See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/2510694

Agency Costs and Ownership Structure

Article  in  The Journal of Finance · March 1999

DOI: 10.1111/0022-1082.00201 · Source: CiteSeer

CITATIONS READS
1,670 15,990

3 authors, including:

James S Ang Rebel Allen Cole

Florida State University Florida Atlantic University
247 PUBLICATIONS   6,767 CITATIONS    127 PUBLICATIONS   7,538 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Corporate Governance View project

Corporate Governance in China View project

All content following this page was uploaded by Rebel Allen Cole on 06 December 2017.

The user has requested enhancement of the downloaded file.

THE JOURNAL OF FINANCE • VOL. LV, NO. 1 • FEBRUARY 2000

Agency Costs and Ownership Structure

JAMES S. ANG, REBEL A. COLE, and JAMES WUH LIN*

ABSTRACT
We provide measures of absolute and relative equity agency costs for corporations
under different ownership and management structures. Our base case is Jensen
and Meckling’s ~1976! zero agency-cost firm, where the manager is the firm’s sole
shareholder. We utilize a sample of 1,708 small corporations from the FRB0NSSBF
database and find that agency costs ~i! are significantly higher when an outsider
rather than an insider manages the firm; ~ii! are inversely related to the manag-
er’s ownership share; ~iii! increase with the number of nonmanager shareholders,
and ~iv! to a lesser extent, are lower with greater monitoring by banks.

THE SOCIAL AND PRIVATE COSTS OF AN AGENT’S ACTIONS due to incomplete align-
ment of the agent’s and owner’s interests were brought to attention by the
seminal contributions of Jensen and Meckling ~1976! on agency costs. Agency
theory has also brought the roles of managerial decision rights and various
external and internal monitoring and bonding mechanisms to the forefront
of theoretical discussions and empirical research. Great strides have been
made in demonstrating empirically the role of agency costs in financial de-
cisions, such as in explaining the choices of capital structure, maturity struc-
ture, dividend policy, and executive compensation. However, the actual
measurement of the principal variable of interest, agency costs, in both ab-
solute and relative terms, has lagged behind.
To measure absolute agency costs, a zero agency-cost base case must be
observed to serve as the reference point of comparison for all other cases of
ownership and management structures. In the original Jensen and Meckling
agency theory, the zero agency-cost base case is, by definition, the firm owned
solely by a single owner-manager. When management owns less than 100
percent of the firm’s equity, shareholders incur agency costs resulting from
management’s shirking and perquisite consumption. Because of limitations
imposed by personal wealth constraints, exchange regulations on the mini-
mum numbers of shareholders, and other considerations, no publicly traded
firm is entirely owned by management. Thus, Jensen and Meckling’s zero
agency cost base case cannot be found among the usual sample of publicly

* Ang is from Florida State University; Cole is from The University of Auckland, New Zea-
land; and Lin is from Montana State University. We appreciate the comments of David Mauer,
Michael Long, René Stulz ~the editor!, and an anonymous referee.

81
82 The Journal of Finance

traded firms for which information is readily available. The absence of in-
formation about sole owner-manager firms explains why agency costs are
often inferred but not directly measured in the empirical finance literature.
No-agency-cost base case f irms, however, can be found among non–
publicly traded firms. Until recently, data on non–publicly traded firms, which
tend to be much smaller than their publicly traded counterparts, have been
sparse. In 1997, the Federal Reserve Board released its National Survey of
Small Business Finances ~NSSBF!, which collected data from a nationally
representative sample of small businesses. Data from the NSSBF enable us
to analyze the relationship between agency costs and ownership structure
because the survey provides financial data on a group of firms whose man-
agement owns 100 percent of equity. These firms enable us to estimate the
expected expense for the no-outside-equity agency-cost base case. Further-
more, the database includes firms with a wide range of ownership and
manager0owner structures, including firms owned by two individuals as well
as firms managed by outsiders with no equity stake. As a consequence, small
firms appear well suited for a study of equity-related agency costs.
We use two alternative measures of agency costs. The first is direct agency
costs, calculated as the difference in dollar expenses between a firm with a
certain ownership and management structure and the no-agency-cost base
case firm. This measure captures excessive expenses including perk con-
sumption. To facilitate cross-sectional comparisons, we standardize expenses
by annual sales. Our second measure of agency costs is a proxy for the loss
in revenues attributable to inefficient asset utilization, which can result
from poor investment decisions ~e.g., investing in negative net-present-value
assets! or from management’s shirking ~e.g., exerting too little effort to help
generate revenue!. This second measure of agency costs is calculated as the
ratio of annual sales to total assets, an efficiency ratio. We can then mea-
sure agency costs as the difference in the efficiency ratio, or, equivalently,
the dollar revenues lost, between a firm whose manager is the sole equity
owner and a firm whose manager owns less than 100 percent of equity.
Monitoring of managers’ expenditures on perquisites and other personal
consumption relies on the vigilance of the nonmanaging shareholders and0or
related third parties, such as the company’s bankers. The lack of specific
operational knowledge on the part of nonmanaging shareholders, and the
lack of an external market for shares, however, may offset the presence of
dominant shareholders. Additionally, heavy reliance of the non–publicly traded
firms on bank financing could give banks a special role in delegated moni-
toring on behalf of other shareholders. Thus, it would seem that determina-
tion of the size of agency costs for these firms is an empirical issue.
Our results provide direct confirmation of the predictions made by Jensen
and Meckling ~1976!. Agency costs are indeed higher among firms that are
not 100 percent owned by their managers, and these costs increase as the
equity share of the owner-manager declines. Hence, agency costs increase
with a reduction in managerial ownership, as predicted by Jensen and Meck-
ling. These results hold true after controlling for differences across indus-
 Agency Costs and Ownership Structure 83

tries, the effects of economies of scale, and differences in capital structure.

We also find some evidence that delegated monitoring of small firms by
banks reduces agency costs.
The paper is organized as follows. In Section I, we discuss the nature of
equity agency costs in various ownership structures and explain the broad
outline of our empirical model. In Section II, we provide a description of the
data. We present results and analysis in Section III, followed by a summary
and conclusions in Section IV.

I. Agency Costs among Small Businesses

When compared to publicly traded firms, small businesses come closest to
the type of firms depicted in the stylized theoretical model of agency costs
developed by Jensen and Meckling ~1976!. At one extreme of ownership and
management structures are firms whose managers own 100 percent of the
firm. These firms, by their definition, have no agency costs. At the other
extreme are firms whose managers are paid employees with no equity in the
firm. In between are firms where the managers own some, but not all, of
their firm’s equity.
Agency costs arise when the interests of the firm’s managers are not aligned
with those of the firm’s owner~s!, and take the form of preference for on-
the-job perks, shirking, and making self-interested and entrenched decisions
that reduce shareholder wealth. The magnitude of these costs is limited by
how well the owners and delegated third parties, such as banks, monitor the
actions of the outside managers.
To illustrate, consider those firms where a single owner controls 100
percent of the stock but hires an outsider to manage the business. On the
one hand, agency costs may be small because the sole owner can internal-
ize all monitoring costs and has the right to hire and fire the manager.
More specifically, such an owner incurs 100 percent of the monitoring costs
and receives 100 percent of the resulting benefits. On the other hand, the
sole owner may not be able to monitor perfectly for the same reasons that
he or she hired an outside manager, such as lack of time or ability. Owners
of small firms typically lack financial sophistication, and may not be ca-
pable of performing random audits or fully understanding the operating or
financial results. Consequently, these firms incur residual agency costs. If
these costs are significant, they must ref lect a failure of the owner’s mon-
itoring activities. Potential explanations for this failure are lax monitoring
by the owners and the lack of an adequate monitoring technology available
for the owners. In this case, the separation of the management function
~initiation and implementation! versus the control function by nonmanag-
ing owners0shareholders ~ratification and monitoring!, as suggested by Fama
and Jensen ~1983a, 1983b!, may not be complete or effective. Thus, resid-
ual agency costs are still expected in a sole owner firm when the manager
is an outsider.
84 The Journal of Finance

Agency costs attributable to the divergence of interests vary inversely with

the manager’s ownership stake. As the number of shareholders increases
from one, the ownership of the owner0manager falls to a, where 0 # a , 1.
Because the manager gains 100 percent of each dollar spent on perks, but
only a percent of each dollar in firm profit, the manager who owns less than
100 percent of the firm has the incentive to consume perks rather than to
maximize the value of the firm to all shareholders. At the extreme is the
manager with zero ownership ~a 5 0!, who gains 100 percent of perquisite
consumption, but zero percent of firm profits ~in the case when salary is
independent of firm performance!.
Aggregate expenditure on monitoring by the nonmanaging shareholders
decreases as their individual ownership shares decline. This is due to the
well-known free-rider problem in spending for quasi-public goods, such as
monitoring effort. Each monitoring shareholder, with ownership l i must in-
cur 100 percent of the monitoring costs, but realizes only l i percent of the
monitoring benefits ~in the form of reduced agency costs!. A nonmonitoring
shareholder, however, enjoys the full benefits of a monitoring shareholder’s
activity without incurring any monitoring cost. Thus, as the number of non-
manager shareholders increases, aggregate expenditure on monitoring de-
clines, and the magnitude of owner-manager agency-cost problems increases.
Offsetting this relationship are concerns among shareholders about an in-
crease in the probability that the firm will be unable to pay off bank debt or
secure future financing from the same or new investors, which may produce
some restraint in agency behavior. However, as noted by Williams ~1987!,
these countervailing forces to agency behavior are expected to decline in
effectiveness when the firm is not in imminent danger of insolvency.
To summarize, against the null hypothesis that agency costs are indepen-
dent of the ownership and control structure,1 we postulate the following
hypotheses derived from agency theory when compared to the base case:
~i! agency costs are higher at firms whose managers own none of the firm’s
equity, ~ii! agency costs are an inverse function of the managers’ ownership
stake, and ~iii! agency costs are an increasing function of the number of
nonmanager shareholders.

II. Data
Our empirical approach utilizes two fundamental assumptions about agency
costs: ~1! A firm managed by a 100 percent owner incurs zero agency costs and,
~2! agency costs can be measured as the difference in the efficiency of an im-
perfectly aligned firm and the efficiency of a perfectly aligned firm. To opera-

1
Theoretical support for the null hypothesis is due to Demsetz ~1983!, who suggests that the
sum of amenities for on-the-job consumption and take-home pay for similar quality managers is
the same for both high-cost and low-cost monitoring organizations. The proportion paid to the
managers, however, differs according to the cost of monitoring. Here, it would seem that total
operating expense, which include direct pay to the managers as well as perks and firm level
monitoring cost, is the appropriate measure to test the hypothesis.
 Agency Costs and Ownership Structure 85

tionalize this approach for measuring agency costs, we need certain data inputs:
~i! data on firm efficiency measures; ~ii! data on firm ownership structure, in-
cluding a set of firms that are 100 percent owned by managers; and ~iii! data on
control variables, including firm size, characteristics, and monitoring technology.
Of these data requirements, the most demanding in terms of availability
is item ~ii! because sole-ownership firms typically are not publicly listed, and
because financial information on U.S. private firms usually is not available
to the public. The Federal Reserve Board’s National Survey of Small Busi-
ness Finances ~NSSBF!, fortunately, does provide financial information about
privately held firms, including their ownership structure, and does include a
set of firms entirely owned by managers. Consequently, we use data from
the NSSBF to measure agency costs.2
The NSSBF is a survey conducted by the Federal Reserve Board to gather
information about small businesses, which have largely been ignored in the
academic literature because of the limited availability of data. The survey
collected detailed information from a sample of 4,637 firms that is broadly
representative of approximately 5 million small nonfarm, nonfinancial busi-
nesses operating in the United States as of year-end 1992. Cole and Wolken
~1995! provide detailed information about the data available from NSSBF.
For this study, we limit our analysis to small C-corporations, collecting
information on the governance structure, management alignment, extent of
shareholder and external monitoring, size, and financial information. We
focus on corporations to minimize problems associated with the financial
statements of proprietorships, which typically commingle personal and busi-
ness funds. We eliminate partnerships and S-corporations because, unlike
C-corporations, they are not subject to corporate taxation, and this may lead
owner-managers to take compensation in the form of partner distributions
or dividends rather than salary expense because there is no double taxation
of such earnings at the firm level. By focusing solely on C-corporations, we
avoid the complications of comparing operating expenses across organiza-
tional forms. This restriction on the NSSBF database yields an analysis
sample of 1,708 firms.3

A. Agency Costs
To measure agency costs of the firm, we use two alternative efficiency
ratios that frequently appear in the accounting and financial economics
literature: the expense ratio, which is operating expense scaled by annual

2
Data from the NSSBF yield significant and interesting results that appear in several re-
cent published papers. See the studies on banking relationships and credit markets by Petersen
and Rajan ~1994, 1995, 1997!, Berger and Udell ~1995!, and Cole ~1998!.
3
The staff at the Federal Reserve Board partially edited the financial statement items for
violations of accounting rules, such as when gross profit is not equal to sales less cost of goods
sold, and some improbable events such as when accounts receivable are greater than sales, or
cost of goods sold equals inventory.
86 The Journal of Finance

sales,4 and the asset utilization ratio, which is annual sales divided by total
assets. The first ratio is a measure of how effectively the firm’s management
controls operating costs, including excessive perquisite consumption, and other
direct agency costs. More precisely, the difference in the ratios of a firm with
a certain ownership and management structure and the no-agency-cost base
case firm, multiplied by the assets of the former, gives the excess agency
cost related expense in dollars.
The second ratio is a measure of how effectively the firm’s management
deploys its assets. In contrast to the expense ratio, agency costs are in-
versely related to the sales-to-asset ratio. A firm whose sales-to-asset ratio is
lower than the base case firm experiences positive agency cost. These costs
arise because the manager acts in some or all of the following ways: makes
poor investment decisions, exerts insufficient effort, resulting in lower rev-
enues; consumes executive perquisites, so that the firm purchases unpro-
ductive assets, such as excessively fancy office space, office furnishing,
automobiles, and resort properties.
These efficiency ratios are not measured without error. Sources of mea-
surement error include differences in the accounting methods chosen with
respect to the recognition and timing of revenues and costs, poor record-
keeping typical of small businesses, and the tendency of small-business own-
ers to exercise f lexibility with respect to certain cost items. For example,
owners may raise0lower expenses, including their own pay, when profits are
high0low. Fortunately, these items are sources of random measurement er-
rors that may be reduced with a larger sample across firms in different
industries and age.

B. Ownership Structure
The corporate form of organization, with the limited-liability provision that
makes it more efficient for risk-sharing than proprietorships or partner-
ships, allows the firm to expand and raise funds from a large number of
investors.5 Thus, it has a richer set of ownership and management struc-
tures. The NSSBF provides four variables that we use to capture various

4
Operating expenses are defined as total expenses less cost of goods sold, interest expense,
and managerial compensation. Excessive expense on perks and other nonessentials should be
ref lected in the operating expenses. Strictly speaking, agency costs that are measured by this
ratio are those incurred at the firm level ~i.e., shirking and perquisite consumption by the
managers!. This may underestimate total agency costs since this ratio does not fully measure
firm-level indirect agency costs, such as the distortion of operating decisions due to agency
problems. ~See Mello and Parsons ~1992! for an attempt to measure such costs in the presence
of debt.! Nor does it measure off-income-statement agency costs, such as the private monitoring
costs by the nonmanagement shareholders or the private costs of bonding incurred by the manager.
5
Manne ~1967! and Alchian and Demsetz ~1972! agree that limited liability is an attractive
feature of the corporate form of organization. Jensen and Meckling ~1976! point out that al-
though unlimited liability gives more incentive for each shareholder to monitor, in the aggre-
gate it leads to excessive monitoring. Thus, it may be more economical to offer a single high
premium to creditors to bear risk of nonpayment and, thus, monitoring in exchange for limited
liability.
 Agency Costs and Ownership Structure 87

aspects of the ownership structure of small-business corporations: ~i! the

ownership share of the primary owner, ~ii! an indicator for firms where a
single family controls more than 50 percent of the firm’s shares, ~iii! the
number of nonmanager shareholders,6 and, ~iv! an indicator for firms man-
aged by a shareholder rather than an outsider.
According to theory, agency costs should be inversely related to the own-
ership share of the primary owner. For a primary owner who is also the
firm’s manager, the incentive to consume perquisites declines as his owner-
ship share rises, because his share of the firm’s profits rises with ownership
while his benefits from perquisite consumption are constant. For a primary
owner who employs an outside manager, the gains from monitoring in the
form of reduced agency costs increase with his ownership stake. Here, the
primary owner fulfills the monitoring role that large blockholders perform
at publicly traded corporations.
Agency costs should be lower at firms where a single family controls more
than 50 percent of the firm’s equity. At a small, closely held corporation
where a single family controls the firm, the controlling family also fulfills
the monitoring role that large blockholders perform at publicly traded cor-
porations. Due to more diffused ownership among older businesses with larger
families, however, monitoring by family members whose interests may not
always be aligned should be less effective than monitoring by a sole owner.
Agency costs should increase with the number of nonmanager sharehold-
ers. As the number of shareholders increases, the free-rider problem reduces
the incentives for limited-liability shareholders to monitor. With less moni-
toring, agency costs increase. Hence, we hypothesize that the expense and
asset-utilization ratios should be positively and negatively related to the
natural logarithm of one plus the number of nonmanaging shareholders,
respectively.7
Finally, agency costs should be higher at firms managed by an outsider.
This relationship follows directly from the agency theory of Jensen and Meck-
ling ~1976!. As noted above, this is the extreme case where the manager
gains 100 percent of perquisite consumption, but little of the firm’s profits.

C. External Monitoring by Banks

Banks play a pivotal role in small business financing because they are the
major source of external funds for such firms. Cole, Wolken, and Woodburn
~1996! report that more than 60 percent of the dollar amount of small busi-
ness credit outstanding takes the form of bank loans. Petersen and Rajan

6
Technically, the survey does not provide a variable for the number of nonmanager share-
holders. Rather, we define this variable as the number of shareholders for firms that have an
outside manager and as the number of shareholders less one for firms that have an insider
manager.
7
This formulation recognizes the unequal and diminishing role of additional shareholders,
and the problem of undefined zero when there is no other shareholder.
88 The Journal of Finance

~1994!, Berger and Udell ~1995!, and Cole ~1998! argue and present evidence
that firm-creditor relationships generate valuable information about bor-
rower quality.
Because banks generally require a firm’s managers to report results hon-
estly and to run the business efficiently with profit, bank monitoring com-
plements shareholder monitoring of managers, indirectly reducing owner-
manager agency costs. That is, by incurring monitoring costs to safeguard
their loans, banks lead firms to operate more efficiently by better utilizing
assets and moderating perquisite consumption in order to improve the firm’s
reported financial performance to the bank. Thus, lower priority claimants,
such as outside shareholders, should realize a positive externality from bank
monitoring, in the form of lower agency costs. Additionally, local bankers’
ability to acquire knowledge concerning the firms from various local sources,
such as churches, social gatherings, and interactions with the firm’s cus-
tomers and suppliers, makes them especially good monitors. We use two
variables to represent bankers’ incentive, cost, and ability to monitor: the
number of banks used by the firm and the length of the firm’s longest bank-
ing relationship.8
The bank’s cost of monitoring is proxied by the number of banks from
which the firm obtains financial services. The incentive for each bank to
monitor may decrease as the number of banks with which the firm deals
increases ~Diamond ~1984!!. Part of the reduced incentive to monitor is due
to a form of lenders’ free-rider problems, and part is due to the shorter
expected length of banking relationships when there is a greater perceived
likelihood of the firm switching its banking business between banks.
The bank’s ability to monitor is proxied by the length of a firm’s relation-
ship with its primary bank. A longer relationship enables the bank to gen-
erate information about the firm that is useful in deciding its creditworthiness
~Diamond ~1984!!. Both Petersen and Rajan ~1994! and Cole ~1998! find that
longer relationships improve the availability of credit to small firms while
Berger and Udell ~1995! find that longer relationships improve the terms of
credit available to small firms.
The bank’s incentive to monitor is proxied by the firm’s debt-to-asset ratio.
Because our sample consists entirely of small businesses, virtually all of the
firm’s debt is private rather than public, and the majority of this debt is in
the form of bank loans. As leverage increases, so does the risk of default by
the firm, hence the incentive for the lender to monitor the firm. While the
primary purpose of this monitoring is to prevent risk-shifting by sharehold-
ers to debtholders, increased monitoring should also inhibit excessive per-
quisite consumption by managers. ~Most of the sample firms’ nonbank debt

8
These are also the same governance variables used by Berger and Udell ~1995! and Cole
~1998! in their studies of banking relationship. However, none of their variables, except for the
corporate form dummy, are found to significantly affect either the loan term or the use of
collateral.
 Agency Costs and Ownership Structure 89

Figure 1. Operating expense-to-sales ratio by one-digit SIC for a sample of 1,708 small
corporations.

is in the form of loans from finance companies and other nonbank private
lenders, who also have greater incentive to monitor the firm as leverage
increases.!

D. Control Variables
We realize that the length of banking relationship variable may be corre-
lated with firm age, which in turn could be related to a firm’s efficiency. Due
to the effects of learning curve and survival bias, older firms are likely to be
more efficient than younger ones and, especially, than start-up firms. Hence,
we include firm age as a control variable in all our tests involving the vari-
able measuring the length of the firm’s relationship with its primary bank.
Both of our efficiency ratios vary widely across industries because of the
varying importance of inventory and fixed assets. Figure 1 shows the ratio
of operating expenses to sales by one-digit SIC. These ratios vary from a low
of 0.39 for construction and manufacturing to a high of 0.65 for finance and
real estate and professional services. Figure 2 shows the ratio of annual
sales to total assets by one-digit SIC. This efficiency ratio ranges from 3.6
for manufacturing to 6.2 for professional services. Hence, these figures un-
derscore the importance of controlling for differences across industries in
our analysis of agency costs. We do this by including a set of 35 dummy
variables, one for each two-digit SIC that accounts for more than one per-
cent of our sample of firms.
Small firms such as those surveyed by the NSSBF seem likely to realize
scale economies in operating expenses ~e.g.,overhead items!. Thus, there is a
need to control for firm size. This adjustment is especially important for
90 The Journal of Finance

Figure 2. Sales-to-asset ratio by one-digit SIC for a sample of 1,708 small corporations.

comparisons of operating expenses across firms where the difference in av-

erage size is of several orders of magnitude, as it is with the small busi-
nesses in our sample. Figure 3 confirms this, showing that the operating-
expense-to-sales ratio declines monotonically by sales quartile, decreasing
from 0.56 for the smallest quartile to 0.38 for the largest quartile. If we
regress the expense-to-sales ratio against annual sales, we find a negative
relationship that is statistically significant at better than the 1 percent level
~t 5 26.9!.
It is not clear, however, that efficiency in scale economies is realized as
measured by the ratio of sales to assets, where both the numerator and
denominator are popular alternative measures of size. Indeed, Figure 4 shows
that the ratio of sales to assets is higher for the two middle sales quartiles
than for either the largest or smallest quartile, suggesting, if any, a qua-
dratic relationship. When we regress the sales-to-asset ratio against sales
we find a positive but statistically insignificant relationship ~t 5 0.18!. Sim-
ilar results are obtained when the sales-to-asset ratio is regressed against
the natural logarithm of sales.

III. Results and Analysis

A. Some Preliminary Results Regarding the Separation
of Ownership and Control
We first examine how agency costs vary with the separation of ownership
and control—that is, whether the firm’s manager is a shareholder or an
outsider with no ownership stake. This analysis may offer some insights into
 Agency Costs and Ownership Structure 91

Figure 3. Operating expense-to-sales ratio by sales quartile for a sample of 1,708 small
corporations.

the effects of managerial alignment with owners on equity agency costs.9

Table I compares the agency costs of firms under two types of managers:
owners versus outsiders. Panel A shows results when agency costs are mea-
sured by the ratio of operating expenses to annual sales; Panel B shows
results when agency costs are measured by the ratio of annual sales to total
assets. It is important to note here and in all subsequent analyses that the
expected signs for the expense ratio and the asset utilization ratio are op-
posite to each other. Higher sales-to-assets ratios are associated with greater
efficiency and lower agency costs, whereas higher expense-to-sales ratios
are associated with less efficiency and higher agency costs.

A.1. Agency Costs as Measured by the Ratio of Operating

Expenses to Annual Sales
In Panel A of Table I, columns 2 and 3 show the number of observations
and the mean ~median! ratios of operating expenses ~which does not include
salary to managers!, to sales for firms whose manager is an owner. Columns
4 and 5 show the same information for firms whose manager is an outsider.
Consistent with our prior expectations, most small businesses are managed
by shareholders rather than by outsiders ~1,249, or 73 percent of the 1,708
sample firms!. However, there is not an insignificant number of firms that
hire outside managers ~459, or 27 percent of the sample!. Thus, there appear
to be a sufficient number of firms in these two groups for making meaning-
ful statistical comparisons of their operating expense ratios.
We find that both the median and average ratios of operating expenses to
annual sales are considerably higher for firms managed by outsiders ~col-
umn 5! than for firms owned by shareholders ~column 3!. For the full sample

9
In Fama and Jensen ~1983a!, the delegation of decision control management and residual
owner ~i.e., hiring of outsiders as managers! is related to the decision skill and the accompa-
nying specialized knowledge that are needed to run the firm. Shareholders, however, still have
to bear the costs of monitoring.
92 The Journal of Finance

Figure 4. Sales-to-asset ratio by sales quartile for a sample of 1,708 small cor-
porations.

~line 1 of Panel A!, the average ratios of operating expenses to assets at

insider-managed firms and outsider-managed firms are 46.9 percent and
51.9 percent, respectively; the 5.0 percentage-point difference in these means
is statistically significant at the 1 percent level.
Our data enable us to provide a rough estimate of the agency costs per
year attributable to the nonalignment of outside managers and sharehold-
ers. A back-of-the-envelope calculation shows that, in absolute dollars, a five-
percentage-point difference implies that the operating expenses at a firm
with median annual sales of $1.3 million are $65,000 per year higher when
an outsider rather than a shareholder manages the firm. The present values
of these residual equity agency costs are of course several times higher.10
Included in the full sample are 515 firms in which the primary owner
controls all of the firm’s equity. At 368 of these 515 firms, the owner also
serves as manager; at 147 firms, the owner employs an outsider as manager.
The former group fits the definition of our no-agency-cost base case, where
the manager owns 100 percent of the firm and the interests of manager and
owner are completely aligned. For the latter group, the interests of owner
and manager are completely unaligned. Thus, these groups are of interest
because they represent the two ends of the Jensen and Meckling’s spectrum
of ownership and managerial structures. Line 2 of Panel A in Table I shows
that the ratio of operating expenses to sales for the no-agency-cost base case
firm, where the manager owns 100 percent of the firm’s equity, is 46.4 per-
cent, as compared with 49.8 percent for firms whose owners hold all of the

10
As a way of comparison, Dong and Dow ~1993! estimate that 10 to 20 percent of total labor
hours are attributed to supervision or monitoring in the Chinese collective farms. Dobson ~1992!
finds that X-inefficiency measures 0.2 percent of sales among large U.S. manufacturing firms.
 Agency Costs and Ownership Structure 93

Table I
Agency Costs, Ownership Structure, and Managerial Alignment
with Shareholders
Agency costs are presented for a sample of 1,708 small corporations divided into two groups of
firms: those managed by owners ~aligned with shareholders! and those managed by an outsider
~not aligned with shareholders!. Agency costs are proxied alternatively by the ratio of operating
expenses to annual sales and the ratio of annual sales to total assets. Separate analyses are
presented for each agency cost proxy and for subgroups where the primary owner owns 100 per-
cent of the firm, where the primary owner owns more than half of the firm, where a single
family owns more than half of the firm, and where no owner or family owns more than half of
the firm. The last column shows the difference between the mean ~median! ratios of the outsider-
managed firms and the insider-managed firms. Statistical significance of the differences in the
mean ratios is based on the t-statistic from a parametric test ~based on the assumption of
unequal variances! of whether the difference in the mean ratios of the two groups of firms is
significantly different from zero. Statistical significance of the differences in the median ratios
is based on a chi-square statistic from a nonparametric test of whether the two groups are from
populations with the same median ~Mood ~1950!!. Data are taken from the Federal Reserve
Board’s National Survey of Small Business Finances.

Type of Manager

Owner-Manager Outsider-Manager Difference

Number Ratio Mean Number Ratio Mean in Means

of Firms ~Median! of Firms ~Median! ~in Median!

Panel A: Operating Expense-to-Annual Sales Ratio

All firms 1,249 46.9 459 51.9 5.0***

~42.0! ~52.2! ~10.2!***
Primary owner owns 368 46.4 147 49.8 3.4
100 percent of the firm ~41.7! ~47.6! ~5.9!**
Primary owner owns 743 46.8 258 49.6 2.8
.50 percent of the firm ~41.5! ~47.7! ~6.2!**
A single family owns 943 46.2 306 50.1 3.9**
.50 percent of the firm ~41.7! ~49.0! ~7.3!***
No owner or family owns 220 48.1 116 55.3 7.2**
.50 percent of the firm ~42.7! ~55.6! ~12.9!***

Panel B: Annual Sales-to-Total Assets Ratio

All firms 1,249 4.76 459 4.35 20.41*

~3.18! ~2.88! ~20.30!*
Primary owner owns 368 5.35 147 4.78 20.57
100 percent of the firm ~3.54! ~3.33! ~20.21!
Primary owner owns 743 5.08 258 4.49 20.59*
.50 percent of the firm ~3.33! ~3.13! ~20.20!
A single family owns 943 4.74 306 4.41 20.33
.50 percent of the firm ~3.19! ~3.07! ~20.12!
No owner or family owns 220 4.63 116 3.89 20.74
.50 percent of the firm ~3.14! ~2.49! ~20.65!**

*, **, *** indicate statistical significance at the 10, 5, and 1 percent levels, respectively.
94 The Journal of Finance

firms’ equity but hire an outside manager. For these two groups of firms,
the difference in operating expense ratios is 3.4 percentage points. Al-
though this univariate difference in means is not statistically significant, a
multiple regression model that corrects for size and industry effect, shown
in Table III later, indicates that firms hiring outside managers have oper-
ating expenses that are 5.4 percent greater than those at firms managed
by a shareholder.
Also included in the full sample are 1,001 firms in which the primary
owner holds a controlling interest of more than half of the firm’s equity. As
shown in Table I, Panel A, line 3, the ratio of operating expenses to sales for
these firms is 2.8 percentage points lower when the owner manages the firm
than when the owner hires an outside manager. However, this difference is
not statistically significant.
There are also 1,249 firms in which a single family holds a controlling
interest of more than half of the firm’s equity. As shown in line 4 of Panel A,
the average ratio of operating expenses to sales for these firms is 3.9 per-
centage points higher when the firm is managed by an outsider than when
the firm is managed by a shareholder. This difference is statistically signif-
icant at the 5 percent level.
One final group of interest is composed of 336 firms in which no person or
family holds a controlling interest of more than 50 percent of the firm’s
equity. As predicted, because of the more diffuse ownership of these firms,
the average ratio of operating expenses to sales is indeed much higher: 7.2 per-
centage points more at firms managed by outsiders than at firms managed
by shareholders. This difference is statistically significant at the 5 percent
level. To confirm that our finding is robust with respect to sample distribu-
tions, we also perform nonparametric tests on the difference between the
medians, and find similar results.

A.2. Agency Costs as Measured by the Ratio

of Annual Sales to Total Assets
In Panel B of Table I we present results from a similar analysis of agency
costs, but here we measure agency costs by the ratio of annual sales to total
assets rather than the ratio of operating expenses to annual sales.11 As pre-
dicted, the results show that the sales-to-asset ratios are higher in all cat-
egories of shareholder-managed firms versus outsider-managed firms. This
is true for the full sample of 1,249 firms ~line 1! and for the subsamples
where the primary owner holds all of the firm’s equity ~line 2!, where the
primary owner holds a controlling interest in the firm ~line 3!, where a sin-
gle family holds a controlling interest in the firm ~line 4!, and where no
individual or family owns more than half of the firm ~line 5!.

11
The distribution of the sales-to-asset ratio is highly skewed by the presence of large out-
liers. Consequently, the ratio is capped at the value found at the 95th percentile, a ratio of 19.0.
 Agency Costs and Ownership Structure 95

For the full sample, displayed in line 1, the average sales-to-asset ratio at
insider-managed f irms is almost 10 percent higher than at outsider-
managed firms at 4.76 and 4.35, respectively. The 0.41 difference in these
means is statistically significant at the 10 percent level. This difference im-
plies that the revenues of a median-size firm, which has $438,000 in total
assets, are $180,000 per year higher when a shareholder rather than an
outsider manages the firm. In each of the remaining four comparisons ~lines
2–5 of Panel B!, the average ratio of annual sales to total assets also is
greater when the firm is managed by a shareholder than when the firm is
managed by an outsider. However, this difference is statistically significant
at least at the 10 percent level only when the primary owner holds a con-
trolling interest in the firm ~line 3!.
Overall, the results displayed in Table I suggest that both the ratio of
operating expenses to annual sales and the ratio of annual sales to total
assets are adequate proxies for small corporations’ agency costs. Each pro-
vides results consistent with the predictions of agency theory for a wide
range of potentially high to low agency cost organizational and management
structures.

A.3. Determinants of High- and Low-Agency Cost Firms

Table II presents descriptive statistics for the variables hypothesized to
explain agency costs. Statistics are presented both for the entire sample
and for two groups of firms constructed by dividing the entire sample in
half, based on the sample’s median ratios of agency costs. For the entire
sample ~Panel A!, ownership and control is highly concentrated. On aver-
age, a shareholder manages the firm 73 percent of the time, the primary
owner controls 65 percent of the firm’s equity, and a single family owns a
controlling interest in the firm 73 percent of the time. The average number
of nonmanager shareholders is 3.51, but this statistic is strongly inf lu-
enced by extreme values, as the median number of nonmanager share-
holders is one. The average firm’s longest banking relationship is 10.6 years.
The average firm maintains relationships with 1.65 banks, reports $5.9
million in annual sales, is 17.6 years old, and has a debt-to-asset ratio of
0.60.
When we split the sample into low-expense and high-expense ratio groups
~Panel B!, we observe strong differences in the two groups. Based on t-tests
for significant differences in the means of the two groups, the high-expense
firms are less likely to be managed by a shareholder, are less likely to be
controlled by one family, have fewer nonmanaging shareholders, have shorter
and fewer banking relationships, report lower sales, and are younger than
the low-expense firms. Similar results are obtained when the top third and
bottom third of the sample are compared.
When we split the sample into low and high asset-utilization groups
~Panel C!, we also find strong differences in the two groups. Low-efficiency
firms are less likely to be managed by a shareholder, have lower percentage
 96
Table II
Descriptive Statistics for Variables Used to Analyze Agency Costs
Selected variables used to study agency costs at a sample of 1,708 small corporations are identified in column 1, the sample means and medians
appear in columns 2 and 3. In columns 4 and 5 ~columns 7 and 8! are the means for two groups of firms constructed by splitting the sample into
two equal-size groups of 854 firms based on the entire sample’s median operating expense-to-annual sales ratio ~annual sales-to-total asset ratio!.
Column 6 ~column 9! shows the difference in the two groups’ means, and the results from a t-test for significant differences in the means of the
low- and high-ratio groups of firms. Data are from the Federal Reserve Board’s National Survey of Small Business Finances.

Panel A Panel B Panel C

The Journal of Finance

Operating Expense-to-Sales Sales-to-Assets
Ratio Groups Ratio Groups

Below Above Below Above

Mean Median Median Median Difference Median Median Difference
~1! ~2! ~3! ~4! ~5! ~6! ~7! ~8! ~9!

Ownership variables
Firm manager is a shareholder 0.73 1 0.78 0.69 0.09*** 0.71 0.75 20.04*
One family owns .50 percent of the firm 0.73 1 0.75 0.71 0.04* 0.72 0.74 20.02
Ownership share of primary owner 0.65 0.54 0.66 0.64 0.02 0.62 0.68 20.06***
Number of nonmanager shareholders 3.51 1 3.89 3.14 0.75** 2.35 4.68 22.33***

External monitoring variables

Length of the longest banking relationship ~years! 10.6 8 11.3 10.0 1.3*** 11.5 9.7 1.8***
Number of banking relationships 1.65 1 1.75 1.56 0.19*** 1.73 1.58 0.15***
Debt-to-asset ratio 0.60 0.52 0.61 0.58 0.03 0.55 0.64 20.09***
Control variables
Annual sales ~$millions! 5.9 1.3 8.2 3.6 4.6*** 6.1 5.7 0.4
Firm age ~years! 17.6 14 18.4 16.8 1.6** 19.0 16.2 2.8***

*, **, *** indicate statistical significance at the 10, 5, and 1 percent levels, respectively.
 Agency Costs and Ownership Structure 97

ownership by the primary owner, have fewer nonmanaging shareholders,

have longer and more numerous banking relationships, have lower debt-to-
asset ratios, and are older than high-efficiency firms.

B. Multivariate Regression Results Explaining Agency Costs

Tables III and IV present the results obtained from estimating multivar-
iate regressions to explain the determinants of our two proxies for agency
costs, the ratio of operating expenses to annual sales and of annual sales to
total assets. Each proxy is regressed against the ownership, external mon-
itoring, and control variables introduced and discussed in Section II. These
regressions compare the relative, as well as the absolute, agency costs of
various ownership structures vis-à-vis the no-agency-cost base case—the
100 percent manager-owned firm.

B.1. Agency Costs as Measured by the Ratio of Operating

Expenses to Annual Sales
Table III presents the results from multivariate regressions analyzing agency
costs as measured by the ratio of operating expenses to annual sales. Col-
umn 1 identifies the explanatory variable and columns 2 through 9 display
parameter estimates for eight different model specifications. In columns 2
through 8 we analyze each of the seven ownership structures, external mon-
itoring, and capital structure variables independently. In column 9 we test
whether the independent results stand up when all seven variables are in-
cluded in a single regression. Because of the importance of industry struc-
ture and economies of scale, as established in Section II, we include in each
regression variables to control for firm size and industry effects. Our mea-
sure of size is the logarithm of annual sales, and our controls for industry
are 35 two-digit SIC indicator variables, one for each two-digit standard
industrial classification that accounts for more than one percent of our sam-
ple of firms.
In column 2 of Table III we find that a firm managed by a shareholder has
agency costs that are 5.4 percentage points lower than those at firms man-
aged by an outsider. This is very close to the 5.0 percentage point difference
reported for all firms in Panel A of Table I. For a firm with the $1.3 million
median annual sales, the coefficient in column 2 of Table III implies agency
costs of approximately $70,000.
In column 3 of Table III we find that a firm in which one family owns a
controlling interest has agency costs that are 3.0 percentage points lower
than other firms. For the median-size firm, this implies agency costs of ap-
proximately $39,000. In column 4, we find that agency costs decline by 0.082
percentage points for each percentage point increase in the ownership share
of the firm’s primary owner. This implies that a median-size firm where the
primary owner has a 100 percent share has agency costs that are approxi-
mately $105,000, or 8.1 percentage points, lower than those at a firm where
 98
Table III
Determinants of Agency Costs at Small Corporations
The dependent variable proxying for agency costs is the ratio of operating expense to annual sales. There are four groups of independent
variables: common ownership0managerial alignment variables, external monitoring variables, capital structure variables, and control variables.
Sample size is 1,708. Each specification includes a set of 35 dummy variables indicating each two-digit SIC that accounts for more than one
percent of the sample of firms. Data are from the Federal Reserve Board’s National Survey of Small Business Finances.

~1! ~2! ~3! ~4! ~5! ~6! ~7! ~8! ~9!

94.7*** 91.7*** 74.9*** 91.3*** 88.3*** 87.4*** 88.7*** 104.5***
Intercept ~17.4! ~16.9! ~17.1! ~17.2! ~16.9! ~16.6! ~16.9! ~17.3!

Ownership variables
25.4*** 25.7***

The Journal of Finance

Manager is a shareholder
~24.1! ~24.3!
One family owns .50 percent of the firm 23.0** 20.4
~22.2! ~20.3!
Ownership share of primary owner 28.2*** 28.6***
~23.9! ~22.8!
Log of the number of nonmanager stockholders 1.9*** 0.01
~2.8! ~0.1!
External monitoring variables
Length of the longest banking relationship 20.22*** 20.3***
~23.2! ~23.4!
Number of banking relationships 21.1* 21.0*
~21.9! ~21.7!
Debt-to-asset ratio 20.6 20.7
~20.6! ~20.7!
Control variables
Two-digit SIC dummies Yes Yes Yes Yes Yes Yes Yes Yes
Log of annual sales 22.9*** 23.1*** 23.1*** 23.1*** 22.6*** 22.6*** 22.8*** 22.9***
~28.8! ~28.6! ~29.0! ~28.8! ~27.8! ~27.4! ~28.3! ~27.9!
Firm age 0.03 0.02
~0.6! ~0.4!
Regression summary statistics
Adjusted R 2 0.246 0.241 0.245 0.242 0.244 0.240 0.239 0.259
F-statistic 16.04*** 15.63*** 16.00*** 15.73*** 15.50*** 15.57*** 15.46*** 14.57***

*, **, *** indicate statistical significance at the 10, 5, and 1 percent levels, respectively.
 Table IV
Determinants of Agency Costs at Small Corporations
The dependent variable proxying for agency costs is the ratio of annual sales to total assets. There are four groups of independent variables:
common ownership0managerial alignment variables, external monitoring variables, capital structure variables, and control variables. Sample
size is 1,708. Each specification includes a set of 35 dummy variables indicating each two-digit SIC that accounts for more than one percent of
the sample of firms. In column 10, the four ownership variables have been orthogonalized. Data are from the Federal Reserve Board’s National
Survey of Small Business Finances.

~1! ~2! ~3! ~4! ~5! ~6! ~7! ~8! ~9! ~10!

Agency Costs and Ownership Structure

2.01* 2.51** 1.27 1.27 2.49** 1.95* 3.26*** 1.34 1.17
Intercept ~1.9! ~2.4! ~1.2! ~1.2! ~2.5! ~1.9! ~3.2! ~1.2! ~21.0!

Ownership variables
Manager is a shareholder 0.51** 0.30 0.46*
~2.0! ~1.2! ~1.8!
One family owns .50% of the firm 0.087 20.34 20.22
~0.3! ~21.2! ~20.8!
Ownership share of primary owner 0.012*** 20.011* 0.012***
~3.0! ~21.8! ~3.1!
Log of the number of nonmanager 20.82*** 21.05*** 21.05***
stockholders ~26.2! ~25.4! ~25.4!
External monitoring variables
Length of the longest banking 20.025* 20.01 20.01
relationship ~21.7! ~20.4! ~20.4!
Number of banking relationships 21.09*** 20.50*** 20.50***
~24.7! ~24.3! ~24.3!
Debt-to-asset ratio 1.05*** 1.11*** 1.11***
~5.3! ~5.5! ~5.5!
Control variables
Two-digit SIC dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes
Log of annual sales 0.05 0.03 0.07 0.18*** 0.07 0.11* 0.02 0.28*** 0.28***
~0.7! ~0.5! ~1.1! ~2.7! ~1.1! ~1.7! ~0.3! ~3.9! ~3.9!
Firm age 20.012 20.01 20.01
~21.2! ~20.8! ~20.8!
Regression summary statistics
Adjusted R 2 0.032 0.030 0.035 0.051 0.035 0.042 0.045 0.080 0.080
F-statistic 2.51*** 2.40*** 2.65*** 3.49*** 2.63*** 3.02*** 3.20*** 4.37*** 4.37***

99
*, **, *** indicate statistical significance at the 10, 5, and 1 percent levels, respectively.
100 The Journal of Finance

the primary owner has only a one percent share. Each of the variables an-
alyzed in columns 2 through 4 is statistically significant at least at the
5 percent level.
In column 5 of Table III we analyze ~the natural logarithm of one plus! the
number of nonmanager shareholders. We expect a positive relationship be-
tween agency costs and this variable, as the returns to monitoring decrease
and free-rider problems increase with the number of nonmanager sharehold-
ers. We use the natural logarithm rather than the level of this variable
because we expect that the relationship is stronger at smaller values of the
variable. The estimated coefficient is positive and significant at better than
the one percent level, confirming our expectations. For a firm with 30 non-
manager shareholders, the maximum value imposed by our cap at the 95th
percentile, the estimated coefficient of 1.9 implies that agency costs are 6.5 per-
centage points higher, or $85,000 greater, than at a firm with zero nonman-
ager shareholders.
In columns 6 and 7 of Table III we analyze the two bank monitoring vari-
ables: the length of the firm’s longest banking relationship and the firm’s
number of banking relationships. As discussed in Section II, we expect agency
costs to vary inversely with the length of the longest banking relationship
and directly with the number of banking relationships. To distinguish be-
tween the private information generated by bank monitoring and the public
information generated by a firm’s durability, we also include firm age in the
specification analyzing the length of the firm’s longest banking relationship.
As shown in column 6 of Table III, agency costs are reduced by a statis-
tically significant 0.22 percent for each additional year in the length of the
firm’s longest banking relationship. The coefficient on firm age is not sig-
nificantly different from zero. In column 7, however, a related variable, the
number of banking relationships, is negative and statistically significant at
better than the 10 percent level. This finding conf licts with our hypothesis
in which multiple banking relationships reduce each bank’s incentive to mon-
itor, and, therefore, increase agency costs. One possible explanation recon-
ciling the two seemingly contradictory results is that the number of banking
relationships may proxy for factors other than the banks’ incentive to mon-
itor the firm. The most prominent explanations are the increasing financial
sophistication and maturity of the firms and their managers, and regulatory
limitations on loans to a single borrower, which may constrain a small bank’s
ability to supply funds to a larger firm.
In column 8 of Table III we analyze the complex relation between capital
structure and ownership on agency costs. As discussed in Section II, we ex-
pect an inverse relationship between agency costs and the debt-to-asset ra-
tio. We do, indeed, find a negative relationship, but the coefficient is not
significantly different from zero.
In each of the seven specifications displayed in columns 2 through 8 of
Table III, observe that our size variable, the natural logarithm of annual
sales, is negative and statistically significant at better than the 1 percent
level, which is strong evidence of economies of scale. Not shown in Table III
 Agency Costs and Ownership Structure 101

are statistics indicating that at least 20 of the 35 two-digit SIC indicator

variables included in each specification are statistically significant at least
at the 5 percent level. These findings underscore the critical importance of
controlling for differences across industries when examining the operating
expense-to-sales ratio. The adjusted R 2 for each of the seven specifications
in columns 2 through 8 indicates that the models explain approximately
one-quarter of the variability in the ratio of operating expenses to annual
sales.
Our final specification appears in column 9 of Table III, where we in-
clude each of four of the ownership variables, the two bank-monitoring
variables, and the capital structure variable, along with the control vari-
ables for firm size, age, and industrial classification. We find that each of
the four ownership variables has the predicted sign. However, only two of
the four—the indicator variable for shareholder-managed firm and the vari-
able for the ownership share of the primary owner—are statistically sig-
nificant, but each is significant at better than the 1 percent level. The
statistical insignificance of the other two ownership variables may be at-
tributable to the high correlation among the ownership variables. The sig-
nificant coefficients indicate that agency costs at a firm managed by a
shareholder are 5.7 percentage points lower than those at a firm managed
by an outsider, and that agency costs are reduced by 0.086 percentage
points for each percentage point increase in the primary owner’s ownership
share. This latter result supports the hypothesis that large shareholders
make more effective monitors ~Shleifer and Vishny ~1986! and Zeckhauser
and Pound ~1990!!.12
Both of the external monitoring variables are negative and significant,
just as they are in columns 6 and 7. The debt-to-asset ratio is negative but
not significantly different from zero, just as it is in column 8. Overall, the
results displayed in column 9 generally confirm the findings when the analy-
sis variables are examined independently in columns 2 through 8.

B.2. Agency Costs as Measured by the Ratio

of Annual Sales to Total Assets
Table IV displays the results from multivariate regressions analyzing
agency costs as measured by the ratio of annual sales to total assets. In
interpreting these results, it is important to remember that the sales-to-
asset ratio varies inversely with agency costs. As in Table III, column 1
identifies the explanatory variables and columns 2 through 9 display pa-
rameter estimates for different specifications of the regression model. In
columns 2 through 8, we analyze each of seven ownership structure, exter-

12
Fama and Jensen ~1983! realize that for shareholders to monitor the firm’s management
they must hold sufficient ownership; however, the cost of large ownership shares is suboptimal
risk-taking, and, possibly, underinvestment. Also, Demsetz ~1983! suggests that firms with con-
centrated ownership have lower monitoring costs.
102 The Journal of Finance

nal monitoring, and capital structure variables independently. In column 9,

we test whether the independent results stand up when all seven are in-
cluded in a single regression.
Because of the importance of industry structure, established in Section II,
we include in each regression a series of 35 two-digit SIC indicator vari-
ables, one for each two-digit standard industrial classification that accounts
for more than one percent of our sample of firms. We include the natural
logarithm of annual sales as a measure of size, because Figure 4 suggests a
possible quadratic relationship between sales and the ratio of sales-to-asset
ratio.
Column 2 of Table IV shows that a firm managed by a shareholder has a
sales-to-asset ratio that is 0.51 greater than that of a firm managed by an
outsider, and this coefficient is statistically significant at better than the
5 percent level. This evidence supports the hypothesis that agency costs are
higher when an outsider manages the firm. In column 3, we see that the
variable indicating those firms in which one family owns a controlling in-
terest has a coefficient that is positive but not significantly different from
zero. Column 4 shows that the coefficient on the ownership share of the
primary owner is positive and significant at better than the 1 percent level.
The coefficient indicates that the sales-to-asset ratio increases by 0.012 for
each percentage point increase in the ownership share of the firm’s primary
owner. This finding supports the hypothesis that agency costs decrease as
the ownership becomes more concentrated. Column 5 shows that the coeffi-
cient on ~the natural logarithm of! the number of nonmanager stockholders
is negative and statistically significant at better than the 1 percent level,
supporting the hypothesis that agency costs increase as the free-rider prob-
lem worsens.
In columns 6 and 7 of Table IV we analyze the two bank monitoring
variables: the length of the firm’s longest banking relationship and the
number of the firm’s banking relationships. Once again, to distinguish be-
tween the private information generated by bank monitoring and the pub-
lic information generated by a firm’s durability, we also include firm age in
the specification analyzing the length of the firm’s longest banking rela-
tionship. As shown in column 6, the length of the firm’s longest banking
relationship variable is inversely related to the sales-to-asset ratio, and is
statistically significant at better than the 10 percent level. This runs counter
to our hypothesis that agency costs are lower when a firm’s bank has had
more time to develop valuable private information about the firm. How-
ever, this variable is not significantly different from zero in the full spec-
ification shown in column 9. In column 7 we see that a related variable,
number of banking relationships, is negative and statistically significant
at better than the 1 percent level. This latter finding supports the hypoth-
esis that the values of a bank’s monitoring effort and private information
about a firm are dissipated when the firm obtains financial services from
multiple sources, but, on the whole, the results regarding the bank moni-
toring variables are ambiguous.
 Agency Costs and Ownership Structure 103

In column 8 we analyze the effect of capital structure on the sales-to-asset

ratio. The results indicate that firms with higher debt ratios have higher
sales-to-asset ratios, and that this relationship is statistically significant at
better than the 1 percent level. This finding is supportive of a version of the
theory put forth by Williams ~1987! that additional debt decreases agency
costs.
Not shown in Table IV are the results concerning the industry indicator
variables. In each specification, at least 20 of the 35 two-digit SIC indicator
variables are significant at the 5 percent level. Once again, this finding
underscores the critical importance of controlling for differences across in-
dustries when comparing agency costs.
Our final specification appears in column 9 of Table IV, where we include
each of the four ownership variables, the two monitoring variables, and the
capital structure variable, along with the control variables for firm size, age,
and industrial classification. We find that only two of the four correlated
ownership variables are statistically significant at better than the 10 per-
cent level. The natural logarithm of the number of nonmanager sharehold-
ers varies inversely with the sales-to-asset ratio and is significant at better
than the 1 percent level. The primary owner’s ownership share switches
from positive and significant at better than the 1 percent level in column 4
to negative and significant at better than the 10 percent level in column 9.
This counterintuitive finding may be attributable to the high correlation of
the ownership share variable with the log of number of nonmanager share-
holders ~ r 5 20.75!.
Both of the external monitoring variables are inversely related to the sales-
to-asset ratio, as they are in columns 6 and 7, when these variables are
analyzed independently. However, the length of the longest banking rela-
tionship variable no longer even approaches statistical significance ~t 5 20.3!,
but the number of banking relationships is significant at better than the
1 percent level. The debt-to-asset ratio remains positive and significant at
better than the 1 percent level, as it is in column 8. Overall, the results
displayed in column 9 tend to confirm the findings previously discussed when
the analysis variables are examined one-by-one in columns 2 through 8.
To test whether the correlations among the ownership variables are re-
sponsible for this finding, we orthogonalize the four ownership variables
and then reestimate the model specification appearing in column 9. The
results of this reestimation, which appear in column 10, confirm our suspi-
cions. Three of the four ownership variables are statistically significant at
least at the 10 percent level.13 The log of the number of shareholders re-

13
The sales-to-asset ratio is likely subject to additional biases that render it much noisier
than the operating expense-to-sales ratio. Note that the adjusted-R 2 statistics appearing at the
bottoms of Tables III and IV indicate that we are able to explain about 26 percent of the
variability in the latter ratio but only about eight percent of the variability in the former ratio.
This led us to investigate additional control variables in the sales-to-asset regression, including
the ratio of operating expenses to sales. Results from specifications including the operating
104 The Journal of Finance

mains negative and significant at better than the 1 percent level, but the
primary owner’s ownership share switches back from negative to positive
and also is significant at better than the 1 percent level. The dummy indi-
cating that the firm is managed by a shareholder also is positive, but is
significant at only the 10 percent level. The dummy indicating that a family
controls the firm is not significantly different from zero. In sum, the results
for the four ownership variables are not qualitatively different from those
reported in columns 2 through 5, when each of these variables is examined
independently, and provide strong support for the agency-cost theory of Jensen
and Meckling ~1976!.14

IV. Summary and Conclusions

In this article, we use data on small businesses to examine how agency
costs vary with a firm’s ownership structure. Because the managerial own-
ership of small firms is highly variable, with a range from zero to 100 per-
cent, we are able to estimate a firm’s agency costs across a wide variety of
management and ownership structures.15 By comparing the efficiency of firms
that are managed by shareholders with the efficiency of firms managed by
outsiders, we can calculate the agency costs attributable to the separation of
ownership and control.
We also examine the determinants of agency costs in a multivariate re-
gression framework and find that our results support predictions put forth
by the theories of Jensen and Meckling ~1976! and Fama and Jensen ~1983a!
about ownership structure, organizational form, and the alignment of man-
agers’ and shareholders’ interests. First, we find that agency costs are higher
when an outsider manages the firm. Second, we find that agency costs vary
inversely with the manager’s ownership share. Third, we find that agency
costs increase with the number of nonmanager shareholders. Fourth, we
also find that, to a lesser extent, external monitoring by banks produces a
positive externality in the form of lower agency costs.

expense-to-sales ratio as an additional regressor are not qualitatively different from those in
Table IV. In no case was the operating expense-to-sales ratio statistically significant at even the
20 percent level. Because we view the operating expense-to-sales ratio as an endogenous vari-
able, we also tested a specification that included a predicted value of this ratio rather than the
actual value. The predicted value was obtained using the model appearing in column 9 of Table III.
Again, the results from this robustness check are not qualitatively different from those appear-
ing in Table IV
14
For the sake of completeness, we also perform the same procedure on the regression equa-
tion in column 9 of Table III. We find that orthogonalizing the ownership variables does not
qualitatively affect the results. Only the same two ownership variables are statistically signif-
icant. Overall, although both measures of agency costs provide qualitatively similar results, the
expense ratio regression yields greater explained variations.
15
There are few empirical studies in related areas of corporate finance that analyze own-
ership, organizational, and management structures in detail. For example, see a study of ex-
ecutive compensation in Israel by Ang, Hauser, and Lauterbach ~1997!.
 Agency Costs and Ownership Structure 105

REFERENCES

Alchian, Armen A., and Harold Demsetz, 1992, Production, information costs, and economic
organization, American Economic Review 62, 777–795.
Ang, James S., Samuel Hauser, and Beni Lauterbach, 1997, Top executive compensation under
alternate ownership and governance structure, in Mark Hirschey and M. Wayne Marr, eds.:
Advances in Financial Economics ~JAI Press Inc., Greenwich, Conn.!.
Arnold, R. J., 1985, Agency costs in banking firms: An analysis of expense preference behavior,
Journal of Economics and Business, 103–112.
Barclay, Michael J., and Clifford W. Smith Jr., 1995, The maturity structure of corporate debt,
Journal of Finance 50, 609–632.
Berger, Allen N., and Gregory F. Udell, 1995, Relationship lending and lines of credit in small
firm finance, Journal of Business 68, 351–382.
Cole, Rebel A., 1998, The importance of relationships to the availability of credit, Journal of
Banking and Finance, forthcoming.
Cole, Rebel A., and Hamid Mehran, 1998, CEO compensation, ownership structure, and taxa-
tion: Evidence from small businesses, mimeo, Northwestern University.
Cole, Rebel A., and John D. Wolken, 1995, Sources and uses of financial services by small
businesses: Evidence from the 1993 National Survey of Small Business Finances, Federal
Reserve Bulletin 81, 629–667.
Cole, Rebel A., John D. Wolken, and Louise Woodburn, 1996, Bank and nonbank competition for
small business credit: Evidence from the 1987 and 1993 National Surveys of Small Busi-
ness Finances, Federal Reserve Bulletin 82, 983–995.
Demsetz, Harold, 1983, The structure of ownership and theory of the firm, Journal of Law and
Economics 26, 375–393.
Demsetz, Harold, and Kenneth Lehn, 1985, The structure of corporate ownership: Causes and
consequences, Journal of Political Economy 93, 1155–1177.
Denning, Karen C., and Kuldeep Shastri, 1993, Changes in organizational structure and share-
holder wealth, Journal of Financial and Quantitative Analysis 28, 553–564.
Diamond, Douglas W., 1984, Financial intermediation and delegated monitoring, Review of Eco-
nomic Studies 51, 393–414.
Dobson, John, 1992, Agency costs in U.S. manufacturing: An empirical measure using X-efficiency,
Journal of Economics and Finance 16, 1–10.
Dong, Xiao Y., and Gregory K. Dow, 1993, Monitoring costs in Chinese agricultural teams,
Journal of Political Economy 101, 539–553.
Fama, Eugene F., and Michael C. Jensen, 1983a, Separation of ownership and control, Journal
of Law and Economics 26, 301–325.
Fama, Eugene F., and Michael C. Jensen, 1983b, Agency problems and residual claims, Journal
of Law and Economics 26, 327–349.
Fama, Eugene F., and Michael C. Jensen, 1985, Organizational forms and investment decisions,
Journal of Financial Economics 14, 101–119.
Jensen, Michael C., 1986, Agency costs of free cash f low, corporate finance and takeovers,
American Economic Review 76, 323–329.
Jensen, Michael C., and William H. Meckling, 1976, Theory of the firm: Managerial behavior,
agency costs and capital structure, Journal of Financial Economics 3, 305–360.
Kim, Wi S., and Eric H. Sorensen, 1986, Evidence on the impact of the agency costs of
debt in corporate debt policy, Journal of Financial and Quantitative Analysis 21, 131–
144.
Knight, Frank H., 1921, Risk, Uncertainty and Profit ~University of Chicago Press, Chicago!.
Manne, Henry G., 1967, Our two corporate systems: Law and economics, Virginia Law Review
53, 259–284.
Mello, Antonio S., and John E. Parsons, 1992, Measuring the agency cost of debt, Journal of
Finance 47, 1887–1904.
Milgrom, Paul, and John Roberts, 1995, Economics, Organization, and Management ~Prentice
Hall, Englewood Cliffs, N.J.!.
 106 The Journal of Finance

Petersen, Mitchell A., and Raghuram G. Rajan, 1994, The benefits of lending relationships:
Evidence from small business data, Journal of Finance 49, 3–38.
Petersen, Mitchell A., and Raghuram G. Rajan, 1995, The effect of credit market competition on
lending relationships, Quarterly Journal of Economics 110, 407–442.
Petersen, Mitchell A., and Raghuram G. Rajan, 1997, Trade credit: Theories and evidence,
Review of Financial Studies 10, 661–691.
Shleifer, Andrei, and Robert Vishny, 1986, Large shareholders and corporate control, Journal of
Political Economy 94, 461–488.
Williams, Joseph, 1987, Perquisites, risk, and capital structure, Journal of Finance 42, 29–48.
Zeckhauser, Richard, and John Pound, 1990, Are large shareholders effective monitors? An
investigation of share ownership and corporate performance, in R. G. Hubbard, ed.: Asym-
metric Information, Corporate Finance, and Investment ~University of Chicago Press, Chicago!.

View publication stats