0.002
0.003
0.002
0.010
0.002
0.010
0.002
0.009
0.002
0.015
0.002
∆ Profitability
-0.010
0.006
-0.017
0.007
-0.020
0.007
-0.022
0.006
-0.045
0.006
-0.031
0.006
-0.024
0.006
-0.040
0.006
-0.015
0.005
-0.003
0.004
Financial deficit
0.866
0.004
0.833
0.004
0.805
0.004
0.761
0.005
0.706
0.005
0.574
0.005
0.452
0.006
0.328
0.005
0.236
0.005
0.151
0.004
Adj. R-squared
0.851
0.814
0.789
0.733
0.678
0.556
0.430
0.312
0.231
0.155
Number of Observations
9893
9996
10046
10023
10043
10032
10040
9959
9869
9559
- 47 -
Table 9
Financing the deficit across size and asset volatility quintiles
The regression Dit  a  bnD DEFit   it is estimated for each size/asset volatility group. The table reports coefficients of
the financial deficit. Firms are sorted in quintiles according to book assets, and then within each size quintile, firms are
ranked in 5 groups based on daily standard deviation of the return on market value of assets during the previous calendar
year. OLS standard errors reported below the coefficients in italics.
Asset volatility quintile
Size quintile 1 (Small)
1 (Low)
0.505
0.008
2
0.309
0.008
3
0.235
0.007
4
0.143
0.006
5 (High)
0.127
0.005
Size quintile 2
0.836
0.007
0.624
0.008
0.456
0.008
0.291
0.008
0.189
0.007
Size quintile 3
0.866
0.006
0.771
0.008
0.676
0.008
0.490
0.008
0.265
0.008
Size quintile 4
0.873
0.006
0.821
0.007
0.798
0.007
0.703
0.007
0.519
0.008
Size quintile 5 (Big)
0.839
0.006
0.823
0.006
0.788
0.006
0.750
0.007
0.713
0.007
Figure 3
Financing the deficit across size and asset volatility quintile
The regression Dit  a  bnD DEFit   it is estimated for each size/ asset volatility group. Firms are sorted in quintiles
according to book assets, and then within each size quintile, firms are ranked in 5 groups based on daily standard
deviation of the return on market value of assets during the previous calendar year. The figure plots coefficients on
financial deficit for the size quintiles.
1
0.9
0.8
0.7
Size quintile 1 (Small)
0.6
Size quintile 2
0.5
Size quintile 3
Size quintile 4
0.4
Size quintile 5 (Big)
0.3
0.2
0.1
0
1
2
3
4
Asset volatility quintile
- 48 -
5
Table 10
Financing the deficit across age and asset volatility quintiles
The regression Dit  a  bnD DEFit   it is estimated for each age/ asset volatility group. The table reports coefficients of
the financial deficit. Firms are sorted in quintiles according to age (years since it first appeared in CRSP), and then within
each age quintile, firms are ranked in 5 groups based on daily standard deviation of the return on market value of assets
during the previous calendar year. OLS standard errors reported below the coefficients in italics.
Asset volatility quintile
1 (Low)
2
3
4
5 (High)
Age quintile 1 (Young)
0.771
0.615
0.374
0.250
0.155
0.006
0.007
0.007
0.006
0.005
Age quintile 2
0.841
0.006
0.705
0.009
0.531
0.009
0.292
0.008
0.157
0.007
Age quintile 3
0.856
0.007
0.785
0.008
0.607
0.008
0.397
0.009
0.180
0.007
Age quintile 4
0.879
0.005
0.795
0.006
0.703
0.007
0.521
0.008
0.277
0.007
Age quintile 5 (Old)
0.889
0.007
0.844
0.008
0.795
0.009
0.760
0.010
0.504
0.010
Figure 4
Financing the deficit across age and asset volatility quintile
The regression Dit  a  bnD DEFit   it is estimated for each age/ asset volatility group. Firms are sorted in quintiles
according to age (years since it first appeared in CRSP), and then within each age quintile, firms are ranked in 5 groups
based on daily standard deviation of the return on market value of assets during the previous calendar year. The figure
plots coefficients on financial deficit for the age quintiles.
1.000
0.900
0.800
0.700
Age quintile 1 (Young)
0.600
Age quintile 2
0.500
Age quintile 3
Age quintile 4
0.400
Age quintile 5 (Old)
0.300
0.200
0.100
0.000
1
2
3
4
Asset volatility quintile
- 49 -
5
Table 11
Financing the deficit across tangibility and asset volatility quintiles
The regression Dit  a  bnD DEFit   it is estimated for each tangibility/ asset volatility group. The table reports
coefficients of the financial deficit. Firms are sorted in quintiles according to tangibility (Compustat item8/Compustat
item6), and then within each tangibility quintile, firms are ranked in 5 groups based on daily standard deviation of the
return on market value of assets during the previous calendar year. OLS standard errors reported below the coefficients in
italics.
Asset volatility quintile
1 (Low)
2
3
4
5 (High)
Tangibility quintile 1 (Low)
0.764
0.568
0.271
0.156
0.110
0.007
0.008
0.007
0.006
0.005
Tangibility quintile 2
0.844
0.006
0.743
0.008
0.428
0.008
0.294
0.007
0.132
0.006
Tangibility quintile 3
0.830
0.006
0.780
0.007
0.691
0.008
0.458
0.008
0.187
0.006
Tangibility quintile 4
0.855
0.006
0.817
0.006
0.754
0.007
0.567
0.008
0.278
0.007
Tangibility quintile 5 (High)
0.866
0.006
0.849
0.006
0.790
0.007
0.685
0.008
0.445
0.008
Figure 5
Financing the deficit across tangibility and asset volatility quintile
The regression Dit  a  bnD DEFit   it is estimated for each size/asset volatility group. Firms are sorted in quintiles
according to tangibility (item8/item6), and then within each tangibility quintile, firms are ranked in 5 groups based on
daily standard deviation of the return on market value of assets during the previous calendar year. The figure plots
coefficients on financial deficit for the size quintiles.
1
0.9
Tangibility quintile 1
0.8
(Low)
0.7
Tangibility quintile 2
0.6
Tangibility quintile 3
0.5
0.4
Tangibility quintile 4
0.3
0.2
Tangibility quintile 5
0.1
(High)
0
1
2
3
4
Asset volatility quintile
- 50 -
5
Table 12
Financing the deficit order across market-to-book ratio and asset volatility quintiles
The regression Dit  a  bnD DEFit   it is estimated for each MTB/asset volatility group. The table reports coefficients of
the financial deficit. Firms are sorted in quintiles according to market-to-book ratio MTB ((market value of equity+book
value of debt)/book value of assets), and then within each MTB quintile, firms are ranked in 5 groups based on daily
standard deviation of the return on market value of assets during the previous calendar year. OLS standard errors reported
below the coefficients in italics.
Asset volatility quintile
1 (Low)
2
3
4
5 (High)
MTB quintile 1 (Low)
0.880
0.891
0.888
0.774
0.388
0.005
0.005
0.005
0.007
0.008
MTB quintile 2
0.903
0.005
0.886
0.005
0.863
0.006
0.797
0.007
0.603
0.008
MTB quintile 3
0.833
0.006
0.801
0.007
0.777
0.007
0.695
0.008
0.476
0.008
MTB quintile 4
0.799
0.007
0.684
0.008
0.572
0.009
0.444
0.009
0.292
0.008
MTB quintile 5 (High)
0.518
0.009
0.261
0.008
0.194
0.007
0.141
0.006
0.099
0.005
Figure 6
Financing the deficit across market to book and asset volatility quintiles
The regression Dit  a  bnD DEFit   it is estimated for each size/ asset volatility group. Firms are sorted in quintiles
according to market-to-book ratio MTB ((market value of equity+book value of debt)/book value of assets), and then
within each market-to-book quintile, firms are ranked in 5 groups based on daily standard deviation of the return on
market value of assets during the previous calendar year. The figure plots coefficients on financial deficit for the
market-to-book quintiles.
1.000
0.900
0.800
0.700
MTB quintile 1 (Low)
0.600
MTB quintile 2
0.500
MTB quintile 3
0.400
MTB quintile 4
0.300
MTB quintile 5 (High)
0.200
0.100
0.000
1
2
3
4
Asset volatility decile
- 51 -
5
Table 13
Financing the deficit order across Z-score and asset volatility quintiles – unrated firms
The regression Dit  a  bnD DEFit   it is estimated for each Z-score/asset volatility group firms that do not have an S&P
credit rating. The table reports coefficients of the financial deficit. Firms are sorted in quintiles according to their Z-score
(= 3.3*(#170, pretax income)+(#12, sales)+1.4*(#36, retained earnings)+1.2*[(#4, current assets)-(#5, current
liabilities)]/(#6, assets) (see MacKie-Mason (1990))), and then within each Z-score quintile, firms are ranked in 5 groups
based on daily standard deviation of the return on market value of assets during the previous calendar year. OLS standard
errors reported below the coefficients in italics.
Asset volatility quintile
1 (Low)
2
3
4
5 (High)
Z-score quintile 1 (Low)
0.603
0.259
0.182
0.123
0.095
0.012
0.010
0.009
0.007
0.007
Z-score quintile 2
0.821
0.014
0.773
0.014
0.740
0.012
0.491
0.013
0.270
0.010
Z-score quintile 3
0.836
0.013
0.786
0.013
0.710
0.013
0.497
0.012
0.269
0.010
Z-score quintile 4
0.889
0.010
0.839
0.011
0.726
0.013
0.445
0.012
0.348
0.011
Z-score quintile 5 (High)
0.849
0.010
0.711
0.012
0.617
0.012
0.456
0.012
0.349
0.010
Figure 7
Financing the deficit order across Z-score and asset volatility quintiles – unrated firms
The regression Dit  a  bnD DEFit   it is estimated for each Z-score/asset volatility group for firms that do not have an
S&P credit rating. The table reports coefficients of the financial deficit. Firms are sorted in quintiles according to their
Z-score (= 3.3*(#170, pretax income)+(#12, sales)+1.4*(#36, retained earnings)+1.2*[(#4, current assets)-(#5, current
liabilities)]/(#6, assets) (see MacKie-Mason (1990))), and then within each Z-score quintile, firms are ranked in 5
groups based on daily standard deviation of the return on market value of assets during the previous calendar year. The
figure plots coefficients on financial deficit for the Z-score quintiles.
Non-rated firms
1
0.9
0.8
0.7
Z-score quitile 1(low)
Z-score quintile 2
Z-score quintile 3
Z-score quintile 4
Z-score quintile 5 (high)
0.6
0.5
0.4
0.3
0.2
0.1
0
1
2
3
4
Asset volatility
- 52 -
5
Table 14
Financing the deficit order across Z-score and asset volatility quintiles – rated firms
The regression Dit  a  bnD DEFit   it is estimated for each Z-score/asset volatility group firms that do have an S&P
credit rating. The table reports coefficients of the financial deficit. Firms are sorted in quintiles according to their Z-score
(= 3.3*(#170, pretax income)+(#12, sales)+1.4*(#36, retained earnings)+1.2*[(#4, current assets)-(#5, current
liabilities)]/(#6, assets) (see MacKie-Mason (1990))), and then within each Z-score quintile, firms are ranked in 5 groups
based on daily standard deviation of the return on market value of assets during the previous calendar year. OLS standard
errors reported below the coefficients in italics.
Asset volatility quintile
1 (Low)
2
3
4
5 (High)
Z-score quintile 1 (Low)
0.801
0.714
0.495
0.602
0.319
0.016
0.039
0.081
0.133
0.241
Z-score quintile 2
0.882
0.009
0.832
0.010
0.810
0.016
0.811
0.023
0.657
0.053
Z-score quintile 3
0.878
0.011
0.854
0.011
0.804
0.020
0.869
0.023
0.757
0.058
Z-score quintile 4
0.819
0.015
0.733
0.016
0.733
0.022
0.841
0.027
0.883
0.047
Z-score quintile 5 (High)
0.780
0.020
0.678
0.024
0.572
0.034
0.788
0.040
0.886
0.068
Figure 8
Financing the deficit order across Z-score and asset volatility quintiles – rated firms
The regression Dit  a  bnD DEFit   it is estimated for each Z-score/asset volatility group for firms that do have an
S&P credit rating. The table reports coefficients of the financial deficit. Firms are sorted in quintiles according to their
Z-score (= 3.3*(#170, pretax income)+(#12, sales)+1.4*(#36, retained earnings)+1.2*[(#4, current assets)-(#5, current
liabilities)]/(#6, assets) (see MacKie-Mason (1990))), and then within each Z-score quintile, firms are ranked in 5
groups based on daily standard deviation of the return on market value of assets during the previous calendar year. The
figure plots coefficients on financial deficit for the Z-score quintiles.
Rated firms
1
0.9
0.8
0.7
Z-score quitile 1(low)
Z-score quintile 2
Z-score quintile 3
Z-score quintile 4
Z-score quintile 5 (high)
0.6
0.5
0.4
0.3
0.2
0.1
0
1
2
3
4
Asset volatility
- 53 -
5
Table 15
Financing the deficit across asset volatility deciles for firms with investment grade debt
Pooled panel OLS regressions of net debt issues D on the financing deficit DEF are estimated for each decile: Dit  a  bnD DEFit   it ,
Ranking based on the daily standard deviation of the return on market value of firms assets during the previous calendar year. Firms with rank 10 have highest standard deviation.
Sample consists of firms with Z-score higher than 1.671. This cut-off value is the median Z-score for companies with S&P Domestic Issuer credit rating of BBB. Standard errors
are reported below the coefficients, in italics. All coefficients on financial deficit are significant at the 1 % level.
Dependent Variable: Net debt issued
Decile
2
3
5
6
7
8
9
-0.001
0.000
0.002
0.000
0.002
0.000
0.002
0.000
0.001
0.000
-0.001
0.001
-0.002
0.001
-0.002
0.001
-0.002
0.001
-0.002
0.001
Financial deficit
0.899
0.004
0.828
0.005
0.822
0.006
0.775
0.006
0.756
0.006
0.676
0.007
0.634
0.007
0.539
0.007
0.454
0.007
0.386
0.007
Adj. R squared
0.870
0.803
0.780
0.714
0.713
0.630
0.588
0.492
0.406
0.361
Number of Observations
6249
6230
6237
6230
6231
6239
6234
6233
6233
6220
Intercept
1 (Low)
- 54 -
4
10 (High)
Table 16
Financing the deficit across asset volatility deciles in the 70s, 80s and 90s
Pooled panel OLS regressions of net debt issues D on the financing deficit DEF are estimated for each decile in each period separately: Dit  a  bnD DEFit   it . Ranking based
on the daily standard deviation of the return on market value of firms assets during the previous calendar year. Firms with rank 10 have highest standard deviation. OLS standard
errors are reported below the coefficients, in italics.
Panel A: 1971-1980
Decile
1 (Low)
2
3
4
5
6
7
8
9
10 (High)
Intercept
-0.002
-0.001
-0.001
-0.001
-0.001
-0.001
-0.001
-0.001
-0.001
0.001
0.001
0.001
0.000
0.001
0.001
0.001
0.001
0.001
0.001
0.001
Financial deficit
0.916
0.007
0.838
0.007
0.900
0.006
0.862
0.007
Adj. R squared
0.880
0.861
0.898
1 (Low)
-0.005
0.001
2
-0.004
0.001
3
-0.001
0.001
0.848
0.869
Panel B: 1981-1990
4
5
-0.002
-0.004
0.001
0.001
Financial deficit
0.891
0.006
0.792
0.008
0.824
0.007
0.802
0.007
Adj. R squared
0.889
0.765
0.813
1 (Low)
-0.006
0.001
2
0.001
0.001
3
0.000
0.001
0.782
0.711
Panel C: 1991-2001
4
5
-0.002
-0.004
0.001
0.001
Financial deficit
0.829
0.006
0.837
0.006
0.771
0.007
0.717
0.008
Adj. R squared
0.804
0.809
0.741
0.667
Decile
Intercept
Decile
Intercept
- 55 -
0.887
0.007
0.842
0.007
0.798
0.008
0.788
0.008
0.725
0.009
0.534
0.010
0.847
0.789
0.781
0.709
0.504
6
-0.006
0.001
7
-0.008
0.001
8
-0.008
0.001
9
-0.009
0.002
10 (High)
-0.005
0.002
0.720
0.009
0.623
0.009
0.531
0.010
0.356
0.009
0.210
0.008
0.686
0.578
0.485
0.327
0.186
6
-0.005
0.001
7
-0.006
0.001
8
-0.005
0.001
9
-0.005
0.002
10 (High)
-0.004
0.002
0.648
0.008
0.454
0.008
0.337
0.008
0.209
0.007
0.150
0.006
0.100
0.005
0.600
0.423
0.298
0.181
0.136
0.089
0.758
0.008
Figure 9
Financing the deficit across asset volatility deciles in the 70s, 80s and 90s.
Pooled panel OLS regressions of net debt issues D on the financing deficit DEF are estimated for each decile in each
period separately: Dit  a  bnD DEFit   it . Ranking based on the daily standard deviation of the return on market
value of firms assets during the previous calendar year. Firms with rank 10 have highest standard deviation. OLS
standard errors are reported below the coefficients, in italics.
1.000
0.900
0.800
0.700
0.600
1971-1980
0.500
1981-1990
1991-2001
0.400
0.300
0.200
0.100
0.000
1
2
3
4
5
6
7
Asset volatility decile
- 56 -
8
9
10