Asset Specificity, Firm Heterogeneity and Capital Structure

Author(s): Srinivasan Balakrishnan and Isaac Fox

Source: Strategic Management Journal, Vol. 14, No. 1 (Jan., 1993), pp. 3-16
Published by: John Wiley & Sons
Stable URL: [Link]
Accessed: 13/09/2008 05:10

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
[Link] JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
[Link]

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the
scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that
promotes the discovery and use of these resources. For more information about JSTOR, please contact support@[Link].

John Wiley & Sons is collaborating with JSTOR to digitize, preserve and extend access to Strategic
Management Journal.

[Link]
 Strategic Management Journal, Vol. 14, 3-16 (1993)

ASSET SPECIFICITY,FIRM HETEROGENEITYAND

CAPITAL STRUCTURE
SRINIVASAN BALAKRISHNAN
Curtis L. Carlson School of Management, University of Minnesota, Minneapolis,
Minnesota, U.S.A.
ISAAC FOX
College of Business and Economics, Washington State University, Pullman,
Washington, U.S.A.

This paper presents an empirical investigation of the importance of specialized assets and
other unique characteristics of a firm in explaining the variance in capital structure across
firms. The results show thatfirm-specific effects contribute most to the variance in leverage,
suggesting a strong link between strategy and capital structure.

INTRODUCTION is that the firm's ability to finance these assets

with debt may be limited by the nature of the
Recent developments in transaction costs eco- assets themselves-their redeployability. Never-
nomics suggest that a firm's capital structure may theless, the firm also has an opportunity to
have more to do with strategic and control factors design and manage governance structures and
than with purely financial factors (Williamson, relationships such that this problem is mitigated.
1988). A firm often invests in firm specific assets The firm's ability to manage its relationships with
in order to enhance its uniqueness and competitive lenders thus becomes a key source of competitive
advantage.' Such assets also adversely affect the advantage. An empirical implication of this is
firm's ability to borrow. By definition, firm- that individual firm effects will have a significant
specific assets cannot be costlessly redeployed to impact on the firm's capital structure.
other uses and therefore cannot be used effec- Previous empirical studies, mostly by financial
tively as collateral for borrowing. Many firm economists, have tended to focus on tax shelter
specific assets are intangible-for example R&D effects and business risk in the presence of
and advertising-and difficult to measure and bankruptcy costs, as explanations for cross-
evaluate. Transactions in such assets will be sectional variations in leverage. Associated with
affected by informational asymmetry between the the business risk-bankruptcy costs hypothesis is
firm's insiders and outsiders. Consequently, they the notion that firms in the same industry, facing
may be thinly traded and if the firm faces similar supply and demand conditions, will have
liquidation, their market value can fall precipi- basically similar risk characteristics. This implies
tously. For the firm, specialized assets create that leverage will vary systematically across
both a problem and an opportunity. The problem industries. In this paper, we extend previous
empirical work on capital structure in an
important way. We simultaneously investigate the
Key words: Strategy, firm specific assets, capital importance of unique, firm specific characteristics
structure and industry characteristics in determining the
I For our purposes, firm specificity implies that the value of capital structure of the firm. Our results indicate
the asset is greater in its use within a particular firm rather that firm-specific effects are more important in
than in its next best use in some other firm. explaining the observed cross-sectional variation

0143-2095/93/010003-14$12.00 Received 25 June 1991

( 1993 by John Wiley & Sons, Ltd. Final revision received 2 June 1992
4 S. Balakrishnan and I. Fox

in leverage. The relationships between leverage the firm is entirely debt financed. This is
and certain determinants of capital structure, completely independent of the particular firm,
such as tax shelters and business risk, are also its strategy or industry. There are two problems
conditioned by the presence of firm-specific with this theory. First, we rarely observe firms
effects. Taken together, these two pieces of entirely financed with debt. Second, debt has
evidence lead us to believe that a firm's ability existed long before the advent of the corporate
to borrow and hence its capital structure, are income tax. A number of theories have attempted
closely linked to its business strategy and the to find an optimal capital structure that is not
nature of the assets and skills required to 100% debt. The main thrust of the arguments in
implement that strategy. these theories is that the tax benefits of debt are
This paper is organized as follows. First, offset by a variety of costs that are due to capital
we review several financial theories of capital market imperfections (Miller, 1989).
structure and the empirical evidence for these
theories. Next, we develop the arguments that
Nondebt tax shields
link strategy and firm-specific assets to capital
structure and derive testable implications. Finally, Under the current U.S. tax laws, some invest-
after a brief description of the techniques of ments may generate tax benefits that are unrelated
panel data analysis, we present the results of an to how the investments are financed (DeAngelo
empirical investigation of these implications using and Masulis, 1980). Depreciation, investment
a sample of 295 single business firms. The results tax credits, and depletion allowances are tax
indicate a strong link between strategy and capital deductible in the same way as interest on debt.
structure. Advertising, and R&D can also be thought of as
generating nondebt tax shields. These nondebt
tax shields do not involve any debt related costs
FINANCIAL THEORIES OF CAPITAL but act as substitutes for debt tax shields,
STRUCTURE particularly when for some reason, part of the
firm's total tax shields must go unused. For
Assuming perfect and complete capital markets, example, if the firm faces financial distress, it is
costless and symmetric information, value-maxi- more likely that the tax benefits of debt will go
mizing decision makers, and no taxes, Modigliani unused. Nondebt shields will then tend to act as
and Miller (1958) show that capital structure substitutes for debt shields. We would therefore
choices are irrelevant to the firm's investment expect a negative relationship between the extent
decisions, its overall cost of capital, and its value. of nondebt tax shields and leverage.
In this simplified Modigliani and Miller world,
there will not be any systematic variation in
Agency theory and debt
capital structure across firms. The implications
for managers are quite clear. They should ignore Significant agency costs arise from the fundamen-
financing issues and undertake investments that tal conflict of interest between stockholders and
maximize profits and firm value. Violations of bondholders. Stockholders are mainly concerned
the underlying assumptions of this model provide about the upper part of the probability distri-
the basis for the notion that there exists an bution of possible performance outcomes-over
optimal capital structure which maximizes the and above the amount required to repay debt.
overall value of the firm. Bondholders on the other hand receive only the
specified payment in the debt contract and
nothing of the cash flows above that payment.
Debt tax shields
Therefore, they are concerned with only the
When tax laws permit the deductibility of interest lower end of the probability distribution of
expenses, firms can increase their net cash outcomes. Riskier projects therefore reduce the
flows by financing with debt rather than equity expected pay-offs to bondholders. This can lead
(Modigliani and Miller, 1963). Borrowing lowers to what is called the 'asset substitution problem'
the real after-tax cost of capital and raises the in the finance literature (Myers, 1977). Once
value of the firm. Total value is maximized when the bondholders have advanced money to the
 Strategy and Capital Structure 5
stockholders, stockholders have an incentive to managers of losing their jobs and/or damaging
take on projects that are riskier than what the their reputation. If the firm defaults and faces
bondholders would prefer. The bondholders bankruptcy, there is a change of control over the
recognize this incentive and will charge a higher underlying assets of the firm from the stockholders
price for debt capital. These agency costs may and managers to the bondholders. This change
be attenuated by structuring the contract in of control is usually costly.2 The direct costs are
certain ways-debt covenants, conversion rights, the legal fees, delays and bargaining costs
call provisions, etc.-but, they can never be involved in the bankruptcy proceeding itself.
completely eliminated as long as the bondholders Indirect bankruptcy costs may be more important
cannot perfectly discern whether the outcome for managers. Titman (1984) argues that the
was the result of uncertainty or the actions of value of a durable good (which requires service,
the owner-manager to take on an excessively maintenance or updating) falls as the likelihood
risky project (Jensen and Meckling, 1976). of liquidation increases. The customers recognize
Another agency issue associated with debt is that they may be 'left holding the bag' and take
the 'under-investment problem'. There are some that risk into account and reduce demand. Firms
positive net present value projects that the selling these kinds of products face higher
stockholders would accept if the firm were totally bankruptcy costs. The managers personally bear
equity financed, but would reject when the firm a substantial portion of these bankruptcy costs
is partially debt financed. While the pay-offs to (Titman, 1984).
the investment may be large enough to be Faced with the burden of potential bankruptcy
profitable they may not be sufficient to repay from increased debt, managers are less likely to
the debt holders. In this case the lenders would take on unprofitable investments that they
get the rights to the positive pay-offs and the otherwise would. The 'free cash flow' (Jensen,
stockholders would get nothing. Both asset 1986) or the residual cash flow after all positive
substitution and underinvestment problems result net present value investment needs have been
in some 'good' projects being rejected because met, will be used to pay off the lenders rather
of the distribution of pay-offs and the capital than to expand the manager's empire. If the
structure of the firm. Myers (1977) argues that incumbent management team is unwilling to
this sort of problem is especially serious for assets increase debt, then another team has an incentive
that give the firm the option to undertake growth to take over the firm and restructure it. The
opportunities in the future. The greater the firm's types of firms that are most likely candidates for
investment in such assets the less it would be using debt to discipline managers would be
debt financed, indicating a negative relationship firms whose cash flows exceed their growth
between leverage and growth opportunities. opportunities. Thus the agency related problems
between managers and stockholders will result
in a negative relationship between leverage and
Free cashflow, debt, and bankruptcy
growth opportunities, in just the same way as
So far, we have ignored any conflict of interest the agency issues between stockholders and
between stockholder and managers. Agency bondholders.
theory (Jensen and Meckling, 1976) suggests that
managers have an incentive to over-expand the
Industry and capital structure
size and scope of the firm to satisfy their own
ends at the expense of the shareholders. A larger Although bankruptcy is a firm level phenomenon,
firm may have more career and promotion Scott (1980) has argued that industry structure
opportunities, more status, and less risk of may be an important determinant of bankruptcy
bankruptcy, especially if it is diversified. Gross-
man and Hart (1982) and Jensen (1986) have 2
Stiglitz (1972) has shown that capital structure was irrelevant,
argued that there can be 'agency benefits' to even with the possibility of default, as long as the transfer
debt if it reduces the scope of managerial of ownership that takes place during bankruptcy could be
discretion. done costlessly. Risky debt will then be priced like any other
risky asset. Contracts can still be costlessly written and
Higher debt increases the probability of bank- enforced. For bankruptcy to be costly there must be some
ruptcy. This, in turn, increases the risks to the effect on the real cash flows of the firm.
6 S. Balakrishnan and I. Fox

risk and bankruptcy costs. For example, when industry effects on capital structure when they
there are decreases in demand, lenders may estimated their model with a set of industry
believe that it will be more likely that small firms dummies. The magnitude and significance levels
that are members of the competitive fringe will of the coefficient fall but the signs remain the
be pressured to liquidate and exit, imputing same and all of the variables in their model
higher bankruptcy costs for these firms. There remain significant. Several industrial organization
may also be large interindustry differences in the economists have also studied the industry effects
variability of cash flows. Firms that produce on capital structure. The results are mixed. Caves
similar products, and use similar technology, may and Pugel (1985) find that of all their structural
face the same level of uncertainty from demand variables only advertising is significant and
and factor market shocks. This implies industry negatively correlated to leverage. Spence (1985)
level effects on capital structure operating through finds that advertising and market share are
bankruptcy risk. significant, the former negatively and the latter
positively correlated to leverage. Scott (1980)
finds that the relative use of short-term debt
Empirical evidence
decreases with market concentration, implying
The various theories that we have briefly surveyed that lenders may view firms in more concentrated
imply that the important financial determinants industries as less risky and are willing to lend
of capital structure are: (i) nondebt tax shields, for longer periods.
(ii) earnings volatlity, (iii) growth opportunities, The upshot of these empirical studies is that
and (iv) industry structure. On balance, evidence neither the financial factors such as nondebt tax
from previous empirical studies lend support to shields and income volatility nor the structural
a negative relationship between leverage and characteristics of the industry seem to explain
proxies representing earnings volatility. Support much of the observed variation in capital structure
for predicted relationships between leverage and across firms. The contradictory evidence on
the other variables is weak and inconclusive. nondebt tax shields-the positive effect of
Among the various empirical studies by finan- depreciation and the negative effect of R&D and
cial economists, the most comprehensive and advertising on debt raises the doubt that other
relevant to this paper is the one by Bradley, factors may be at play here. Recently, Williamson
Jarrell, and Kim (1984). Bradley et al. find that (1988) has offered a transaction costs based
the relationship between earnings volatility and perspective on debt. Combined with the resource
leverage as measured by the ratio of the book based view of the firm, Williamson's approach
value of the firm's long-term debt to the market allows us to examine the linkage between a firm's
value of its equity, is significant and negative. capital structure and its business strategy.
One of the most important nondebt tax shields,
depreciation, was significant and positively
related to leverage, in the study by Bradley et RESOURCES, STRATEGY AND
al. This is opposite to what is predicted by CAPITAL STRUCTURE
theory. One explanation is that depreciation is
Corporate governance and business strategy
actually proxying for collateral assets that can be
held as security against the debt, in which case Williamson (1988) has argued that debt and
a positive relationship with debt would be equity ought to be viewed as different forms of
expected. Bradley et al. also find that the 10- governance structure. Debt, like market, is less
year average sum of advertising and R&D to sales interventionist and the bond-holders can seize
ratio is significantly and negatively correlated with control over the firm's assets if and only if the
leverage. Bradley et al. conclude that this measure firm had defaulted or violated the covenants of
is acting as a proxy for either or both nondebt the debt contract in some way. Equity is similar
tax shields and the agency costs of investment. to hierarchical control. The rights of the equity-
Both Long and Malitz (1985) and Titman and holders are much more general than those of the
Wessels (1988) find qualitatively similar results bond-holders. They can exercise these rights
for R&D and advertising. through the board of directors, by monitoring
Bradley et al. (1984) also found significant the conduct of management and intervene in
 Strategy and Capital Structure 7

strategic decisions whenever it is deemed necess- against such assets afford limited protection.
ary. Using his well known transaction costs Consequently, the cost of financing these assets
framework, Williamson argues that the type of with debt will increase. The trade-off between
governance chosen-debt or equity-will depend asset specificity and leverage works not only
on the characteristics of the assets employed in through the firm's ability to make its own
the venture, particularly their redeployability to specialized investments, but also through the
other uses. This relationship between the nature willingness of its trading partners to make the
of the assets and their control derived from the kinds of specialized investments needed to further
type of financing, provides the link to the firm's enhance the firm's competitive advantage. If
business strategy. employees think that the probability of continued
Many researchers in business strategy have employment is eroding, they may not make
stressed the importance of unique and inimitable investments in firm-specific human capital
assets, resources, skills, relationships and invest- (Titman, 1984). Distributors may be unwilling to
ments as the primary sources of a firm's sell the firm's product line if they expected that
competitive advantage (Barney, 1986; Lippman any product-specific investment they might make
and Rumelt, 1982; Montgomery and Wernerfelt, would lose value.
1988; Rumelt, 1991). The relationship between A key issue is the extent to which lenders can
firm specific assets and capital structure, however, evaluate and monitor the manager's investments
has not been the subject of inquiry until recently and activities. When informational asymmetries
(Barton and Gordon, 1988; Bettis, 1983; Turk cause well managed, 'good' quality firms, to be
and Hoskisson, 1991). Assets specially tailored pooled with 'poor' quality firms, lenders cannot
to the firm's strategy and technology can reduce perfectly differentiate between borrowers of
costs, improve quality and enable the firm to different risk types creating a version of the
differentiate its products and services from those 'lemon problem' (Akerlof, 1970). As they
of its competitors. Such firm specific assets- increase the interest rate to cover high risk
especially intangible assets like R&D, brand borrowers, the better quality (lower risk)
name and other reputational investments-may borrowers drop out of the market and eventually
be difficult for outsiders to monitor, understand, only the worst quality borrowers will be willing
and evaluate. These assets are less redeployable to accept the contract. Further, these riskier
to other uses than general purpose assets-an borrowers have an even stronger incentive to
emergency power plant, for example. Secondary take on even higher risk projects because their
market for such assets may not value them as expected pay-offs increase with risk. Beyond a
much as the firm and sometimes may not even certain threshold quality of the risk pool,
exist (Klein, Crawford, and Alchian, 1978; lenders will begin to ration credit among
Williamson 1975, 1986). This property of assets, borrowers of different risk classes instead of
known as 'asset specificity,' affects the firm's adjusting the interest rates, just as firms lay off
capital structure through bankruptcy costs which employees instead of adjusting their wage rates
reflect the loss in firm value due to the occurrence (Stiglitz and Weiss, 1981). Faced with this
of, or an increase in, the likelihood of financial form of credit rationing, firms may tend to
distress.3 underinvest in R&D and other intangible assets
that rely on private, firm specific information,
the release of which may reduce the value of
Firm-specific assets and debt
the project as a source of competitive advantage
In the event of bankruptcy and liquidation, the (Myers and Majluf, 1984; Turk and Hoskisson,
firm's more specialized assets face a greater loss 1991).
in value and the pre-emptive claims of the lenders The following propositions relating the nature
of the firm's assets and its leverage can be derived
3If all that is involved in a default is a costless transfer of from this discussion:
assets between stockholders and bondholders, it will not
affect firm value. For bankruptcy costs to be relevant to firm Proposition 1: A firm's leverage should be
value they must involve some sort of payments to third
parties besides the two classes of claimants. See Stiglitz, positively related to the redeployability of its
1972. existing assets.
8 S. Balakrishnan and I. Fox

Proposition 2: A firm's leverage will be name etc. become realized as reputational assets,
positively related to its investments in tangible they may require continued expenditure much
assets. like maintenance to keep physical facilities from
depreciating too rapidly. The presence and
Proposition 3: A firm's leverage will be maintenance of reputational assets may therefore
negatively related to its investments in intangible, be taken by lenders as commitment by the firm
firm-specific assets. to be an aggressive competitor in its product
market, leading to lower cost of debt financing
for such firms.
Signaling
When the capital market cannot adequately Proposition 4: A firm's leverage will be
differentiate between good and bad borrowers, positively related to its investment in reputational
the good firms have an incentive to signal their assets that signal commitment to the product
quality by undertaking some action that would market in which it competes.
be too costly for poor firms to imitate (Spence,
1974). Most of the signaling models in financial To sum up, the resource-based view suggests
economics involve firms using their capital that investments that rely on private information
structure, dividends, or managerial ownership to and unique, firm-specific assets are essential
signal to the capital market their underlying value for superior performance. The transaction-costs
(Myers and Majluf, 1984; Ross, 1977). We can framework suggests that debt governance will
go beyond the formalistic notion of signaling inhibit investments in specialized assets because
equilibria in these models and use the term in a they cannot be readily redeployed and are
broader sense to include situations where a firm therefore poor security to lenders. Such invest-
wants to commit itself to taking a particular ments are more likely to be financed with
course of action. It may do so to gain credibility equity. Signaling models, however, suggest that
with the other players in the game. Brander and reputational assets such as brand names, although
Lewis (1986, 1988) have developed models in intangible, may be interpreted by lenders as
which firms commit themselves to being aggres- commitments to the product market and viewed
sive in their product markets by taking on positively. All of these theories imply that unique,
additional debt. The firm that increases its debt firm-specific factors play an important role in the
commits itself to increasing output for each level determination of the capital structure of the firm.
of its competitor's output. In this game-theoretic
model, the firms' strategic options are limited to
output decisions. However, aggressiveness in the METHODOLOGY
product market can take many other forms-
increased advertising, quality improvement, prod- In the rest of the paper we present an empirical
uct differentiation, segmentation and focus, etc. investigation of some testable implications derived
An alternative scenario is that in some markets from the business strategy-transaction costs-
firms may find it necessary to invest in repu- capital structure linkage. Our methodology differs
tational, and brand name assets in order to from previous empirical studies of capital struc-
compete and be successful (Klein and Leffler, ture in several ways. First, in contrast to the
1981; Shapiro, 1982, 1983). After some period several studies that had examined the industry
of time, these reputational assets, even though effects on capital structure, we will assess the
they are firm specific and intangible, form a basis relative importance of industry and firm effects
of security for lenders. The firm's reputation can on the firm's capital structure using panel data.
be seriously damaged if it defaulted on a loan. We will also explore whether the predicted
The firm may also be in a situation in which, if relationships between leverage and the explana-
it reduced its level of advertising, the action can tory variables such as firm-volatility, collateral,
be construed by the capital market as evidence R&D, and advertising growth opportunities have
that the firm is in some sort of trouble-perhaps a significant firm-specific component. Second,
it is planning to harvest and exit the market. The most of the studies that investigate industry
implication is that once investments in brand effects on leverage are variations of one-way
 Strategy and Capital Structure 9

ANOVA where the implicit assumption is that are close to either of the extreme values:
the industry effect is a fixed effect. This is log[Leverage / (1-Leverage)]. Flath and Knoeber
(1980) use this transformation in their leverage
reasonable if we are primarily concerned about study. The logistic transformation will allow
the effect of particular levels (i.e., particular us to maintain the assumption of normally
industries) in the study. However, if the levels distributed error terms and use linear estimation
in the sample are more appropriately thought of procedures.
as a random sample from a larger population, it Earnings Volatility (RISK): Various measures
is more appropriate to use the random effects have been used in the literature to measure the
model otherwise called the variance or error earnings volatility or business risk which is
components model (Dielman, 1989; Hsiao, 1986). predicted to be negatively related to leverage.
Bradley et al. (1984) use standard deviation of
Finally, virtually all the previous studies seem to the annual percentage change in cash flows
have been biased against finding industry effects (earnings before interest, depreciation and taxes)
by including diversified firms in the sample. for the previous 10 years. Titman and Wessels
Capital structure decisions are made at the (1988) use operating income instead of cash flow
in constructing their measure of risk. We used
corporate level and not at the business unit level.
both measures in our regressions and did not
In a diversified firm two possible effects on debt find any qualitative difference in the results. The
are confounded-corporate diversification and results presented are for the measure of business
industry. Diversification may lead to the reduction risk used by Bradley et al.
of total risk, paving the way for increased debt- Depreciation (DEPRN): As we discussed earlier,
what is called the 'coinsurance effect' in the nondebt tax shield are substitutes for interest
finance literature (Kim and McConnell, 1977). deduction for tax purposes. Bradley et al.
In this study, we control for diversification by used the ratio of the sum of depreciation,
restricting our sample to single business firms. amortization, and investment tax credits to
earnings before interest, depreciation and taxes
to assess the effect of nondebt tax shields on
leverage). Titman and Wessels (1988) used the
Sample and data
ratios of depreciation and investment tax credit
Our sample consists of 295 single business firms to total assets as separate constructs. In our
operating in 30, 3-digit SIC industries for the regressions, we used the same measure that
Bradley et al. used as a proxy for redeployable
years 1978-87. The panel data set was constructed assets. Propositions 1 and 2 suggest that
as follows. First, all the firms operating in a depreciation should be positively related to
single 4-digit SIC industry were extracted from leverage.
the Disclosure data base which includes all R&D Intensity (R&D) and Advertising Intensity
firms filing with the Securities and Exchange (ADVERT): Previous studies have used the ratio
Commission. From these, we selected a subset of R&D and advertising expenses to net sales
of firms operating in mining and manufacturing as proxies for nondebt tax shields as well as
intangible assets of the firm (Bradley et al.,
(SIC 1000 to SIC 3999) and that are included in
1984; Titman and Wessels, 1988). To the extent
the COMPUSTAT data base. We excluded all that it is difficult for outsiders to assess their
firms that had more than two consecutive missing value, they may also be interpreted as invest-
observations for any variable in the model ments in assets that cannot be readily redeployed.
during the period 1978-87. The firms were then This interpretation implies that R&D and adver-
aggregated into 3-digit SIC industries to give us tising intensity will be negatively related to
leverage. These two measures thus test Prop-
the final sample of 295 firms with a minimum of ositions 1 and 3.
four firms per industry. The following variables
were constructed using the raw data from Growth Opportunities (GROWTH): The firm's
growth opportunities are difficult to measure
COMPUSTAT: and guidance from previous studies is scant. As
proxies for growth opportunities, previous studies
Leverage (LEVER): Leverage is measured as have used the ratio of capital expenditure to
the ratio of the book value of total debt (long- total assets, percentage change in total assets
term and short-term debt) to market value of (Titman and Wessels, 1988), and operating cash
equity and book value of debt. Because this flows (Long and Malitz, 1985). We constructed
ratio is bounded at zero and one, we also used a similar proxy for the firm's future growth
a logistic transformation to avoid problems that prospects as the ratio of capital expenditures on
may occur if a large number of observations property plant and equipment to earnings before
10 S. Balakrishnan and L. Fox
interest, depreciation and taxes (EBIT), to We assume that the firm effect (FIRM), the
test the agency theory prediction that growth industry effect (INDUSTRY), the year effect
opportunities will be negatively related to lever-
age. (YEAR) are the independent random variables.
The total variance in leverage (LEVER or
Apart from these explanatory variables, we LOGLEVER), can be decomposed into com-
used year (YEAR), the 3-digit SIC code ponents associated with each of these independent
(INDUSTRY), and the COMPUSTAT identifi- random variables. The formal models take the
cation number for the firm (FIRM), as class following form:
variables to assess the relative importance of
time, industry, and firm effects in explaining cross- Var(LEVER) = Var(CUS1P) + Var(lND US-
sectional and longitudinal variation in capital TRY) + Var(YEAR) + Var(ERROR)
structure. The year effect is expected to pick Var(LOGLEVER) = Var(CUS1P) + Var(1NDUS-
up any year to year changes in policies and TRY) + Var(YEAR) + Var(ERROR)
regulations-for example, tax reforms-that
might affect capital structure decisions. The We used the MIVQUE method in SAS to
industry effect is meant to capture various estimate the variance components.
unspecified structural characteristics that may
affect leverage. These will include entry barriers, Error components model
capacity conditions, industry growth rate, factor
market conditions, and industry-wide business In addition to the variance components model,
risk and industry-specific capital market imperfec- we also estimated two regression models to
tions. The firm effects will take into account identify the significant determinants of capital
unspecified resources, growth opportunities, con- structure, using leverage and its logarithm as
tractual relationships (formal or implicit) and dependent variables. The general forms of the
skills. regression models are:
We estimated a restricted form of this general
model which allowed for some omitted variables.
Variance components model
We used a variance components model to LEVER i,t = o i + 1 jRiSK + P2,iDEPRN +
decompose the total sample variance in capital 3jR&D + 134ADVERT + 5jGROWTH + Efit

structure into firm, industry and time com- LOGLEVER i,t = P i + 1PI,RISK + P2,
ponents. The underlying assumption in the DEPRN + P3,iR&D + P4,IADVERT + P5,
variance components model is that we may not GROWTH + Elit
be able to correctly specify all the explanatory
variables that determine capital structure. Sometimes referred to as error components
Instead, we try to assess how much of the overall model, it assumes that the slopes of the explana-
variance in capital structure is attributable to tory variables are the same across firms but that
unspecified firm, industry, and year to year the intercepts are different, that is,
differences. The percentages of these 'variance
components' should inform us of their relative N. i $ o,,, for all i $ j
importance. In contrast with standard analysis of
variance (ANOVA) and multiple comparison The cross-sectionally varying intercept captures
tests which try to assess whether the leverage the combined effects of firm level omitted
significantly differs between class levels such as
any two industries or firms, the approach here explanation for observed differences in firm performance,
is to assess how important these differences are accounting for nearly 20 percent of the variance in firm
profits. Rumelt (1991), on the other hand, finds that although
in determining capital structure.4 industry level effects account for nearly 16 percent of the
variance in firm returns, more than 46 percent of the variance
was attributable to stable business-unit level effects. While
4Schmalensee (1985) and Rumelt (1991) have used this Schmalensee argues that his results are consistent with the
method to assess the relative importance of firm, industry focus on industry level analysis, Rumelt argues strongly that
and time effects on industry and corporate returns. Schma- we should concentrate our attention on the business unit
lensee finds that industry level effects constitute the prime level and focus on the sources of firm heterogeneity.
 Strategy and Capital Structure 11

variables. Rather than treating the intercepts as sample of single business firms. If industry effects
fixed effects (as one would in a dummy variable dominate, single business firms operating in the
regression), they are assumed to be random same industry should face fairly similar conditions
variables. As before, we assume independence for raising funds in the capital market. On the
among the error components. Additionally, the contrary, our results show that the industry and
effects of the error components are assumed to time effects are of little importance when
be independent of the errors in the explanatory compared to issues related to firm uniqueness,
variables. in determining the variability in capital structure.
For strategy researchers this is further evidence
that we should focus on the unique characteristics
RESULTS of the firm to understand its behavior rather than
assuming firm homogeneity and looking at the
Variance components of leverage
structural characteristics of the industry or the
Table 1 presents the results from the two variance larger economy.
components model using both leverage and its
logistic transformation as dependent variables.
Error components regression
The results show unambiguously that the firm
effect is by far the most important. Using the Table 2 presents the results for the regressions.
first measure, it accounts for over 52 percent of the Apart from minor difference in magnitude and
total variance in capital structure. Interindustry significance, the results are the same for leverage
differences account for 10 percent of the total and logarithm of leverage as dependent variables.
variance and the year effect accounts for only Aside from the sign on the advertising to
about 1 percent of the variation in leverage. sales ratio these results are consistent with
When LOGLEVER is used as the measure of the predictions for capital structure based on
leverage, firm effects account for 50.5 percent of transaction costs and resource-based theories.
the variance in capital structure and the industry The significant negative sign on the volatility
effects account for only about 5 percent. measure RISK is evidence for the business-risk
There are several implications from these based arguments. The positive and significant
results. If one expects to find industry effects to slope for the nondebt tax shield measure DEPRN
be important, it should have been evident in a is consistent with the results obtained in previous
studies and supports Propositions I and 2.
Depreciation is usually proportional to the value
Table 1. Variancecomponent models of physical assets such as buildings, plant and

Dependent Variable:LEVER
Table 2. Regressionresults
Variancecomponent Estimate Percent
Dependent variable Dependent variable
FIRM 0.03001 52.1 Variable LEVER LOGLEVER
INDUSTRY 0.00602 10.5
YEAR 0.00064 1.1 ,B, -0.00016 -0.00262
ERROR 0.02090 36.3 RISK (0.00026) (0.00202)
TOTAL 0.05757 100 P2 0.006*** 0.02821*
DEPRN (0.00153) (0.01190)
Dependent variable:LOGLEVER P33 -0.03927* -0.36502*
R&D (0.02366) (0.18541)
P4 0. 14778*** 0.84011**
Variancecomponent Estimate Percent ADVERT (0.03419) (0.26347)
,B5 -0..00171* -0.00498
FIRM 1.5278 50.5 GROWTH (0.00094) (0.00730)
INDUSTRY 0.1662 5.5 R2 0.853 0.824
YEAR 0.0495 1.6 F value 38.51*** 28.80***
ERROR 1.2801 42.3
TOTAL 3.0236 99.9 (t-statistics in parentheses). ***p < 0.0001, **p < 0.01,
*p < 0.05.
12 S. Balakrishnan and I. Fox

equipment, that act as collateral and provide intangibility to specificity or redeployability of

security for lenders. The ratio of depreciation to assets.
total assets is therefore an index of the extent to We did an F-test to see whether the interfirm
which the assets of the firm are redeployable. differences in the intercepts are significant given
Debt will be the preferred method of financing that the slopes are identical across firms. The
such assets. The coefficient on capital expendi- tests reject the hypothesis that the firm intercepts
tures on plant and equipment as a percentage of are identical across firms at the significance level
earnings (GROWTH), is negative and weakly 0.01 or better. This result, in conjunction with
significant @.07 when using leverage as the the results from the variance components model,
dependent variable. This is in accordance with strengthens our belief that unspecified firm-level
the prediction from agency theory. The slope variables account for a substantial portion of the
was negative but not significant when we used variance in capital structure. The intercepts, as
the logistic transformation of leverage as the we mentioned earlier, may be seen as capturing
dependent variable. Our interpretation of this the effects of each firm's unique ability to manage
weak result is that the existence of substantial its relationships with lenders. We may infer that
physical plant reduces the firm's bankruptcy risk firms with higher intercepts find it relatively
and may make scale-increasing expansion of these easier to borrow and on average have a greater
assets irrelevant to leverage decisions. percentage of their total value made up of debt,
The negative and significant sign of the slope notwithstanding the common effects of capital
for R&D intensity is consistent with Proposition market imperfections due to asset redeployability,
3. Firms that tend to invest heavily in R&D agency costs etc. that they may face.
which potentially creates intangible and firm- The significant differences in the intercepts
specific knowhow will find it more difficult to raise another interesting issue. After all, if firm
fund such investments with debt. To the extent heterogeneity is the dominant characteristic of
that this intangible knowhow is not redeployable, business, the relationship between capital struc-
this result supports Proposition I as well. One ture and R&D, advertising, earnings volatility
of the interesting results of this study is the and growth opportunities can also be expected
positive and significant coefficient for the advertis- to vary across firms. A separate regression for
ing to sales ratio, ADVERT. This result implies each firm is desirable, but currently infeasible
that firms that spend more on reputational assets for lack of a time-series of adequate length for
such as brand name can leverage more. Even the firms in our sample. As a compromise we
though image and brand names are intangible divided the firms into four groups based on the
they may be bought and sold without much quartiles of their standardized intercepts which
transaction costs as some recent buyouts and are rough indices of the firms' competence in
subsequent asset sales have indicated.5 Also, the managing lender relationship. Table 3 reports
firm's investments to gain customer loyalty may the means and standard deviations for the
form a basis of security for lenders once those dependent and independent variables for each of
reputational investments have been realized as the four groups. Not surprisingly, the upper
capital assets. Brand names are like durable quartile groups on average have higher levels of
goods that require a high level of maintenance advertising intensity and decreasing levels of
expenditure and therefore higher levels of adver- R&D intensity. Both capital expenditures and
tising may be seen as a signal that the firm nondebt tax shields are higher for the higher
intends to stay in a market and not harvest or intercept groups. However, there is a large
liquidate. Thus investments in reputation assets decrease in income volatility for the highest level
that increase the firm's uniqueness may actually quartile.
increase its ability to borrow. The result also In Table 4 we present the results from OLS
underscores the importance of not equating regressions for each quartile group.6 We assume
that within each quartile group the 3 coefficients

I See Bhagat, Shleifer, and Vishny (1990) for a related 6 We present only the results obtained by using the logistic
discussion on the gains from hostile takeovers and subsequent transformation of leverage as the dependent variable. The
sell-offs of brand names and other assets to strategic buyers. results were similar for leverage as the dependent variable.
 Strategy and Capital Structure 13
Table 3. Group means and standard deviation

Dependent variable: LOGLEVER

Variable All firms QUARTI QUART2 QUART3 QUART4

Std -3.919 -8.867 -4.829 -2.439 -0.4465

Intercept (3.614) (1.825) (0.7391) (0.643) (1.332)
Loglever -1.442 -3.239 -1.825 -0.9033 0.146
(1.729) (1.517) (1.100) (1.004) (1.516)
RISK 4.257 3.532 5.604 4.899 2.946
(16.703) (10.359) (25.485) (17.467) (6.684)
DEPRN 0.4004 0.2086 0.2784 0.4317 0.6871
(4.547) (3.576) (4.231) (5.683) (4.425)
R&D 0.0627 0.0877 0.0844 0.0444 0.0237
(0.1991) (0.2549) (0.2449) (0.1374) (0.0412)
ADVERT 0.0392 0.0113 0.0184 0.0658 0.0624
(0.2866) (0.0265) (0.0628) (0.522) (0.2295)
GROWPRO 0.6365 0.3167 0.4865 0.8602 0.8851
(6.498) (7.986) (3.997) (6.989) (6.340)

Standard Deviation in parenthesis.

Table 4. Regression results

Dependent variable: LOGLEVER

Variable QUART1 QUART2 QUART3 QUART4

Intercept -3.3033*** - 1.703*** -0.8318*** 0.2473**

(-40.437) (27.14) (18.056) (3.51)
PIBRISK -0.01625** -0.00367* -0.00065 -0.00517
(-2.68) (2.107) (-1.847) (-0.384)
P2 DEPRN 0.02993 0.05965** -0.002 0.03538
(0.748) (2.623) (0.061) (1.795)
33 R&D -0.26345 -1.453** -1.5343*** -0.49334
(-0.993) (3.309) (-5.407) (-0.357)
P4 ADVERT 13.8098*** 0.65239 0.0705 0.5205**
(4.11) (0.562) (1.69) (2.585)
,B5GROWTH 0.00353 -0.03933 0.00453 -0.00438
(0.197) (1.53) (0.222) (-0.358)
R2 0.0608 0.0552 0.0728 0.0361
F value 5.746*** 5.025** 6.487*** 2.522*

(t-statistics in parentheses), ***p < 0.0001, **p < 0.01, *p < 0.05

are common across firms but we allow the R&D may be negative for a representative firm,
coefficients to vary between groups. The results some firms-possibly because of their reputation
indicate that both the magnitudes and significance in the capital market for converting research
of the 1 coefficients are less for the higher expenditures into successful projects-may face
quartile groups. This may be interpreted as a few credit constraints if any, in financing their
weakening of the underlying relationship between R&D with debt. The slope for R&D will then
the redeployability of the firm's assets and its be zero or even positive for these firms. In a
ability to borrow for firms that have a better market where research and product innovation
relationship with their lenders. For example, are important aspects of competitive strategy,
although the relationship between leverage and firms that face lesser credit constraints will have
14 S. Balakrishnan and 1. Fox

a competitive advantage. We do not therefore dominant characteristic of business, the relation-

claim any statistical validity for the varying slopes ship between capital structure and R&D, advertis-
because the small number of separate regressions ing, earnings volatility and growth opportunities
does not allow any meaningful test of homogen- can be expected to vary across firms. The
eity of slopes across the groups. Besides, the differences in the magnitudes and significance of
partitioning of the sample into quartile groups the slopes for these variables between the
based on the estimated intercepts may have restricted error components model and the
introduced a bias. Nevertheless we believe that regressions for the four groups of firms, indicates
the results are sufficiently interesting to merit that this may indeed be so. Fourth, we had
further study of cross-sectional variance in the focused our attention on the strategic and control
determinants of capital structure. factors driving the capital structure. There is
possibly a simultaneity in the firm's strategic and
financing decisions. It may be worthwhile to look
CONCLUSIONS at the random-effects model in a simultaneous
equation framework. Currently, data limitation
It is important for both strategy and finance and methodological difficulties preclude us from
researchers to recognize that once the separation doing this. Finally, we have ignored certain
between the firm's financing and investment qualitative issues in finance and strategy that
decisions breaks down, the firm's competitive clearly are related to the firm's resource base,
strategy becomes an important determinant of for example, internal vs. external financing and
its capital structure. Evidence from our study of the operation of the internal capital market
295 mining and manufacturing firms strongly (Williamson, 1975; Hill, 1988; Hoskisson and
suggests that unique firm-specific assets and skills Hitt, 1988). Such issues may have a greater
are by far the most important determinants of impact on the relationship between strategy and
capital structure. Structural characteristics of capital structure in diversified firms which have
industry and/or the notion of large interindustry been excluded from our sample.
differences in risk are not nearly as important as
the firm-specific aspects of the management of
this risk and its implications. ACKNOWLEDGEMENTS
There are several possible extensions of this
study that we think may give more insight into The authors gratefully acknowledge the comments
the relationship between the firms strategy and and suggestions made by several colleagues
capital structure. First, managerial ownership and including Tom Brush, Elaine Mosakowski,
control are intrinsically related to the agency and Howard Thomas, Birger Wernerfelt, Oliver
transaction cost issues. Future studies may Williamson, and the participants at the Minnesota
identify the particular governance mechanisms Conference on Corporate Governance and the
that can attenuate the agency, bankruptcy and Academy of Management National Conference
transaction cost problems associated with the at Miami where earlier versions of this paper
strategy-capital structure linkage. Second, we were presented. The errors and omissions remain
have ignored in our study, noncommercial risks entirely ours.
that a firm may face in making strategic and
capital structure decisions-for example, risks
arising from changes in government tax policy
and other regulatory changes. The empirical REFERENCES
models used in this study can be extended to Akerlof, G. A. 'The market for 'lemons': Qualitative
look at the impact of changes in tax regimes uncertainty and the market mechanism', Quarterly
and/or the business cycle on capital structure and Journal of Economics, 84, 1970, pp. 488-500.
strategic investment decisions. Third, a natural Barney, J. 'Strategic factor markets: Expectations,
extension of this work is to investigate whether luck, and business strategy', Management Science,
32, 1986, pp. 1231-1241.
there are significant differences in slopes as well Barton, S. and P. Gordon. 'Corporate strategy and
as intercepts with a separate regression for each capital structure', Strategic Management Journal, 9,
firm. After all, if firm heterogeneity is the 1988, pp. 623-632.
 Strategy and Capital Structure 15
Bettis, R. 'Modern finance theory, corporate strategy, Long, M. and E. Malitz. 'Investment patterns and
and public policy: Three conundrums', Academy of financial leverage'. In B. Friedman (ed.), Corporate
Management Review, 8, 1983, pp. 406-415. Capital Structure in The United States, University
Bhagat, S., A. Shleifer and R. W. Vishny. 'Hostile of Chicago Press, Chicago, IL, 1985, pp. 325-348.
takeovers in the 1980s: The return to corporate Miller, M. 'The Modigliani-Miller proposition after
specialization', Brookings Papers: Microeconomics, thirty years', Journal of Applied Corporate Finance,
Brookings Institute, Washington, DC, 1990. 2, 1989, pp. 6-18.
Bradley, M., G. Jarrell and E. Kim. 'On the existence Modigliani, F. and M. Miller. 'The cost of capital,
of an optimal capital structure', Journal of Finance, corporation finance and the theory of investment',
39, 1984, pp. 857-878. American Economic Review, 48, 1958, pp. 261-297.
Brander, J. and T. Lewis. 'Oligopoly and financial Modigliani, F. and M. Miller. 'Taxes and the cost of
structure: The limited liability effect', American capital', American Economic Review, 53, 1963,
Economic Review, 76, 1986, pp. 956-970. pp. 433-443.
Brander, J. and T. Lewis. 'Bankruptcy costs and Montgomery, C. A. and B. Wernerfelt. 'Diversifi-
the theory of oligopoly', Canadian Journal of cation, Ricardian rents, and Tobin's q', Rand
Economics, 21, 1988, pp. 221-243. Journal of Economics, 19, 1988, pp. 623-632.
Caves, R. and T. Pugel. 'Firm size, financial structure Myers, S. 'Determinants of corporate borrowing',
and market structure: A study based on intraindustry Journal of Financial Economics, 9, 1977,
differences', Harvard Institute of Economic pp. 147-176.
Research, Discussion Paper #1129, 1985. Myers, S. and N. Majluf. 'Corporate finance and
DeAngelo, H. and R. Masulis. 'Optimal capital investment decisions when firms have information
structure under corporate and personal taxation', investors do not have', Journal of Financial Eco-
Journal of Financial Economics, 8, 1980, pp. 3-29. nomics, 13, 1984, pp. 187-221.
Dielman, T. E. Pooled Cross-sectional and Time Series Ross, S. 'The determination of financial structure: The
Analysis, Marcel Dekker, Inc., New York, 1989. incentive signalling approach', Bell Journal of
Flath, D. and C. Knoeber. 'Taxes, failure costs and Economics, 8, 1977, pp. 23-40.
optimal industry capital structure: An empirical Rumelt, R. 'How much does industry matter?',
test', Journal of Finance, 35, 1980, pp. 99-117. Strategic Management Journal, 12, 1991,
Grossman, S. and 0. Hart. 'Corporate financial pp. 167-185.
structure and management incentives'. In J. McCall Schmalensee, R. 'Do markets differ much?', American
(ed.), The Economics of Information and Uncer- Economic Review, 75, 1985, pp. 341-351.
tainty, University of Chicago Press, Chicago, IL, Scott, D. 'Evidence on the importance of financial
1982, pp. 107-141. structure', Financial Management, 2, 1972,
Hill, C. W. L. 'Internal capital market controls and pp. 45-50.
the financial performance of multidivisional firms', Scott, J. 'Corporate finance and market structure'. In
Journal of Industrial Economics, 37, 1988, R. Caves, M. Porter, M. Spence and J. Scott
pp. 67-83. (eds.), Competition in the Open Economy, Harvard
Hoskisson, R. and M. A. Hitt. 'Strategic control University Press, Cambridge, MA, 1980,
systems and relative R&D investment in large pp. 325-359.
multiproduct firms', Strategic Management Journal, Shapiro, C. 'Consumer information, product quality,
9, 1988, pp. 605-621. and seller reputation', Bell Journal of Economics,
Hsiao, C. Analysis of Panel Data, Cambridge Univer- Spring, 1982, pp. 20-35.
sity Press, Cambridge, MA, 1986. Shapiro, C. 'Premiums for high quality as returns
Jensen, M. 'Agency costs of free cash flow, corporate to reputation', Quarterly Journal of Economics,
finance and takeovers', American Economic Review, November, 1983, pp. 659-686.
76, 1986, pp. 323-329. Spence, M. 'Competitive and optimal responses to
Jensen, M. and W. Meckling. 'Theory of the firm: signals: An analysis of efficiency and distribution',
Managerial behavior, agency costs and ownership Journal of Economic Theory, 7, 1974, pp. 296-332.
structure', Journal of Financial Economics, 3, 1976, Spence, M. 'Capital structure and the corporation's
pp. 305-360. product market environment'. In B. Friedman
Kim, H. and J. McConnell. 'Corporate merger and (ed.), Corporate Capital Structure in The United
the co-insurance of corporate debt', Journal of States, University of Chicago Press, Chicago, IL,
Finance, 32, 1977, pp. 349-366. 1985, pp. 353-377.
Klein, B. and K. Leffler. 'The role of price guaranteeing Stiglitz, J. and A. Weiss. 'Credit rationing in markets
quality', Journal of Political Economy, 89, 1981, with imperfect information', American Economic
pp. 615-641. Review, 71, 1981, pp. 393-410.
Klein, B., R. Crawford and A. Alchian. 'Vertical Stiglitz, J. 'Some aspects of the pure theory of
integration, appropriable rents, and the competitive corporate finance: Bankruptcies and takeovers',
contracting process', Journal of Law and Eco- Bell Journal of Economics, 3, 1972, pp. 458-482.
nomics, 21, 1978, pp. 297-326. Titman, S. 'The effects of capital structure on the
Lippman, S. and R. Rumelt. 'Uncertain imitability: An firm's liquidation decision', Journal of Financial
analysis of interfirm differences under competition', Economics, 13, 1984, pp. 137-151.
Bell Journal of Economics, 13, 1982, pp. 418-439. Titman, S. and R. Wessels. 'The determinants of
16 S. Balakrishnan and I. Fox
capital structure choice', Journal of Finance, 43, talism, The Free Press, New York, 1986.
1988, pp. 1-19. Williamson, 0. 'Corporate finance and corporate
Turk, T. A. and R. E. Hoskisson. 'The strategic governance', Journal of Finance, 43, 1988,
implications of capital structure decisions', Working pp. 567-591.
Paper, Texas A&M University, November, 1991. Williamson, 0. 'Transaction cost economics'. In R.
Williamson, 0. Markets and Hierarchies: Analysis and Schmalensee and R. Willig (eds.), Handbook of
Antitrust Implications, The Free Press, New York, Industrial Organization, Vol. 1. North-Holland,
1975. Amsterdam, 1989, pp. 135-182.
Williamson, 0. The Economic Institutions of Capi-