Financial Accountability & Management, 16(4), November 2000, 0267-4424

GOVERNANCE AND FUNDS ALLOCATION IN UNITED

STATES MEDICAL RESEARCH CHARITIES
Edward A. Dyl, Howard L. Frant and Craig A. Stephenson*

INTRODUCTION

How can boards of directors influence managers to act in the interest of the
organization, rather than in their personal interest? Much financial
economics literature has focused on answering this question for directors of
for-profit corporations. Less is known about governance arrangements in
nonprofit corporations. Although some researchers have advanced theories,
the finance literature contains little systematic, empirical work regarding
nonprofits. The organization-theory literature does pay considerable
attention to nonprofit boards, but as we will see, it focuses on the external role
of boards. Relatively few studies consider the internal control function of
boards.
Nonprofit organizations are of interest not only in themselves, as a large and
growing part of the economy, but also for what they may say about the
corporate control problem in general. Because their ownership is non-
tradable, nonprofits face neither monitoring by equity markets nor the threat
of takeover. Boards of directors are thus virtually only mechanism for
controlling nonprofit managers. In this respect, nonprofits, although unlike
publicly traded corporations, resemble other types of for-profit organizations
(e.g., privately-held or closely-held corporations).
We examine the relationship between governance structure and resource
allocation decisions in a sample of nonprofit corporations. Our sample
comprises medical charities that fund medical research and patient services.
We find strong evidence that board structure and composition make a
difference in the performance of the organization. In short, managerial self-
interest is an issue in nonprofits, as it is in for-profits.
The following section discusses the role of the board of directors in the
governance of for-profit and nonprofit corporations. We then describe our
sample of medical charities and report the results of regressions examining
the effect of various board characteristics on the corporations' funds allocation
decisions. We summarize our conclusions in the final section.
* The authors are respectively Christopher Sheafe/Estes Homes Professor of Finance at the University
of Arizona; Senior Lecturer in Political Science at Haifa University; and Assistant Professor of
Finance at Babson College. Authors' names are in alphabetical order.
Address for correspondence: Howard Frant, Department of Political Science, University of
Haifa, Mt. Carmel, Haifa 31905, Israel.
email: frant@poli.haifa.ac.il

ßBlackwell Publishers Ltd. 2000, 108 Cowley Road, Oxford OX4 1JF, UK
and 350 Main Street, Malden, MA 02148, USA. 335
336 DYL, FRANT AND STEPHENSON

BOARDS OF DIRECTORS AND CORPORATE GOVERNANCE

Boards and Management in the For-Profit Sector

Berle and Means (1932) are generally credited with raising the issue of
separation of ownership and control in corporations. They point out that,
given the then-recent tendency toward widely diffuse stock ownership,
. . .the position of ownership has changed from that of an active to that of a passive
agent. . . The owner is practically powerless through his own efforts to affect the
underlying property (p. 64).

Berle and Means consider the locus of control in corporations to rest in the
board of directors:
Since direction over the activities of a corporation is exercised by the board of
directors, we may say for practical purposes that control lies in the hands of the
individual or group who have the actual power to select the board of directors. . . (p.66).

Although a large part of their book discusses ways in which some stockholders
can act to the detriment of other stockholders, they also consider the
possibility that management could gain effective control through its influence
on the nominating committee that determines board composition:
Where ownership is sufficiently sub-divided, the management can thus become a self-
perpetuating body even though its share in the ownership is negligible. . . For the most
part, the stockholder is able to play only the part of the rubber stamp. Occasionally he
may have the opportunity to support an effort to seize control, a position not unlike
that of the populace supporting a revolution. In either case, the usual stockholder has
little power over the affairs of the enterprise. . . The separation of ownership and
control has become virtually complete (pp. 82^83).

As is clear from the quotation, Berle and Means had little faith in the ability
of shareholders to change management. But this option became more credible
as takeovers began to replace proxy fights. Manne (1965) explicitly put
forward the idea that takeovers may function as a management control device.
During the hostile takeover wave of the 1980s, it was widely believed that
managers were indeed being disciplined by market forces, making the issue of
control by the board less pressing.
Thus, Fama and Jensen (1983, p. 301) argue:
Since the takeover market provides an external court of last resort for protection of
residual claimants [i.e. stockholders], a corporate board can be in the hands of agents
who are decision experts. Given that the board is to be composed of experts, it is natural
that its most influential members are internal managers, since they have valuable
specific information about the organization's activities. . . The outside board members
act as arbiters in disagreement among internal managers and carry out tasks that
involve serious agency problems. . . for example, setting executive compensation. . .
(pp. 314^5, 316).

ß Blackwell Publishers Ltd 2000

 GOVERNANCE AND FUNDS ALLOCATION 337

Fama and Jensen suggest that including several top managers on the board is an
essential part of the way boards limit the discretion of managers. Top managers
constitute a `mutual monitoring system' because they possess knowledge about
the firm's operations that may not be available to outside directors and because
they are in competition with each other for career advancement.
Recently there has been renewed attention to the issue of whether boards
exert effective control over managers as institutional investors have become
concerned with corporate governance issues. A variety of proposals has been
advanced to improve the way in which corporate boards of directors monitor
managers. Two ideas gaining considerable currency are (1) to limit the number
of `insiders' on boards, and (2) to reduce the size of boards. Both these changes
are intended to reduce the chief executive's ability to dominate the board.
Lipton and Lorsch (1992, p. 59), for example, assert, `Corporate
governance in the United States is not working the way it should.' They
propose a number of changes; under the heading of `Board Size and
Composition' they state:
We believe that the size of a board should be limited to a maximum of ten directors
(indeed we would favor boards of eight or nine) with a ratio of at least two independent
directors to any director who has a connection with the company. . . A smaller board
will be most likely to allow directors to get to know each other well, to have more
effective discussions with all directors contributing, and to reach a true consensus from
their deliberations (pp. 67^68).
This proposal is seconded by Jensen (1993), who goes even further, and
reverses at least one of his and Fama's earlier prescriptions:
Keeping boards small can help improve their performance. When boards get beyond
seven or eight people they are less likely to function effectively and are easier for the
CEO to control. Since the possibility for animosity and retribution from the CEO is
too great, it is almost impossible for those who report directly to the CEO to participate
openly and critically in effective evaluation and monitoring of the CEO.Therefore, the
only inside board member should be the CEO (p. 865).
In the literature of financial economics, there has been considerable recent
interest in measuring empirically the effect of insider board membership on
firm performance and on monitoring of managers. Baysinger and Butler
(1985), for example, find evidence that firms with a higher percentage of
independent directors perform better in subsequent periods. Weisbach
(1988) examines the relation between CEO resignations, board composition,
and prior poor performance by the firm. He finds that poor performance is a
better predictor of resignations when boards have a significant proportion of
outsiders than when they are dominated by insiders. Weisbach also finds that
resignations following poor performance result in an increase in stock prices
when the board is dominated by outsiders, but not when it is dominated by
insiders. These results suggest that the stock market, at least, believes that
boards are more effective at disciplining management when they are not
insider-dominated.

ß Blackwell Publishers Ltd 2000

338 DYL, FRANT AND STEPHENSON

Hermalin and Weisbach (1988) find that, following poor stock

performance, insiders are more likely to leave the board, and outsiders are
more likely to join the board. Rosenstein and Wyatt (1990) present evidence
that stockholders look favorably on the appointment of outside directors,
finding that stock prices increase following the appointment of a new outside
director. Brickley, Coles and Terry (1994) examine the stock market reaction
to adoption of `poison pills' to make takeovers more difficult. They find a
positive stock price reaction when the board has a majority of outside
directors, but a negative one otherwise, suggesting that stockholders believe
that only boards of the former type are acting in their interest in these
situations. This line of research strongly suggests that the presence of outside
directors on the board inhibits self-interested behavior by managers.
Other studies have looked at more direct measures of managerial self-
interest, such as compensation. Gomez-Mejia et al. (1987) find that the
compensation of CEOs in for-profit firms is affected more by performance
when firms are owner-controlled than when they are manager-controlled.
Dyl (1988) and Santerre and Neun (1986) employ ownership concentration
as a proxy for effective monitoring of managers and report a negative
relationship between ownership concentration and CEO compensation in
for-profit firms. Sridharan (1996) finds that the salaries of CEOs are positively
related to two measures of CEO influence over the board ^ the proportion of
insiders on the board and whether the CEO is also chairman of the board.
A few researchers examine management compensation in situations where
the role of external capital markets is limited. Brickley and James (1987) find
that in states where takeovers are restricted, a greater proportion of outsiders
on the board of directors is associated with a reduction in excess salary
expenditures. In states where takeovers are not restricted, they find no excess
salary expenditures, and interpret this finding as evidence that outside
directors and takeovers serve as substitute control mechanisms. Mayers,
Shivdasani, and Smith (1997) examine mutual insurance companies. These
organizations are customer-owned; they do not have tradable shares, and thus
lack external equity markets. This study finds that mutual insurance
companies have more outside directors than stock insurance companies. It
also finds that among mutuals, but not stock companies, those with a higher
proportion of outside directors spend less on a combined measure of salaries,
wages and rent. They find no such effect for other expenses. The authors
interpret this as evidence that outside directors play a more important role as
a control mechanism when other mechanisms, such as capital market
monitoring, are absent.
There is, then, substantial evidence that boards dominated by insiders
behave differently from those dominated by outsiders in for-profit
corporations. The prescriptions of Lipton and Lorsch (1992) and Jensen
(1993) regarding board size, however, rest on an empirical foundation that,
until recently, was virtually nonexistent. Yermack (1996) reports strong

ß Blackwell Publishers Ltd 2000

 GOVERNANCE AND FUNDS ALLOCATION 339

evidence that the size of a board affects the performance of for-profit

corporations. He finds that the market places a higher value on companies
with smaller boards, that stock prices go up when companies announce
reductions in board size, and that CEO compensation is more strongly linked
to performance when boards are smaller.
The organization-theory literature also deals with the issue of board
composition. In general, this literature, in the words of Middleton (1987, p.
141) `derives primarily from researchers interested in boards of directors as a
mechanism that organizations can use to deal with uncertainties in their
external world', and does not focus on boards' control of managers.
Researchers working in this framework, however, have obtained some results
pertinent to insider versus outsider representation on boards and to board size.
In an early study of 103 companies in twelve industries, Vance (1964) finds
that insider-dominated boards are more effective than outsider-dominated
boards. Pfeffer (1972) studies organization-environment linkages in 80
corporations and finds that firms with greater dependency on external capital
(as measured by their debt-equity ratios) have more outsiders on their boards
and tend to have larger boards. He also finds that those firms that deviate the
most from the empirically predicted proportion of insiders and outsiders
perform relatively poorly, and interprets this finding as evidence that the
insider/outsider ratio is a `rational organizational [response] to the conditions
of the external environment' (p. 226).

Boards and Management in Nonprofit Corporations

To what extent do the above results apply to the nonprofit case? Like mutuals,
nonprofits are not disciplined by an external capital market. Also, the boards
of directors of nonprofits are, in most cases, self-perpetuating bodies, not just
de facto ^ the situation that worried Berle and Means ^ but legally.
These factors suggest that managers in the nonprofit sector may have
greater scope to abuse discretion and pursue self-interest than managers in
the for-profit sector. On the other hand, self-selection by managers might
work in the opposite direction; individuals with primarily selfish motivations
may be less likely to gravitate toward work in nonprofits. Still, recent
examples of managerial excesses in the nonprofit sector, such as the substantial
management perquisites at United Way in the US, point out the need to
monitor managers in nonprofit corporations. The research on the for-profit
sector raises the question: Does board structure in the nonprofit sector affect
the monitoring and control of managers' activities?
Fama and Jensen (1983) attempt to develop a theory of organizational
governance that encompasses a wide variety of organizations, including for-
profit corporations and nonprofits (as well as financial mutuals and
professional partnerships). Discussing the role of boards in nonprofit
corporations, they say:

ß Blackwell Publishers Ltd 2000

340 DYL, FRANT AND STEPHENSON

. . . because a nonprofit lacks alienable residual claims [i.e., shares] the decision agents
are immune from ouster (via takeover) by outside agents.Without the takeover threat
and the discipline imposed by residual claimants with the right to remove members of
the board, nonprofit boards composed of internal agents and outside experts chosen by
internal agents would provide little assurance against collusion and expropriation of
donations.Thus, nonprofit boards generally include few if any internal agents as voting
members . . . (p. 319).
Williamson (1983,p. 359) comments that Fama and Jensen:
. . .have the beginnings of a predictive theory of the performance of nonprofits based on
the composition and character of the board of directors (trustees). For example, I
would predict on their reasoning that the expense to payout ratio will be higher in
nonprofits where the proportion of insiders on the board is high. . .
Financial economists to date, however, have not taken up the challenge of
testing the accuracy of this prediction.
In contrast, the literature in organization theory and sociology has featured
empirical work on nonprofit as well as for-profit boards. Again, however, this
work focuses on the issue of the external, boundary-spanning role of the board.
For example, Pfeffer (1973), in a study of 57 hospitals (almost all nonprofit)
finds that hospitals that depend more on local resources are more likely to
select board members for their linkages to the outside rather than for
management ability, and that their boards tend to be larger. He reports a
negative correlation between the presence of a board selected for management
ability and growth rates of facilities and budget.
When it comes to examining the internal role of boards, and in particular
the issue of board-manager relations, the empirical work is, in Middleton's
(1987, p. 150) words, `scant.' There has been some work on factors affecting
the relative power of boards and managers, with Zald (1969) asserting on
theoretical grounds that, for instance, boards will have less power in highly
complex and technically oriented organizations. Herman and Tulipana
(1989) find weak statistical evidence that nonprofits where the manager is
more influential are rated more effective (by staff and board members), and
that board influence is unrelated to ratings of effectiveness.
Provan (1988) examines CEO ratings of board influence over a variety of
internal management issues, but mainly seeks to relate them to characteristics
of the board's external environment. He does include CEO membership on
the board as an independent variable in regression equations, hypothesizing
that boards with CEO membership will be less influential, but fails to find
any effect.

Hypotheses
Neither the finance literature nor the organization-theory literature, then,
seems to have considered empirically the question of whether the structure of
nonprofit boards affects their ability to control managerial discretion. We

ß Blackwell Publishers Ltd 2000

 GOVERNANCE AND FUNDS ALLOCATION 341

address this issue by examining the relationship between board structure and
spending decisions for a sample of nonprofit corporations. If board structure
matters in the nonprofit sector in ways similar to those demonstrated in the
for-profit sector, then there ought to be observable consequences of different
structures, just as there are in the for-profit sector.
The discussion above suggests several hypotheses. First is what we might
call the `Williamson-Fama-Jensen hypothesis' ^ that when there is a higher
proportion of insiders on the board, we will observe a higher ratio of
management expenses to total spending. Second, by the same logic, insiders on
the board should result in higher executive compensation.
Third, we examine balance-sheet equity, known in nonprofit accounting as
`fund balance' or `net assets.' While growth in equity is a positive sign in for-
profit corporations, nonprofits are supposed to expend their resources to
promote the goals of their donors. Retention of equity beyond a prudent
minimum may indicate that managers are trying to promote their own security
at the expense of the organization's stated goals. This phenomenon is analogous
to the `free cash flow' problem in for-profit corporations (see Jensen, 1986). If
managerial representation on the board makes it easier for managers to pursue
their self-interest, one might hypothesize that fund balance will be positively
associated with managerial representation. Finally, if the literature on for-
profit boards applies to nonprofits, we may expect that increased board size will
have effects on all these variables in the same direction as the effect of
managerial representation, since both imply weaker board oversight.

SAMPLE SELECTION AND DESCRIPTION

Our sample of nonprofit corporations consists of national medical research

charities located in the United States, identified from the 1991 Encyclopedia of
Associations. Because nonprofits differ widely in their missions and the nature of
their activities, we limit our sample to a single, organizationally simple, type
of nonprofit organization in order to reduce the number of confounding
factors. We examine nonprofit organizations that raise funds for medical
research and patient services. We identified eighty-four such medical research
charities and sent them letters requesting copies of their most recent annual
report or Form 990 annual information return filed with the Internal Revenue
Service for the fiscal year ending in 1991. After follow-up letters and telephone
calls, we received a total of 58 responses (69 percent). The remaining
organizations in the sample either chose not to respond, or simply failed to
respond, to our requests for information.1
Upon closer examination, four organizations out of the 58 responding
appeared to violate our goal of examining a homogeneous sample of nonprofit
organizations. Three of the organizations operate large charitable hospitals
rather than functioning purely as a conduit of funds to medical research, and

ß Blackwell Publishers Ltd 2000

342 DYL, FRANT AND STEPHENSON

were therefore eliminated from the sample.2 In addition, the smallest charity
among the respondents had annual revenue of only US$13,692, one-ninth the
annual revenue of the second smallest charity in the sample. One observation
so different from the rest of the sample could have an undue influence on the
study's results, and in fact, several coefficients that were statistically
significant when this observation was included became insignificant once it
was excluded, suggesting the original results were a statistical artifact. This
observation was, therefore, also eliminated from the sample, leaving a final
sample of 54 relatively comparable medical charities. The largest
corporations in the sample are the American Cancer Society and the
American Heart Association, with annual revenues of US$372 million and
US$288 million respectively. The smallest organizations are the National
Reye's Syndrome Foundation and the Rheumatoid Disease Foundation, with
annual revenues of US$190,000 and US$119,000.
An interesting, and potentially important, characteristic of the final sample
involves monitoring by the National Charities Information Bureau (NCIB).
The NCIB is an organization that evaluates and monitors selected nonprofits
according to nine basic standards, to give prospective donors `. . .reasonable
assurance of public service by organizations to which they make their
contributions.'3 The standards promulgated by the NCIB recommend lower
bounds on program spending and upper bounds on fund-raising expenses and
fund balances. By actively monitoring the operations of a subset of national
nonprofit corporations, the NCIB is, in a sense, a part of the governance
structure of these organizations. Organizations listed by the NCIB, which
tend to be larger and better-known than those not listed, may feel additional
pressure to make their financial practices conform to standards. Our analysis
therefore controls for NCIB listing

ANALYSIS

Dependent Variables: Measures of Resource Allocation

The first three dependent variables described in this section are measures of
how the organization allocates its resources. They are the proportion of total
expenditures devoted to program activities, the proportion devoted to fund-
raising, and the proportion devoted to management and general expenses.
These three measures of the organization's resource allocation decisions sum
to one, and so cannot all vary independently, but any two could be
independent of each other. In organizations where managers are acting in
their own interests and counter to organizational interests, we expect, in
accord with the Williamson-Fama-Jensen hypothesis, that management and
general expenses would be a larger proportion, and program expenses a
smaller proportion, of the total.

ß Blackwell Publishers Ltd 2000

 GOVERNANCE AND FUNDS ALLOCATION 343

Earlier in the paper, we also formulated hypotheses concerning

management salaries and fund balances. We operationalize the former
variable as the highest executive salary in the organization, and measure fund
balance as the total net assets of the charity at the end of the 1991 fiscal year
divided by the total expenditures in 1991 ^ yielding a relative measure that
can be interpreted as the number of years' expenditures that could be covered
by the current fund balance.

Independent Variables
In line with the research on for-profits discussed above, our analysis focuses on
managerial representation on the board of directors. In our sample, Fama and
Jensen's (1983) assertion that there are `few, if any' managers on the boards of
nonprofits is correct. None of the medical research charities we examine had
more than one manager ^ the executive director ^ on the board of directors.
This state of affairs stands in marked contrast to boards of for-profit
organizations, where commonly many of the firm's top managers are also
members of the board of directors. Clearly, the executive director alone does
not have enough voting power to control the board, so none of the boards in
our sample is dominated by insiders in a numerical sense. The executive
director does, however, have significant informational advantages that may
result in disproportionate influence, in Sridharan's (1996) sense, on board
decisions. Moreover, the presence of the executive director in the boardroom
may have an inhibiting effect on free discussion of policy issues by other board
members. One of our independent variables, therefore, is a dichotomous
variable with a value of 1 if the executive director is a board member and 0 if
not.
Because proposals for reform in the for-profit sector often include limits on
the number of directors, we also examine the effect of board size on our
dependent variables. Intuitively, it is implausible that the effect is linear; the
difference between a five-member board and a ten-member board is unlikely
to be the same as the difference between a thirty-member board and a thirty-
five-member board. We therefore employ a logarithmic measure of board size,
which implies roughly that equal percentage increases in board size have equal
effect on the dependent variable. This choice of specification is supported by
the empirical findings of Yermack (1996), who reports a logarithmic
relationship between board size and performance in for-profit corporations.
If, as we suggested above, the executive director may have disproportionate
influence on the board, then the dynamics of board meetings may be different
when the director is present. Accordingly, we include an interaction term in
the regression analyses to allow the effect of board size to vary with insider
representation.
Finally, we control for two other organizational characteristics. The first is
size, measured as the organization's annual revenues. If economies of scale

ß Blackwell Publishers Ltd 2000

344 DYL, FRANT AND STEPHENSON

exist for medical charities, as suggested by Hyndman and McKillop (1999),

we expect to observe different patterns of proportional expenditures in large
organizations than in small ones. For reasons similar to those discussed for
board size, we use the natural logarithm of revenues to measure this variable.
The second is NCIB listing, as discussed above. Coverage by the NCIB may
itself influence outcomes (since those listed are presumably aware that they
are being reported on), and listing by the NCIB may also serve as a proxy for
various unmeasured characteristics of nonprofit organizations (e.g., a
willingness to undergo external scrutiny).4

Summary Statistics
Descriptive statistics for the 54 medical research charities in our sample are
shown in Table 1 for the overall sample and categorized according to NCIB
listing and according to whether or not the executive director serves on the
board of directors. The average medical charity in our sample had assets of
US$31.1 million and revenues of US$25.2 million in 1991. Total expenditures
are allocated as follows: 77.5 percent to program activities, 14.3 percent to

Table 1
Descriptive Statistics for 54 Medical Research Charities

NCIB Listed Executive is on Board

Variable Total Sample Yes No Yes No

Number of Observations 54 30 24 14 40
Total Assets $31.1 $56.2 $5.8 $13.3 $42.3
Annual Revenue $25.2 $40.6 $6.0 $16.7 $28.2
Program Expenditures $18.9 $30.4 $4.5 $13.3 $20.2
(77.5%) (77.2%) (77.6%) (78.7%) (74.5%)
Fund Raising $3.5 $5.8 $0.6 $2.2 $3.8
Expenditures (14.3%) (14.7%) (10.3%) (13%) (14.4%)
Mgt. & Gen. $2.0 $3.1 $0.6 $1.3 $2.3
Expenditures (8.2%) (7.9%) (10.3%) (7.7%) (8.5%)
Total $24.4 $39.4 $5.8 $16.9 $27.1
Expenditures (100%) (100%) (100%) (100%) (100%)

Notes:
Mean values ($ million) are shown for the total sample and for the sample categorized according
to coverage in the NCIB's Wise Giving Guide and according to whether or not the executive director
serves on the board of directors. The numbers in parentheses are the percentage of total expendi-
tures represented by each expenditure category.

ß Blackwell Publishers Ltd 2000

 GOVERNANCE AND FUNDS ALLOCATION 345

fund-raising, and 8.2 percent to management and general expenses. Thirty of

the 54 medical charities are monitored by the NCIB. NCIB charities are much
larger than the others, allocate a greater proportion of their resources to fund-
raising, and allocate a smaller proportion to management and general
expenses. The charities where the executive director serves on the board are
smaller than those where he does not, but there are no major differences
between the two groups with regard to the proportion of their expenditures
in each category.
We have data on management salaries for 43 of the charities in our sample.
We focus on the highest management salary in the organization. The median
annual salary is US$78,898, with a mean of US$90,305 and a standard
deviation of US$50,605. The salaries range from a high of US$208,231 for
the American Cancer Society to lows of zero for the Dysautonomia
Foundation and the National Reyes Syndrome Foundation. Although these
latter two charities do have paid employees, apparently the managers serve
without remuneration. These two charities are relatively small, with annual
expenditures of US$693,247 and US$177,991 respectively. When these two
charities are excluded, the median and mean salaries are US$83,500 and
US$94,710 respectively.
We measure fund balance as the charity's net assets divided by its annual
expenditures. The median (mean) fund balance of the 54 charities in our
sample represents 0.44 (0.74) years' expenditures.

Regression Results
Results of regressions examining the relationship between financial
performance and various governance variables and other organizational
characteristics are shown in Table 2. Panel A shows results for ordinary least
squares regressions, using White (1980) heteroskedasticity-consistent
standard errors.5
The regressions with program expenditures and fund-raising expenditures
as the dependent variables reveal significant effects for two variables ^
executive director membership on the board, and the interaction term
between board size and executive director membership on the board. Having
an executive on the board significantly decreases program expenditures and
significantly increases spending on fund-raising. When an executive is on the
board (and only then), however, these effects are to some degree counteracted
by increased board size. The size of the board is not significant, except when an
executive serves on the board. Stated differently, having an executive on the
board is associated with decreased spending on the charity's medical support
program and increased spending on fund-raising, but only when the board is
small. The board size at which the interactive variable offsets the effect of the
executive director being on the board is roughly 23 for the program activity
and fund-raising regressions in Table 2.

ß Blackwell Publishers Ltd 2000

346 DYL, FRANT AND STEPHENSON

Table 2
Expenditures Regressed on Governance Variables and Other Variables

Dependent Total Listed by Exec. on Board Exec. x R2(adj.)

Variable Intercept Revenue NCIB Board Size Board Size [N]

Panel A. Ordinary Least Squares Regressions

Program 0.813 ÿ0.0024 0.011 ÿ0.417 ÿ0.003 0.138 0.090
Activity (8.62)** (ÿ0.42) (0.42) (ÿ3.13)** (ÿ0.15) (3.44)** [54]
Fund ÿ0.085 0.0094 0.022 0.457 0.0097 ÿ0.145 0.232
Raising (ÿ1.49) (2.07)* (1.23) (4.95)** (0.59) (ÿ4.82)** [54]
Mgt./Gen. 0.272 ÿ0.007 ÿ0.032 ÿ0.040 ÿ0.0066 0.007 0.091
Expenses (4.64)** (ÿ1.69) (ÿ1.61) (ÿ0.58) (ÿ0.42) (0.33) [54]
Panel B: Ordinary Least Squares Regressions With Influential Outliers Deleted
Program 0.880 0.0005 ÿ0.023 ÿ0.388 ÿ0.025 0.124 0.092
Activity (10.10)** (0.07) (ÿ1.02) (ÿ3.01 )** (ÿ1.13 ) (2.97)** [51]
Fund ÿ0.099 0.0073 0.035 0.387 0.020 ÿ0.121 0.174
Raising (ÿ1.17) (1.05) (1.55) (3.09)** (0.90) (ÿ2.98)** [51]
Mgt./Gen. 0.219 ÿ0.0078 ÿ0.011 0.0007 0.006 ÿ0.003 0.020
Expenses (3.69)** (ÿ1.60) (ÿ0.71) (0.01) (0.37) (ÿ0.11) [51]
Panel C: Least Absolute Deviations Regressions
Program 0.905 ÿ0.00002 ÿ0.015 ÿ0.531 ÿ0.029 0.167 0.049
Activity (16.01)** (ÿ0.32) (ÿ0.79) (ÿ5.79)** (ÿ1.91) (5.42)** [54]
Fund ÿ0.162 0.015 0.0009 0.390 0.006 ÿ0.119 0.201
Raising (ÿ3.46)' (3.70)** (0.06) (5.13)** (0.46) (ÿ4.63)** [54]
Mgt./Gen. 0.198 ÿ0.007 ÿ0.018 ÿ0.023 0.007 0.005 0.159
Expenses (5.42)** (ÿ2.19)* (ÿ1.43) (ÿ0.39) (0.68) (0.24) [54]

Notes:
The dependent variables are measured as a percentage of total expenditures. Total revenues and
board size are expressed as natural logarithms. A 0,1 indicator variable denotes whether the char-
ity is listed by the NCIB and/or if the executive director serves on the Board (yes = 1). The number
of observations is shown in brackets in the last column [N]. The numbers in parentheses are t-sta-
tistics. The t-statistics in Panels A and B are calculated using White's (1980) heteroskedasticity-
consistent method.
** Significant at the 0.01 level.
* Significant at the 0.05 level.

The finding of Pfeffer (1973) that large boards are likely to be more oriented
toward fund-raising is not borne out here. Possibly, larger boards do more
volunteer fund-raising and thus have less need to expend organizational
resources on fund-raising activities. This conjecture fails to explain, however,
why the coefficient on board size is insignificant except when an executive is on
the board.

ß Blackwell Publishers Ltd 2000

 GOVERNANCE AND FUNDS ALLOCATION 347

The relatively small size of the sample means that a few observations could
be highly influential. We use two methods to check our results for their
robustness to outliers. First, we employ standard diagnostic statistics to test
for the presence of influential observations. We find three influential outliers
that could distort the estimates of the regression parameters.6 Panel B shows
results with these observations deleted. They are essentially the same as the
findings reported in Panel A of the table. Second, we use an alternative
estimation technique, least absolute deviations (L1) regression, that is more
robust to outliers than least squares regression. These regressions are shown
in Panel C of Table 2. The results of all three methods are strikingly consistent.
Our findings are not driven by a few outliers.
An obvious interpretation of these results is that executives who serve on the
board are more influential when boards are smaller. This finding is basically in
line with the Williamson-Fama-Jensen hypothesis that the key variable is
proportion of insiders; on a smaller board, the executive has a proportionately
larger voice. This mechanism seems implausible, though; it is hard to believe
that whether a manager is one-eighth or one-twentieth of the board makes a
large difference in his voice. More likely, small boards are different in some
other ways from large ones. For example, smaller boards may be more likely
to be hand-picked by the executive.
Substantively, these results fail to support the Williamson-Fama-Jensen
hypothesis if we interpret it as meaning that management expenses will be a higher
proportion of total expenses when insiders are a larger proportion of the
board. In our data, management expenses appear to be impervious to either
board size or board composition. The way Williamson actually formulates the
conjecture (quoted earlier) is that the expense-to-payout ratio will be higher
when insiders are a larger proportion. This is true in our sample of nonprofits,
but only because fund-raising expenses are a higher proportion of the total and
program expenses lower. Our sample shows no evidence that executives in
nonprofit corporations increase management expenses when they are on the
board. This finding is noteworthy, since a primary concern of the corporate
governance literature is managers pursuing their own interests.
Table 2 reveals that the effect on program activity of having an executive on
the board is opposite in sign and very nearly equal in magnitude to its effect on
fund-raising expenses. This is true both for the direct effect and for the
interaction with board size. We have a clear picture of executives attempting
to shift spending from programs to fund-raising, and of larger boards
successfully resisting this behavior. And this shift appears to be almost entirely
a tradeoff between programs and fund-raising; there is no significant effect on
management expenses. That is, we find no evidence that managers who serve
on the board attempt to increase spending on management and general
activities at the expense of programs.
Our result contrasts with Yermack's (1996) result on for-profit
corporations. When the CEO is on the board, smaller boards appear to be less,

ß Blackwell Publishers Ltd 2000

348 DYL, FRANT AND STEPHENSON

rather than more, successful at controlling managers. When the CEO is not on
the board, we find no independent effect of board size. Thus, we find no
support for the hypothesis that increased board size and CEO board
representation have effects in the same direction.
We find some evidence of economies of scale in management expenses,
consistent with the recent results of Hyndman and McKillop (1999). All three
panels show a negative coefficient on organization size (measured as revenue)
in the management expense regression, though only in Panel C is the effect
significant at the 5 percent level. The numbers indicate that the savings on
management expenses are, however, channeled into fund-raising rather than
program expenses, as larger size correlates with a higher proportion of budget
spent on fund-raising. Of course, size may be endogenous; organizations that
choose to spend proportionately more on fund-raising may naturally end up
being larger than other organizations.
All in all, we find rather mixed evidence for the hypothesis that managerial
self-interest is an important determinant of the policies of our sample of
medical charities. Managerial representation on boards of directors of
nonprofit organizations clearly makes a difference. Managers on small boards
are able to shift spending away from program activities and toward fund-
raising. This finding may be consistent with the self-interest hypothesis.
Increases in fund-raising expenditures presumably result in more revenue
from donations, increasing the job security of managers who have invested
their human capital in the medical charity. In addition, they may be able to
command larger salaries as heads of larger organizations. This argument is
closely analogous to Baumol's (1959) sales maximization hypothesis with
respect to for-profit firms ^ that increased revenue is beneficial to managers.
It may be, moreover, that managers of nonprofits are in part evaluated on the
external labor market by their success in fund-raising.7 They may thus have an
incentive to spend more on fund-raising than is optimal for the organization.
Conversely, however, managers may simply have longer time horizons than
boards and therefore be more willing to invest now to further the charity's
long-term objectives. Feigenbaum (1987, p. 251) suggests that:
. . . .the most important indicator of a research charity's performance is the fraction of
donations actually allocated to research projects.

In cross-sectional data, however, this measure is not an unambiguous index of

good performance. In particular, increased expenditures on fund-raising in
the current period may be an investment that results in increased resources
for medical research in future periods for the organizations in our sample.
When we examine the variable that one would expect to be the most
important indicator of managerial self-interest, management and general
expenses, we find that the empirical analysis does not support the hypothesis
that nonprofit managers use their influence on boards to promote their own
interests. We find no connection between management representation and

ß Blackwell Publishers Ltd 2000

 GOVERNANCE AND FUNDS ALLOCATION 349

Table 3
Executive Salary and Fund Balance Regressed on Governance Variables
and Other Variables

Dependent Total Listed by Exec. on Board Exec. x R2(adj.)

Variable Intercept Revenue NCIB Board Size Board Size [N]

Executive ÿ382,590. 25,277. 23,560. 144,023. 19,298. ÿ43,204. 0.708

Salary (ÿ8.87)** (6.82)** (1.98) (2.98)** (1.62) (ÿ2.49)** [43]
Fund 0.804 0.124 ÿ0.460 ÿ1.722 ÿ0.473 0.386 0.004
Balance (0.61) (1.24) (ÿ0.95) (ÿ0.983) (ÿ1.51) (0.75) [54]

Notes:
Executive salary is the salary of the highest-paid employee. Fund balance is measured as the total
fund balance divided by the charity's annual expenditures. Total revenues and board size are ex-
pressed as natural logarithms. A 0,1 indicator variable denotes whether the charity is listed by the
NCIB and/or if the executive director serves on the Board (yes = 1). The number of observations is
shown in brackets in the last column [N]. The numbers in parentheses are t-statistics calculated
using White's (1980) heteroskedasticity-consistent method.
** Significant at the 0.01 level.

management expenses. We do not find a relation between insider

representation and consumption of non-salary perquisites.
In summary, then, we see that board composition influences allocation
decisions, but not in the way predicted by the hypothesis. We are therefore
uncertain as to whether our findings are evidence of managerial self-interest.
We next turn to the other two measures of managerial self-interest ^ the
relationship between board structure and management salaries and between
board structure and fund balances.
The first row in Table 3 shows results of a regression of the highest
management salary in the organization on the independent variables used in
the previous regressions. Size, not surprisingly, is decisive in all of the
regression models; regressions including size explain a remarkable 70 percent
of the variance in top management salaries.
We also find a striking duplication of the earlier findings regarding program
activity and fund-raising expenses. Managerial salaries are higher when
managers are on the board. The coefficient on the dummy variable suggests
that they are, on average, roughly $144,000 higher when the executive
director serves on the board. This effect is offset by larger board size, however.
When the board size reaches 28 the effect of the executive being on the board is
exactly canceled out. Again, this finding is remarkably similar to the results
reported in the program activity and fund-raising regressions in Table 2. This
duplication makes it more likely that the result on fund-raising indeed reflects
managerial self-interest, rather than, say, differences in time horizon.8

ß Blackwell Publishers Ltd 2000

350 DYL, FRANT AND STEPHENSON

The strong effect of size in the salary regression, moreover, strengthens the
argument for an analogy to the Baumol `sales-maximization' hypothesis. The
allocation of additional resources to fund-raising may be an indirect means of
increasing management salaries over time. If fund-raising expenditures
increase the annual revenues of the organization, and if salary levels are a
function of revenue (as the results in Table 3 suggest), then increased fund-
raising activities may be motivated by a desire to increase management
salaries in the long run.
A large fund balance can be thought of as providing financial stability and
security for a nonprofit organization's managers. The second row in Table 3
shows that the size of the charities' fund balances is unrelated to any of the
independent variables in our regression analysis. In particular, there is no
evidence that executives on boards use their influence to build up large fund
balances. Fund balances are idiosyncratic and unrelated to either board
structure or NCIB listing.

CONCLUSION

There is substantial evidence that managers of for-profit corporations use

their control of boards of directors to further their own interests. We report
here the first evidence of such behavior by managers of nonprofit corporations.
In the organizations we investigate, managerial representation on the board is
not associated with increased management expenses or organizational wealth,
but it is associated with higher salaries and increased attention to fund-raising.
While the finding on fund-raising is consistent with a variety of explanations,
self-interested and otherwise, the fact that it is paralleled by the salary result
suggests self-interest.
The research on managerial discretion in for-profits, both theoretical (e.g.
Williamson, 1964) and empirical (e.g. Brickley and James, 1987) has not
drawn a clear distinction between salary and other management perquisites.
Our results suggest that these two are not perfect substitutes ^ we find evidence
of managers using their influence to increase salary, but not other
management expenses. This is an obvious area for further research.
The question of managerial self-interest in nonprofits may be further
clarified by examining other types of organizations within the nonprofit sector
^ for example, educational, religious, or social service organizations. It might
also be helpful to examine these organizations longitudinally, to see whether
changes over time in board size and composition are reflected in measurable
outcomes. A better understanding of the issues we address here may not only
provide a basis for better prescriptions regarding the governance of nonprofit
corporations, but may also contribute to progress in developing a more
general theory of organizational governance.

ß Blackwell Publishers Ltd 2000

 GOVERNANCE AND FUNDS ALLOCATION 351

NOTES
1 IRS regulations state that individuals are entitled to examine a copy of a nonprofit
organization's three most recent Form 990 informational returns and schedules during regular
business hours at the organization's principal, regional, and district offices, but the
organizations are not required to mail this information.
2 These were the Shriners' Hospitals for Crippled Children, the City of Hope, and ALSAC-St.
Jude's Children's Research Hospital.
3 National Charities Information Bureau, Wise Giving Guide (June 1991, p. 7).
4 Although 100 percent of the NCIB-listed charities responded to our request for information,
only 54 percent of the non-listed charities responded. Controlling for NCIB listing thus reduces
the possibility of selection bias.
5 The American Cancer Society and the American Heart Association are orders of magnitude
larger than the typical charity in our sample. Estimates from OLS regressions omitting these
two observations are essentially identical to the results reported in Panel A of Table 2 for the
full sample.
6 The three outliers are the United Ostomy Association, the National Leukemia Association,
and the National Foundation for Cancer Research. Their common characteristic is unusually
low proportional expenditures on program activities.
7 We thank Keith Provan for this suggestion.
8 As noted earlier, the sample includes two organizations reporting a highest management
salary of zero. Deleting these observations gives similar results, but the coefficients are
significant only at the 10 percent level.

REFERENCES
Baumol, W. J. (1959), Business Behavior, Value and Growth (New York, Harcourt Brace).
Baysinger, B. D. and H. N. Butler (1985), `Corporate Governance and the Board of Directors:
Performance Effects of Changes in Board Composition', Journal of Law, Economics, and
Organization, Vol. 1, No. 1 (Spring), pp. 101^24.
Berle A. A. and G. C. Means (1932), The Modern Corporation and Private Property (Revised ed. 1967,
Harcourt, Brace and World, New York).
Brickley, J. A. and C. M. James (1987), `The Takeover Market, Corporate Board Composition,
and Ownership Structure: The Case of Banking', Journal of Law and Economics, Vol. 30, No. 1
(April), pp. 161^80.
________________________, J. A., J. L. Coles and R. L. Terry (1994), `Outside Directors and the Adoption of Poison
Pills', Journal of Financial Economics, Vol. 35, No. 3 (June), pp. 371^90.
Dyl, E. A. (1988), `Corporate Control and Management Compensation: Evidence on the Agency
Problem', Managerial and Decision Economics, Vol. 9, No. 1 (March), pp. 21^25.
Fama, E. F. and M. C. Jensen (1983), `Separation of Ownership and Control', Journal of Law and
Economics, Vol. 26, No. 2 (June), pp. 301^25.
Feigenbaum, S. (1987), `Competition and Performance in the Nonprofit Sector: The Case of
Medical Research Charities', Journal of Industrial Economics, Vol. 35, No. 3 (March), pp.
241^53.
Gomez-Mejia, L. R., H. Tosin and T. Hihkin (1987), `Management Control, Performance, and
Executive Compensation', Academy of Management Journal, Vol. 30, No. 1 (March), pp. 51^
70.
Hermalin, B. E. and M. S. Weisbach (1988), `The Determinants of Board Composition', RAND
Journal of Economics, Vol. 19, No. 4 (Winter), pp. 589^606.
Herman, R. D. and F. P. Tulipana (1989), `Board-Staff Relations and Perceived Effectiveness in
Nonprofit Organizations', in R.D. Herman and J. Van Til (eds.), Nonprofit Boards of
Directors: Analyses and Applications (Transaction Publishers, New Brunswick), pp. 48^59.
Hyndman, N. and D. McKillop (1999), `Conversion Ratios in Charities in England and Wales:
An Investigation of Economies of Scale', Financial Accountability & Management, Vol. 15, No. 2
(May), pp. 135^53.

ß Blackwell Publishers Ltd 2000

352 DYL, FRANT AND STEPHENSON

Jensen, M. C. (1986), `Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers',
American Economic Review, Vol. 76, No. 2 (May), pp. 323^29.
________________________ (1993), `The Modem Industrial Revolution, Exit, and the Failure of Internal Control
Mechanisms', Journal of Finance, Vol. 48, No 3 (July), pp. 831^80.
Lipton, M. and J. W. Lorsch (1992), `A Modest Proposal For Improved Corporate Governance',
The Business Lawyer, Vol. 48, No. 1 (November), pp. 59^77.
Manne, H. (1965), `Mergers and the Market for Corporate Control', Journal of Political Economy,
Vol. 73, No. 2 (April), pp. 110^20.
Mayers, D., A. Shivdasani and C. W. Smith (1997), `Board Composition and Corporate Control:
Evidence From the Insurance Industry', Journal of Business, Vol. 70, No.1 (January), pp. 33^
62.
Middleton, M. (1986), `Nonprofit Boards of Directors: Beyond the Governance Function', in S.
Rose-Ackerman (ed.), The Economics of Nonprofit Institutions: Studies in Structure and Policy
(Oxford University Press, New York), pp. 141^53.
Pfeffer, J. (1972), `Size and Composition of Corporate Boards of Directors: The Organization and
its Environment', Administrative Science Quarterly, Vol. 17, No. 2 (June), pp. 218^28.
________________________ (1973), `Size, Composition and Function of Hospital Boards of Directors: A Study of
Organization-Environment Linkage', Administrative Science Quarterly, Vol. 18, No. 3
(September), pp. 349^64.
Provan, K. (1988), `Organizational and Decision Unit Characteristics and Board Influence in
Independent Versus Multihospital System-Affiliated Hospitals', Journal of Health and Social
Behavior, Vol. 29, No. 3 (September), pp. 239^52.
Rosenstein, S. and J. G. Wyatt (1990), `Outside Directors, Board Independence, and Shareholder
Wealth', Journal of Financial Economics, Vol. 26, No. 2 (August), pp. 175^91.
Santerre, R. E., and S. P. Neun (1986), `Stock Dispersion and Executive Compensation', Review of
Economics and Statistics, Vol. 88, No. 4 (November), pp. 685^87.
Sridharan, U. V. (1996), `CEO Influence and Executive Compensation', Financial Review, Vol. 31,
No. 1 (February), pp. 51^66.
Vance, S. C. (1964), Boards of Directors: Structure and Performance (University of Oregon Press,
Eugene).
Weisbach, M. S. (1988), `Outside Directors and CEO Turnover', Journal of Financial Economics,
Vol. 20, No. 1 (January/March), pp. 431^60.
White, H. (1980), `A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct
Test For Heteroskedasticity', Econometrica, Vol. 48, No. 4 (May), pp. 817^38.
Williamson, O. E. (1964), The Economics of Discretionary Behavior: Managerial Objectives in a Theory of
the Firm (Prentice Hall, Englewood Cliffs, NJ).
________________________ (1983), `Organization Form, Residual Claimants, and Corporate Control', Journal of Law
and Economics, Vol. 26, No. 2 (June), pp. 351^66.
Yermack, D. (1996), `Higher Market Valuation of Companies With Small Boards of Directors',
Journal of Financial Economics, Vol. 40, No. 2 (February), pp. 185^211.
Zald, M. N. (1969), `The Power and Function of Boards of Directors: A Theoretical Synthesis',
American Journal of Sociology, Vol. 75, No. 1 (July), pp. 97^111.

ß Blackwell Publishers Ltd 2000