The Price of Doing Good:
Executive Compensation in Nonprofit Organizations
Peter FrumkinProfessor of Public Affairs
Lyndon B. Johnson School of Public AffairsBox Y
University of Texas at AustinAustin, Texas 78713
Elizabeth K. KeatingAssistant Professor of Public Policy
Kennedy School of GovernmentHarvard University
79 JFK StreetCambridge, MA 02138
Tel: (617) 495-9856E-Mail: [email protected]
We thank the Aspen Institute's Nonprofit Research Fund for their financial support of this project and the National Center for Charitable Statistics at the Urban Institute for providing us with Form 990 data. We appreciate the helpful advice of Burton Weisbrod, Rachel Hayes and members of the Hauser Center faculty research seminar.
The Price of Doing Good:
Executive Compensation in Nonprofit Organizations
Abstract
This article examines the foundational assumption that nonprofit organizations
operate under a non-distribution constraint, which prohibits paying out excess earnings and
requires their application to the organization’s mission. Examining determinants of
nonprofit executive compensation, we find that nonprofit CEO pay is strongly predicated on
that in similar-sized organizations. Nonprofit executive compensation is modestly affected
by CEO performance, measured by fund-raising results or administrative efficiency. We
find evidence, inconsistent with the principle of not distributing profits, that CEO
compensation is significantly higher in organizations with “free cash flows”. We discuss
implications of this finding on distinctive organizational identity of nonprofit organizations.
Keywords: executive compensation, nonprofit organization, efficiency, accountability.
1
The Price of Doing Good:
Executive Compensation in Nonprofit Organizations
I. Introduction
Nonprofit organizations depend on good will, generosity, and commitment. Existing
studies of nonprofit compensation indicate that the pay of nonprofit workers and executives
is lower than their employees in comparable positions in for-profit firms (Preston, 1989;
Steinberg, 1990; Handy and Katz, 1998; and Ruhm and Borkoski, 2000). However,
appropriate compensation for the leaders of these organizations is central to the long-term
viability and success of the entire nonprofit sector. Quality leadership for nonprofit
organizations must be recruited, motivated and retained. Thus, it is not unexpected that
nonprofit organizations frequently find themselves under competitive pressure to find ways
to offer compensation packages that are comparable to similar nonprofit, or even for-profit,
organizations.
To protect their charitable status, nonprofit organizations are legally prohibited from
distributing earnings that “inure to the private benefit of any private shareholder or
individual.”1 This prohibition, called the “non-distribution constraint,” limits a nonprofit’s
ability to reward nonprofit executives directly for many forms of financial performance.
Historically, nonprofit compensation decisions have not been incentive-based but rather
determined by revenues or earnings or loosely connected to social or programmatic goals.
(Kertz, 1997; Frumkin and Andre-Clark, 1999).
Modest executive compensation packages and limited use of incentives have posed
challenges to nonprofits during the 1980s and 1990s. Due to the commercialization and
1 U.S.C. §501(c)(3).2
increased competition from for-profit and nonprofit providers, nonprofit executive
compensation practices have changed. Some nonprofit organizations have shifted from fixed
salaries to ones containing a variable cash compensation component based on fundraising,
cost reductions or specific programmatic outcomes (Barbeito and Bowman, 1998).
However, these plans have met with resistance because they tend to focus heavily on
financial measures of nonprofit performance rather than on the social dimensions of
performance, namely mission fulfillment.
Nonprofit managers have also sought “comparable pay” (Pappas, 1995; Drucker,
1992) with business managers. Benchmarking of salaries of nonprofit executives has
become more prevalent, encouraged by a new set of IRS regulations that allows sanctions
and fines to be levied on nonprofit organizations that pay their executives excessive
compensation relative to similar nonprofit and for-profit firms. However, for many
nonprofit organizations, increasing executive compensation remains prohibitive because of
budgetary and moral constraints.
To better understand nonprofit compensation practices, we test three main
competing hypotheses. First, we consider whether executive compensation in nonprofit
organizations is a function of the size of the organization. We expect to find that nonprofit
managers are more highly compensated in larger organizations, consistent with pay levels
reflecting managerial responsibility. Second, we examine the prevalence of pay-for-financial
performance in the nonprofit sector. We expect to find little or no connection, given the
weak relation between financial performance and mission fulfillment and the existence of
the non-distribution constraint. Third, we look at the role of liquidity or “free cash flow” and
examine its effect on nonprofit compensation. We expect, if the non-distribution constraint
is indeed operative, that liquidity will not affect CEO compensation decisions. The second
3
and third tests are particularly significant in the nonprofit context. If a strong association
exists between compensation and liquidity or financial performance, it would challenge the
effectiveness of the non-distribution constraint.
The paper proceeds in five steps. First, we present a review of corporate
compensation literature and discuss its applicability to the nonprofit sector. Second, we
develop our research hypotheses. Third, we describe the panel data, the variables, and our
research design. We then present the results of the analysis and interpret their meaning.
Finally, we offer some concluding remarks about the challenges of explaining executive
compensation in nonprofit organizations.
II. Literature Review
A. For-Profit CEO Compensation
To understand the nature of nonprofit compensation, we start by examining the
management literature on the determinants of CEO pay in business firms. Much of this
extensive body of research relates to three general themes. First, compensation studies have
consistently found a link between the size of the company and executive compensation
levels (Gomez-Mejia, Tosi and Hinkin, 1987). Faced with considerable uncertainty,
companies pay their CEO based on the scope of their responsibility and the amount of
resources they are charged with managing. Simon’s early explanation (1957) of this
phenomenon was that firms used compensation to distinguish between different managerial
levels, and because large firms have more levels, they tend to pay their leaders more than
smaller and less hierarchical companies. Subsequently, extensive empirical work has
demonstrated that managers earn more when they have been entrusted with leading large
companies.
4
Second, drawing on agency theory, many studies have examined the linkage
between company financial performance and the executive compensation levels. Some have
found a connection to profitability (Agarwal, 1981; Lewellen and Huntsman, 1970), though
many other studies have concluded that firm performance is not a key driver of CEO
compensation (Benston, 1985; Deckop, 1987; Jensen and Murphy, 1990; Kerr and Bettis,
1987; Murphy, 1985; Redling, 1981; Rich and Larson, 1984). Researchers then have
focused on relative performance evaluation and tested whether CEO pay decisions were
driven by the performance of a manager compared to his peers in a given field (Holmstrom,
1982). One reason why boards might take into consideration the compensation decision of
other companies stems from the possible increased efficiency that such information might
make possible (Antle and Smith, 1986; Kerr and Kren, 1992; Morck, Schleifer and Vishny,
1989). Interestingly, institutional theory has not been actively used to examine
compensation decisions. Outward-oriented decision making has been understood and
rational comparative evaluation of performance, rather than as a mimetic process, an
organizational ritual, or a symbolic legitimizing behavior (Meyer and Rowan, 1977; Tolbert
and Zucker, 1983; DiMaggio and Powell, 1991).
Because of the weak link that has been established between pay and performance,
alternative explanations of compensation patterns have been advanced. A third interesting
explanation of CEO compensation has focused on the independence and relative power of
the board. In situations when the board is non-independent or weak, CEOs may be highly
compensated due to poor oversight by board or by collusion. In either case, the control
systems designed to protect the interests of shareholders fail. In analyzing CEO
compensation levels, board-CEO relations thus becomes a critical factor to consider
(Westfall and Zajac, 1994). Some research has also considered the relative power and
5
influence of shareholders in explaining CEO pay patterns (Gomez-Mejia, Tosi and Hinkin,
1987) in an attempt to understand board decision making.
B. Nonprofit Compensation and the Non-Distribution Constraint
As a whole, nonprofit organizations tend to pay their workers at lower salaries than
their business firm counterparts. Several theories could explain this finding: Many who
choose to work in the nonprofit sector engage in “labor donations,” preferring altruistic and
other non-pecuniary benefits to monetary rewards (Rose-Ackerman, 1986, Preston, 1989).
Wages may be lower in nonprofit jobs as a screening device, attracting only those managers
willing to restrain their desire for profit (Young, 1977; Hansmann, 1980). Other theories
suggest that paying nonprofit executive salaries that rival those in the business would be
highly problematic given expressive character and social orientation of these organizations
(Mason, 1996). However, if the compensation differences between the sectors grows too
large, then nonprofits will be unable to attract personnel with strong management and
leadership skills needed to ensure organizational growth and capacity building (Letts, Ryan
and Grossman, 1999).
Due to these competitive pressures, research has explored the compensation
differences across nonprofit and for-profit sectors (Borjas, Frech III and Ginsburg, 1983;
Frank, 1996; Goddeeris, 1988; Johnson and Rudney, 1987; Mocan and Viola, 1997; Preston,
1989). These cross-sector studies generally apply tests from one or two of the three strands
of for-profit literature. Roomkin and Weisbrod (1999) and Brickley and Van Horn (2000)
focus on profit and nonprofit hospitals, while others concentrated on variations in executive
pay (Oster, 1998; Baber, Daniel and Robert, 1999; Hallock, 2000). These papers often
explain differences between the nonprofit and for-profit sectors using the labor donations,
6
screening or social orientation theories. In contrast, our study focuses on one of the
distinguishing legal features of the nonprofit sector: the non-distribution constraint.
In principle, the nonprofit organizational form allows society to overcome certain
market or "contract failures" (Hansmann, 1980). In exchange for the provision of services to
the needy, nonprofits are provided tax-exemptions and the ability to offer contributors tax-
deductions for their charitable gifts. Hence, nonprofits can enable society to increase the
output of certain goods and services, without moving to direct government provision or the
provision of subsidies to for-profit firms. To ensure that nonprofits do not abuse their
privileged tax position, nonprofits are legally subject to the “non-distribution constraint.”
Hansmann (1980, 840) describes this requirement as:
A nonprofit organization is, in essence, an organization that is barred from distributing its net earnings, if any, to individuals who exercise control over it, such as members, officers, directors, or trustees…. Net earnings, if any, must be retained and devoted in their entirety to financing further production of services that the organization was formed to provide.
By consenting to the non-distribution constraint, nonprofits agree not to distribute profits to
employees or third parties but, instead, to use any excess resources to fulfill the
organizational mission.
The theory of non-distribution creates a line demarcating nonprofits and business
organizations. While for-profit entities can and do freely divide up profits between
shareholders and management, nonprofits are thought to operate differently. For some
nonprofit constituents, the presence of an operating surplus is a sign that nonprofits are
charging too much for their services, either to clients paying a fee or to donors making
contributions. Instead of accumulating surpluses and applying it to future mission-related
work, nonprofits face some pressure to reduce the costs of their services to the break-even
point or expand the volume of services. At the same time, they must be prudent and
7
accumulate enough surplus to sustain a reasonable level of net assets in the event that their
financial position changes unexpectedly.
III. Research Hypotheses
A. Organizational Size and Managerial Responsibility
Extensive for-profit research indicates that corporate executive compensation is a
function of organizational size.2 Murphy (1998) argues that size is a proxy for managerial
skill requirements, job complexity, and span of control. Nonprofit compensation research
also suggests that size may be an important determinant of CEO compensation (Hallock,
2000). Size or organizational scale may actually be a more significant determinant of
compensation in nonprofit than for-profit organization since inputs such as program
expenses and tangible assets are the most visible and measurable element of the
organization’s production process.
Organizational size may also be an important factor in nonprofit pay because
governing boards often determine compensation by benchmarking against senior executives
in nonprofits that are comparable in size and industry focus (Barbeito and Bowman, 1998).
A growing number of professional associations across fields of nonprofit activity now
actively collect and disseminate compensation studies, which report average salaries and
benefits for executives at organizations across different budget categories. Boards are able to
rely on this data to guide their compensation decisions.
Finally, organizational size provides legitimacy (Scott, 1995; Zucker, 1988). Large
institutions typically garner more publicity, have higher prestige, and are viewed as more
effective by virtue of the scope of their activities. Moreover, boards of large institutions are
typically made up of leaders from the community, whose judgment is less likely to be 2 See Murphy (1998) for summary of research.
8
subject to questioning and critical scrutiny. Managers can and do receive larger
compensation packages at these larger institutions because they are simply perceived as
deserving and entitled to earn more. Organizational size can also help overcome norms of
frugality and self-denial that those who work for financially struggling nonprofit
organizations often experience. In a sector where resources are generally scarce, size thus
brings with it financial flexibility and allows for personal rewards. We posit as a first
hypothesis:
H1: CEOs managing large nonprofits will earn more than CEOs at smaller-sized organizations.
B. Incentive Compensation
As nonprofit boards deliberate over the question of CEO compensation, a
compelling criterion is managerial performance. Nonprofit management has become
increasingly understood as a legitimate profession, with its own body of expert knowledge
and a set of best practices (Light, 2000). Leaders of major nonprofit organizations have
come to adopt a more business-like approach to their work, adopting concepts such as
quality management, process reengineering, and benchmarking from the world of corporate
strategy. Pay-for-performance in the nonprofit sector is especially problematic due to the
difficulties in measuring performance and the risk of violating the non-distribution
constraint. Still, two forms of performance have been the focus of most incentive plans:
fund-raising and cost efficiencies.
While many large organizations have development staffs that manage the fund
raising process, the CEO is ultimately responsible for the financial position of their
organization. The ability to raise money is frequently taken as a sign that the organization is
performing well. The logic is that donors reward organizations that are doing good work and
9
punish those that are not by withholding contributions. Hence, fundraising provides an
easily measured metric that proxies for mission fulfillment.
Another way that managerial performance can be judged is by how resources are
used. Frugality is viewed a virtue in nonprofits. Administrative cost-cutting in nonprofits is
often an organizational necessity, particularly when revenues wane or when the community
needs addressed by the nonprofit is extremely pressing. Many funders and watchdog
organizations interpret low ratios of administrative to total expenses as a sign that a
nonprofit is well run and mission-focused.
Traditional agency theory recommends pay-for-performance compensation as way
of aligning agents’ actions with principals’ goals, thereby encouraging effort and reducing
perquisite behavior (Jensen and Meckling, 1979, Fama, 1980). However, paying incentives
based on excess earnings directly conflicts with the non-distribution requirement, since
revenues or cost savings are converted into in higher salaries and benefits for staff rather
than services for clients.3 For this reason, nonprofits have traditionally sought to avoid
paying employees compensation based on financial measures of performance. With all these
factors weighing against pay-for-performance in the nonprofit sector, we hypothesize:
H2: Nonprofit CEOs pay will not be based on the financial performance of the organization.
C. Free Cash Flows and Liquidity
The diversion of “free cash flows” to increase executive pay is another pressing
concern in the nonprofit sector. Most nonprofits seek to achieve stability and sustainability
as a means of improving their capacity to pursue their missions effectively. Given the 3 For example, Tax Court found in People of G-d Community vs. Commissioner
that the payment of a percentage of gross receipts to a pastor violated the non-
distribution constraint (75 T. C. 127, 132 (1980).10
multiple funding streams that support nonprofit organizations, including individual
contributions, foundation grants, fees for service, and government contracts, this task can be
complex and demanding. One tempting response to uncertain funding flows is to build
financial reserves to protect organizations from precipitous changes in one or more revenue
streams. Organizations that are able to increase their liquidity are able, in principle, to cope
more easily with changes in the funding environment. However, this financial slack creates
the temptation to use these resources for personal inurement.
Nonprofits have developed several mechanisms to limit the use of the financial
slack. Donors place restrictions on the use of their funding. Both donors and boards can set
aside funds in permanent or quasi-permanent endowments. Since these funding sources
generally do not fully cover the cost of services, many nonprofits pursue one or more
strategies for developing financial slack. First, nonprofits may actively solicit individuals
for unrestricted funds through special events, direct mail marketing or telemarketing
campaigns. Unlike restricted grants, these funds, which come from many small and/or loyal
donors, do not trigger significant monitoring and oversight.
A second means of enhancing liquidity is through commercial activities, such as fee
for services or product sales. Users and clients tend to focus more on the convenience and
cost of the services rendered, rather than on the underlying financial practices of the
nonprofit organization. Earned income rarely requires program or financial reporting to
outside parties, but instead relies on customer satisfaction.
Finally, some organizations may have endowments. Some or all of the investment
income is used annually to support the general budget or some restricted purposes.
Endowments decrease pressure on managers to raise funds through annual appeals and
reduce the monitoring that may accompany new donations. Given the favorable stock
11
market performance in the 1990s, some nonprofits have been able to use the additional
investment income to cover increases in operating costs. Thus, organizations with
endowments may have more discretionary cash than organizations without these
investments.
The presence of surplus cash in the organization may encourage the board to raise CEO
salaries or pay one-time bonuses. While the non-distribution constraint is legally violated
when “excess earnings” are given over to managers, case law suggests that a large increase
in compensation may be reasonable if compensation has been inadequate in previous years.4
To maintain their privileged tax status and organizational identity, nonprofits may act to
avoid even the appearance of self-interested distributions. Our third research hypothesis is:
H3: Nonprofits CEO compensation will not be determined by liquidity and free cash flows.
In carrying out our analysis, we are interested in isolating the main determinants of
nonprofit compensation and their implications for the strength and meaning of the non-
distribution constraint. While any link between resources growth and compensation might
appear questionable and potentially problematic, only a strong link between fund-raising
results and increased executive compensation would present clear evidence of diversion of
the excess revenues to non-mission related purposes. The potential implications of a
significant relation between organizational size and free cash are more subtle and complex.
IV. Research Design
A. Data and Sample Selection
The sample data used in our analysis originates from the annual Form 990 nonprofit
tax filings, which are unaudited reports completed by either the nonprofits themselves or
4 Medina v. Commissioner, 46 T.C.M. (CCH) 76 (1983).12
with the help of an outside preparer or audit firm. The sample is drawn from the nonprofit
organizations filing Form 990s in the 1998 to 2000 period. The annual data is repackaged
and disseminated to academic researchers by the Urban Institute’s National Center on
Charitable Statistics (NCCS) and the Philanthropic Research Institute (Guidestar). The
annual data files were combined into a single database. The final sample totals 27,319
nonprofit organizations drawn from the 1999 and 2000 dataset, for a total of 33,770
observations. The 1998 dataset is used to help develop lagged variables for estimation.
B. Model Development
We adopt a pooled specification as follows:
Compensationit = α + β1 Total Fixed Assetsit-1 + β2 Total Program Expensesit-1
+ β3 Administrative Efficiencyit-1 + β4 Contribution Growthit-1
+ β5 Commercial Revenue Shareit-1 + β6 Liquid Assets-to-Expensesit-1
+ β8 Investment Portfolio-to-Total Assetsit-1 + ε (1)
The model is employed at the sector-wide level and for each of six major sub-sector
classifications based on the National Taxonomy of Exempt Entities: Arts, Education,
Health, Human Services, Religious and Other (which are primarily public and societal
benefit organizations). We assess the statistical significance of individual variables using a
t-test that controls for firm dependence.5 To assess the relative explanatory power of groups
of variables, we use the Vuong test (1989) z-statistic.
5 The robust estimator of variance assumes the observations are not independent
but that they are divided in M groups (i.e., firms) G1, G2, ..., GM that are
independent. The estimator becomes , where = (2ln
L/2)-1 and uk(G) is the contribution of the kth group to the score ln L/ (Huber
1967; Rogers 1993).13
For our dependent variables, we use three different measures of compensation: CEO
salary, CEO benefits and total CEO compensation. The last variable simply combines
executive salary and benefits. The compensation data is drawn from the salary and benefits
of the officers, directors and key employees reported on Part V of the IRS 990 Form. For
each observation, we identified the CEO through a multi-step process. First, we selected
salaries and benefits information for person’s whose job title was CEO or Chief Executive.
If none were given, then we selected the Executive Director. If that search failed, we
selected the person with the title of President. If the job titles were blank, then we selected
the highest paid person. We included both executive pension plan and expense account
expenditures in our measure of benefits.
To test our first hypothesis, we rely on two variables: lagged total fixed assets and
lagged total program expenses. Prior studies have generally used total assets or log of total
assets to proxy for size (Hallock, 2000). Our field experience with nonprofits leads us to
believe that boards set CEO compensation base on annual budgets and scale of operations in
comparison to industry peers. We chose total fixed assets (which includes land, building,
and equipment) as a proxy for scale of operations and total program expenses as a measure
of the annual budget. In nonprofits, total program expenses include costs of program
services, but exclude administrative and fund-raising expenses. We expect CEO
compensation to be positively associated with both fixed assets and program expenses.
We developed two variables associated with our pay-for-performance hypothesis.
Due to the non-distribution constraint, boards have difficulty rewarding CEOs directly for
cost savings. The ratio of administrative expenses to total expenses is a standard measure of
overhead in the nonprofit industry. Boards view that the lower this ratio, the higher the
efficiency of operations. To measure administrative efficiency, we take one minus the
14
administrative expenses to total expenses.6 To supplement this variable, we include a
second measure of CEO performance: dollar growth in contributed revenue. A critical part
of the work of most nonprofit CEOs is raising money for the organization. The increase in
contributed revenue is a particularly observable measure that boards may correlate with
CEO effort. Other revenue sources, such as program service revenue, investment income,
and special event revenue, may not be as closely tied to CEO performance. If nonprofits are
adopting a more performance-based compensation approach, then we expect that growth in
contributed revenue will be positively associated with compensation. However, since some
incentive pay may be interpreted as a violation of the non-distribution constraint, boards
elect not to reward CEOs directly for increasing contributions. Additionally, we may fail to
find a significant relation because restricted contributions bear donor-imposed restrictions,
which often include limitations on the funds spent on personnel services.7
To test our third hypothesis, we selected three variables that determine whether an
organization is cash constrained or has free cash flows. First, we considered lagged
commercial revenue as a share of total revenue. Commercial revenues are composed of
proceeds from sales of goods as well as program service fees and charges generally paid by 6 Due to accounting flexibility, some nonprofits may allocate a disproportionate
share of joint costs to program rather than administrative activities. Hence, our
variable measures reported rather than actual administrative efficiency. The
variable may be biased of we have omitted a variable correlated with this
misallocation.
7 Due to data limitations, we are unable to distinguish between growth in
unrestricted and restricted contributions.
clients, insurance companies or some government agencies. Often, these funds are relatively
free of donor oversight or outside imposed restrictions. Second, we create a measure of
liquid assets to expenses. Liquid assets are computed using cash plus receivables less
payables. This ratio indicates the proportion of annual expenses that can be paid out of
liquid assets and provides a sense of the organization’s debt-paying ability.8 Finally, we test
to see if CEO compensation is associated with endowments that help pay for general and
administrative costs and that reduce the scrutiny associated with new donations. We use the
ratio of investment portfolio to total assets as our measure.
V. Results
A. Descriptive Statistics
In Table 1, we present descriptive statistics. Panel A indicates that there are 24,532
observations from 1999 and 9,238 from 2000. These observations are divided into six
industry classifications in Panel B. Organizations in the human service and health sectors
compose 38.6% and 21.2% of the sample, respectively. We then examine the key variables
by sector in Panel C. As the statistics indicate, there is considerable variation in most
variables between sectors. Compared to other sectors, the health and education sectors are
composed of institutions that are larger in size, based on fixed assets and program expenses.
The sectors do not vary much in regard to administrative efficiency: means range only
between 79% (arts) and 84.8% (human services). However, the variation in dollar growth in
contributions is considerable with other and education growing $428 thousand and $348 8 We also used an alternative measure, profitability, which we defined as net income
divided by total revenues. The results using profitability were qualitatively unchanged.
thousand per year in contrast to an only $66,000 annual increase for religious organizations.
Health and education have the highest reliance on commercial revenue at over 50%, with
religious groups obtaining only 21% from this source. Educational organizations generate
the highest surplus, equal to 9.6% of total revenues, while human services earn just a 3.0%
margin. Despite this low profit margin, human services, like educational institutions, have a
liquid asset-to-expense ratio of 85%, in contrast to the 43% and 39% for health and religious
organizations, respectively. Finally, educational institutions have the highest endowments,
equal to 17.5% of total assets, as compared to 10-11% for arts, health and other and only
4.6% for religious organizations.
Table 2 presents the mean, standard deviations and correlation matrix. As expected,
there is a high degree of correlation between executive salary and total executive
compensation (0.94), but the relation between executive benefits and total compensation
was considerably weaker (0.65). The explanatory variables that are most strongly correlated
with total compensation are total fixed assets (0.37) and total program expenses (0.45).
Among the other independent variables, a high correlation exists between the two variables
associated with the first hypothesis. Specifically, total fixed assets and total program
expenses are correlated at 0.71. None of the other independent variables are correlated at or
above 0.15.
B. Sector-Wide Regressions
At the outset of our analysis, we ran pooled sector-wide regressions to understand
the overall relation between compensation and the explanatory variables. Table 3, Panel A
provides the results using three different dependent variables (total CEO compensation,
CEO salary, and CEO benefits). For the total compensation model, we find a significant
base salary to pay (as measured by the constant), with each CEO receiving just over $33,000
in annual compensation. When decomposed between salary and benefits, we find that the
fixed component of compensation is essentially salary.
CEO compensation is positively related to both measures of organizational size. For
every thousand dollars of fixed assets or program expenses, a CEO’s total compensation
increases $0.27 and $5.53, respectively. As regards pay-for-performance-type
compensation, we find some evidence that CEOs receive higher compensation when
organizations report higher administrative efficiency. Specifically, the CEO receives about
$105 in added salaries and benefits for a one percent increase in administrative efficiency.
This added compensation is related significantly to both salary and benefits. In contrast,
there is no significant relation between the dollar growth in contributions and compensation.
Of the three free cash flow measures, two were significant. The total compensation
regression suggests CEOs receive $320 in additional compensation when the share of
commercial revenues is higher by one percent. When the organization holds an additional
percent of investments, the CEO on average obtains about $626 in supplemental
compensation.
The sector-wide regressions are then compared to determine if one hypothesis
accounts for significantly more of the variance in compensation (Table 3, Panel B). The
Vuong test indicates that compensation is most strongly related to the first hypothesis. Not
only do the size variables provide more explanatory power than either the pay-for-
performance and free cash flow variables but they explain more variation in compensation
than the other two hypotheses combined (z-statistic = 2.97). When compared to the pay-for-
performance variables, the free cash flow metrics have higher explanatory power, meaning
that compensation is more closely related to free cash flows than to incentive performance.
The results from the Vuong test on the salary alone are identical, while the benefits results
similar in nature but less statistically significant.
C. Industry-Specific Regressions
Since the nonprofit industry is quite heterogeneous, we explore the compensation
question in the major sub-sectors. In Table 4, we provide the results of industry-specific
regressions examining total CEO compensation. In four of the six sub-sectors, executives
receive a significant fixed portion to their compensation, receiving a base ranging from
$25,434 (health) to $42,037 (education).
The examination of the arts sector reveals that CEO compensation is significantly
explained by the regression variables with a R2 of 0.45. CEO compensation is related
significantly to program expenses but not to fixed assets. This latter finding is surprising
given that this category contains numerous performing arts organizations and museums. The
compensation of arts CEOs increases more rapidly in regard to total fixed assets ($28.04 for
each $1,000 in program expenses) than in the other sub-sectors. In contrast, the
remuneration of arts CEOs is negatively associated with commercial revenue share, in
contrast to the positive relation in the remaining sub-sectors. As expected, greater
administrative efficiency, higher liquidity, and a more extensive endowment are associated
with higher compensation. However, generating an annual profit does not translate into
higher compensation. Overall, the organizational size variables explain a substantially
greater proportion of the variation in compensation for arts CEOs than the other two
hypotheses combined (z-stat= 8.19) (Table 4, Panel B).
While arts executive pay is closely related to program expenses, CEOs at educational
institutions receive compensation that is significantly associated with fixed assets. These
organizations include primary and secondary schools as well as colleges and universities.
Unlike the arts CEOs, educational leaders are better compensated when their organizations
have growth in contributions but not when they are more administratively efficient. In the
education sector as in the arts area, two of the three free cash flow variables are significantly
related to compensation. In the educational sector, the free cash flow variables and the
organizational size variables each individually accounted for more of the variation in
compensation than pay-for-performance. Based on the Vuong test, we could not reject the
hypothesis that organizational size and free cash flow had equal explanatory power.
Due to the competition in the health sector between private and nonprofit firms, one
might expect that compensation would be more heavily weighted toward the pay-for-
performance variables. Instead we find that CEO compensation is strongly related to both
organizational size variables. CEO compensation is weakly tied to administrative efficiency
and is not significantly related to growth in contributions. From these results, we conclude
that their compensation is not closely tied to classic pay-for-performance measures. In
regards to free cash flows, we find that the remuneration of CEOs of health organizations is
the most sensitive of the six sub-sectors to increases in the commercial revenue share and is
quite sensitive to the relative size of the endowment. We found no significant relation
between CEO compensation and liquidity. Overall, the organizational size variables explain
20
a greater portion of the variation in pay than the pay-for-performance and free cash flow
variables combined.
CEO compensation in the human services and “other” sub-sectors exhibit
considerable similarities in the magnitude of the coefficients. Both exhibit a fixed
component to compensation of about $35,000, with the component being statistically
significant for the other sector. Total program expenses are significantly related to
compensation, with a $10-$11 gain in compensation for each $1,000 increase in program
expenses. In neither case are total fixed assets significantly associated with remuneration.
CEOs in both sub-sectors can expect to be financially rewarded for greater administrative
efficiency and when the share of commercial revenue is higher and the relative size of the
investment portfolio is larger. One striking difference is that CEOs in the other sub-sectors
receive substantially higher compensation when contributions are increased, while CEOs of
human service providers receive significantly lower compensation when liquidity is higher.
In both sectors, the organizational size variables had higher power to explain compensation
than the other two variable groups combined.
Compensation for religious leaders differs substantially from pay in the other
sectors. First, the “base” pay and both organizational size variables are insignificant. In the
area of pay-for-performance, their compensation is not related to growth in contributions.
More unusually, it is negatively related to administrative efficiency. In one regard, the
religious CEOs are similar to their counterparts: Their compensation is significantly
associated with the commercial revenue share and the relative size of the investment
portfolio. For CEOs of religious organizations, the size hypothesis was most strongly
supported, but it did not dominate the other two hypotheses combined.
21
V. Conclusions
Nonprofits operate to provide a public benefit, and most rely upon donations and
trust to carry out their work. Excessive nonprofit salaries or diversions of resources away
from services to clients can undermine public confidence, hurting not only nonprofit
organizations, but also the sector as a whole. The non-distribution constraint bearing on
nonprofit organizations provides a contractual assurance that the consumer will not be taken
advantage of or betrayed by producers for personal gains. Bound by this promise to use
resources to advance their missions rather than to benefit private parties, nonprofit
organizations emerge as a solution to market or “contract failures.” People seek out
nonprofits in areas where they cannot penetrate and police services using ordinary
contractual devices, in situations where trust and information are scarce, and assessing the
value of the services they receive for their money is difficult.
To better understand whether excessive compensation or violations of the
distribution constraint are frequent in the sector, we examined the factors associated with
CEO compensation. We found that nonprofit CEOs are paid a base salary, and many CEOs
also receive additional pay associated with larger organizational size. Our results indicate
that while pay-for-performance is a factor in determining compensation, it is not prominent.
In fact, in all the sub-sectors, CEOs compensation is more sensitive to organizational size
and free cash flows. While our analysis suggests that nonprofits may not literally be
violating the non-distribution constraint, we did find evidence that CEO compensation is
significantly higher in the presence of free cash flows. However, in only one sector
(education) did we find evidence that free cash flow is a central factor.
New IRS intermediate sanction regulations have recently been put in place to
penalize nonprofits that excessively compensate executives. These regulations determine the
22
reasonableness of executive compensation based on benchmarking against comparable
organizations. Our analysis suggests strong industry-specific similarities in pay are related
to free cash flows and, to a lesser extent, organizational size, rather than to performance.
Hence, the new regulations may not be particularly effective in identifying either absolute
levels of compensation that are too high or organizations that are violating the spirit of the
non-distribution constraint.
There are ultimately two ways to interpret our core results when it comes to the
coherence of nonprofit organizational identity. The first key finding is the strong effect of
organizational size on CEO compensation, implying that compensation is determined
largely by span of managerial responsibility. On this account, one might conclude that
nonprofits do a good job of protecting the coherence of the non-distribution constraint. The
second important insight is that free cash flows appear to have significant bearing on
nonprofit CEO compensation. In this finding, it is possible to see a crack in the moral high
ground that nonprofit organizations occupy by virtue of their pledge to eschew profit
distribution. While the presence of free cash flows and a highly liquid position may explain
less of the observed variance in compensation than organizational size, the significance of
the effect raises important questions about the coherence of a non-distribution constraint in
the nonprofit sector.
Those who are suspicious of the motives of charity workers will embrace this last
finding as proof that the motives and operational systems of businesses and nonprofits are
not so different and that the non-distribution constraint may be more fiction than fact. Those
disposed to trust nonprofit organizations will find some comfort that the non-distribution
constraint appears to limit incentive-based compensation. They may interpret the other
finding in one of two ways. First, nonprofits are generally cash-constrained but increase
23
CEO compensation in more flush periods to compensate nonprofit managers for prior years
work. Alternatively, to attract and retain good management, nonprofits compensate more
entrepreneurial managers that generate substantial commercial revenues and higher levels of
liquidity.
One final implication of our analysis bears on the enduring performance
measurement quandary that confronts so many nonprofit organizations. We believe that
nonprofit organizations may be relying on organizational size to make compensation
decisions and drawing on free cash flows when available rather than address the challenge
of defining, quantifying, and measuring the social benefits that nonprofits produce.
Nonprofit organizations typically produce services that are complex and that produce not
only direct outputs but also indirect, long-term, and societal benefits. These types of services
often make it difficult to both develop good outcome measures and establish causality
between program activity and client effects. In the absence of effective metrics of social
performance and mission accomplishment, many organizations rely on other factors in
setting compensation. Perhaps once better measures of mission fulfillment are developed
and actively implemented, nonprofits will be able to structure CEO compensation in ways
that provide appropriate incentives to managers who successfully advance the missions of
nonprofit organizations, while respecting the full legal and ethical implications of the non-
distribution constraint.
24
References
Agarwal, N., 1981, Determinants of executive compensation, Industrial Relations 20 (1), 36-46.
Alchian, A. A. and H. Demsetz, 1972, Production, information costs, and economic organization, American Economic Review 62, 777-795.
Antle, R. and A. Smith, 1985, Measuring executive compensation: Methods and application, Journal of Accounting Research 23, 296-325.
Baber, W. R., P. L. Daniel, and A. A. Roberts, 1999, Compensation to executives of eleemosynary organizations: An empirical study of the role of accounting performance measures, Working Paper, George Washington University.
Barbeito, C. L., and J. P. Bowman, 1998, Nonprofit compensation and benefits practices (New York, John Wiley & Sons, Applied Research and Development Institute International, Inc.).
Benston, G. J., 1985, The self-serving management hypothesis: Some evidence, Journal of Accounting and Economics 7, 67-84.
Boris, Elizabeth T. and C. Eugene Steuerle, 1999, Nonprofits and Government (Washington, DC, Urban Institute Press).
Borjas, G., H. Frech III, and P. Ginsburg, 1983, Property rights and wages: The case of nursing homes, Journal of Human Resources 1 (2), 103-110.
Brickley, J. and R. Van Horn, 2000, Incentives in nonprofit organizations: Evidence from hospitals, Working Paper, University of Rochester.
Deckop, J, 1987, Top executive compensation and the pay-for-performance issue, in: D.B. Balkin and L.R. Gomez-Mejia, eds., New Perspectives in Compensation (Englewood Cliffs, New Jersey, Prentice Hall) 285-293.
Drucker, P., 1992, Managing the nonprofit organization: Principles and practices (New York, Harper).
Fama, E. F., 1980, Agency problems and the theory of the firm, Journal of Political Economy 88 (2), 288-307.
Fama, E. F. and M. C. Jensen, 1983, Separation or ownership and control, Journal of Law and Economics 26 (2), 301-326.
Fama, E. F. and M. C. Jensen, 1983, Agency problems and residual claims, Journal of Law and Economics 26 (2), 327-350.
25
Finkelstein, S. and D.C. Hambrick, 1989, Chief executive compensation: A study of the intersection of markets and political processes, Strategic Management Journal 10,121-134.
Frank, R., 1996, What price the moral high ground? Southern Economic Journal 63, 1-17.
Frumkin, P. and A. Andre-Clark, 1999, Nonprofit compensation and the market, University of Hawaii Law Review 21 (3), 425-85.
Goddeeris, J., 1988, Compensation differentials and self-selection: An application to lawyers, Journal of Political Economy 84 (1), 411-428.
Gomez-Mejia, L.R., H. Tosi, and T. Hinkin, 1987, Managerial control, performance, and executive compensation, Academy of Management Journal 30 (1), 51-70.
Hallock, K., 2000, Managerial pay and governance in American nonprofits, Working Paper, University of Illinois, Champaign-Urbana.
Handy, F. and E. Katz, 1998, The wage differential between nonprofit institutions and corporations: Getting more by paying less?, Journal of Comparative Economics 26, 246-261.
Hansmann, H., 1980, The role of nonprofit enterprise, The Yale Law Journal 89, 835-98.
Hill, C.W. and P. Phan, 1991, CEO tenure as a determinant of CEO pay, Academy of Management Journal 34, 707-717.
Holstrom, B., 1982, Moral hazard in teams, Bell Journal of Economics 13, 324-340.
Huber, P., 1967, The behavior of maximum likelihood estimates under non-standard conditions, in Proceedings of the Fifth Berkeley Symposium in Mathematical Statistics and Probability, Vol.1, (Berkeley, California, University of California Press) 221-233.
Jensen, M. C. and W. J. Meckling, 1979, Theory of the firm: Managerial behavior, agency costs and ownership structures, Journal of Financial Economics 3 (4), 305-60.
Jensen, M. C., 1986, Agency cost of free cash flow, corporate finance, and takeovers, American Economic Review 76 (2), 323-329.
Jensen, M.C. and K.J. Murphy, 1990, Performance pay and top management incentives, Journal of Political Economy 98, 225-264.
Johnson, D. and G. Rudney, 1987, Characteristics of workers in nonprofit organizations, Monthly Labor Review 110 (7), 28-33.
Kerr, J.L. and R.A. Bettis, 1987, Board of directors, top management compensation and shareholder returns, Academy of Management Journal 30, 645-664.
26
Kerr, J.L. and L. Kren, 1992, Effect of relative decision monitoring on chief executive compensation, The Academy of Management Journal 35 (2), 370-397.
Kertz, C. L., 1997, Executive compensation dilemmas in tax-exempt organizations:
Reasonableness, comparability, and disclosure, Tulane Law Review 71 (3), 819-867.
Krashinsky, M., 1986, Transaction costs and a theory of nonprofit organizations, in: S. Rose-Ackerman, ed., The Economics of Nonprofit Institutions (New York, Oxford University Press).
Lakdawalla, D. and Philipson, T., 1999, The nonprofit sector and industry performance, Working Paper, University of Chicago.
Letts, C., W. Ryan and A. Grossman, 1999, High performance nonprofit organizations: Managing upstream for greater impact (New York: John Wiley and Sons, Inc.).
Light, P.C., 2000, Making Nonprofits Work (Washington, DC, Brookings).
Mason, D., 1996, Leading and managing the expressive dimension: Harnessing the hidden power source of the nonprofit sector (San Francisco, Jossey-Bass Publishers).
Meyer, J.W. and B. Rowan, 1977, Institutionalized organizations: Formal structure as myth and ceremony, American Journal of Sociology 83(2), 340-63.
Mocan, N. and D. Viola, 1997, The determinants of child care workers’ wages and compensation: Sectoral differences, human capital, race, insiders and outsiders, NBER Working Paper No. 6328.
Monsma, S., 1996, When Sacred and Secular Mix (Boston, Rowman and Littlefield).
Morck, R., A. Shleifer, and R.W. Vishny, 1989, Alternative mechanisms for corporate control, American Economic Review 79, 842-852.
Murphy, K., 1998, Executive Compensation, in: O. Ashenfelter and D. Card, eds., Handbook of Labor Economics, Vol. 3 (New York, North-Holland).
Oster, S., 1998, Executive compensation in the nonprofit sector, Nonprofit Management and Leadership 8 (3), 207-21.
Pappas, A., 1995, Reengineering your nonprofit organization (New York: Wiley).
Philipson, T. and R. Posner, 2000, Anti-trust issues in the nonprofit sector, Work in process, University of Chicago.
Preston, A., 1989, The nonprofit worker in a for-profit world, Journal of Labor Economics 7, 438-63.
27
Redling, E.T., 1981, Myth vs. reality: The relation between top executive pay and performance, Compensation Review 13 (4), 16-24.
Rich, J.T. and Larson, J.A., 1984, Why some long term incentives fail, Compensation Review 18 (1), 26-37.
Rogers, W., 1993, Regression standard errors in clustered samples, Stata Technical Bulletin 13, 88-94.
Roomkin, M and B. Weisbrod, 1999, Managerial compensation in for-profit and nonprofit hospitals, Journal of Law, Economics, and Organization 15 (3), 750-81.
Rose-Ackerman, S., 1986, Charitable giving and excessive fundraising, in: S. Rose-Ackerman, ed., The economics of nonprofit institutions (New York, Oxford University Press).
Ruhm, C. and C. Boroski, 2000, Compensation in the nonprofit sector, NBER Working Paper, No. 7562.
Samuels, D. G., and H. Pianko, 1998, Nonprofit compensation, benefits, and employment law (New York, John Wiley & Sons, Nonprofit Series).
Scott, W. R., 1995, Introduction: Institutional theory and organizations, in: W.R. Scott, ed., The Institutional Construction of Organizations (Thousand Oaks, California, Sage) xi-xxii.
Simon, H.A., 1957, Compensation of executives, Sociometry 20, 32-35.
Steinberg, R., 1990, Labor economics and the nonprofit sector: A literature review, Nonprofit and Voluntary Sector Quarterly 19 (2).
Tolbert, P.S. and Lynne G. Zucker, 1983, Institutional sources of change in the formal structure of organizations: The diffusion of civil service reform, Administrative Sciences Quarterly 28, 22-39.
Vuong, Q. H., 1989, Likelihood ratio tests for model selection and non-nested hypotheses, Econometrica 57 (2), 307-333.
Weisbrod, B. A., 1988, The Nonprofit Economy (Cambridge, Massachusetts, Harvard University Press).
Westfall, J.D. and E.J. Zajac,1994, Substance and symbolism in CEOs’ long-term inventive plans, Administrative Sciences Quarterly 39 (3), 367-390.
White. H., 1980, A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity, Econometrica 48, 817-38.
28
Williamson, O., 1983, Organization form, residual claimants, and corporate control, Journal of Law and Economics 26 (2), 351-366.
Zucker, L. G., 1988, Where do institutional patterns come from? In: L. G. Zucker, ed., Institutional Patterns and Organizations (Cambridge, Massachusetts, Ballinger) 3-21.
29
30
Table 1 Descriptive statistics
Panel A: Observations by SOI Panel YearYear Observations1999 24,5322000 9,238Total 33,770
Panel B: Observations by Industry ClassificationClassification Observations Firms
Arts 2,980 2,405Education 4,686 3,522
Health 7,155 5,793Human Services 13,063 10,553
Religious 1,763 1,579Other 4,123 3,467Total 33,770 27,319
Panel C: Characteristics by Industry Classification Human
Arts Education Health Services Religious OtherTotal Fixed Assets (in Thousands of Dollars)Mean 1,681 9,344 7,804 1,476 532 1,649Median 19 91 170 79 8 15St. Deviation 10,500 57,000 31,900 10,800 4,415 24,000Total Program Expenses (in Thousands of Dollars)Mean 357 1,531 2,621 331 175 461Median 69 148 198 90 26 69St. Deviation 1,242 9,719 10,800 1,788 1,067 2,741Administrative EfficiencyMean 79.05% 82.78% 83.83% 84.76% 82.67% 82.47%Median 83.09% 86.02% 86.95% 87.53% 88.20% 86.77%St. Deviation 18.83% 15.19% 14.95% 14.27% 20.84% 17.29%Dollar Growth in Contributions (in Thousands of Dollars)Mean 99 348 198 152 66 428Median 6 0 3 9 4 4St. Deviation 2,637 5,209 5,047 2,141 1,108 5,763Commercial Revenue ShareMean 37.89% 52.20% 52.48% 40.11% 20.75% 29.92%Median 32.34% 61.28% 58.14% 25.34% 2.12% 11.36%St. Deviation 31.76% 38.66% 40.84% 40.46% 31.18% 35.80%Liquid Assets/ExpensesMean 72.14% 85.05% 42.70% 84.88% 39.16% 67.66%Median 25.91% 22.78% 18.49% 18.44% 14.27% 29.57%St. Deviation 375.05% 1194.60% 162.07% 3187.36% 141.82% 261.20%Investment Portfolio/Total AssetsMean 11.26% 17.54% 10.55% 6.54% 4.60% 10.45%Median 0.00% 0.00% 0.00% 0.00% 0.00% 0.00%St. Deviation 23.88% 27.72% 22.38% 17.56% 15.07% 23.25%
Table 2Means, Standard Deviations, and Correlations
Mean St. Dev 1 2 3 4 5 6 7 8 9 101. Total CEO Compensation 74,066 95,271 1.00
2. CEO Salary 67,287 77,256 0.94 1.003. CEO Benefits 6,779 34,415 0.65 0.36 1.004. Total Fixed Assets (per $1,000)
3,898 28,358 0.37 0.37 0.20 1.00
5. Total Program Expenses (per $1,000)
993 6,422 0.45 0.43 0.28 0.71 1.00
6. Administrative Efficiency 83.39% 15.85% 0.00 0.00 0.00 0.02 -0.04 1.00
7. Dollar Growth in Contributions (per $1,000)
214 3,959 -0.01 -0.01 -0.01 0.01 -0.05 0.96 1.00
8. Commercial Revenue Share 41.95% 39.65% 0.08 0.07 0.06 0.12 0.11 0.02 0.02 1.00
9. Liquid Assets/Expenses 70.38% 2038.34% 0.01 0.01 0.00 0.01 0.00 0.03 0.04 0.00 1.00
10. Investment Portfolio/Total Assets
9.71% 21.78% 0.00 -0.01 0.00 0.00 0.00 -0.05 -0.05 0.02 0.01 1.00
Table 3Compensation Analysis by Type of CompensationPanel A: Regression Analysis
Directional Prediction
Total CEO Compensation
CEO Salary
CEO Benefits
Constant + 33,270.96*** 33,434.89*** -163.93Total Fixed Assets (per $1,000) + 0.27* 0.27*** -0.01
Total Program Expenses (per $1,000) + 5.33*** 3.89*** 1.45*
Administrative Efficiency + 10,659.50*** 8,041.88*** 2,617.62**Dollar Growth in Contributions (per $1,000) + 0.63 0.41 0.22
Commercial Revenue Share + 32,007.09*** 27,858.92*** 4,148.17***Liquid Assets/ Expenses + -1.81 -4.14 2.33Investment Portfolio/ Total Assets + 62,657.94*** 52,261.62*** 10,396.32***
Plus: Primary Metropolitan Statistical Areas
Adjusted R2
0.258 0.248 0.091Observations 33,770 33,770 33,770Number of Firms 27,319 27,319 27,319
* p-value (two-sided) < .10 ** p-value (two-sided) < .05 *** p-value (two-sided) < .01The p-values in the regressions are computed using White’s robust standard errors (White [1980]). In addition, the estimator of variance used assumes the observations are not independent but that they are divided in M groups (i.e., firms)
G1, G2, ..., GM that are independent. Specifically, the estimator is . , where = (2ln L/2)-1 and
uk(G) is the contribution of the kth group to the scores ln L/ (Huber [1967] and Rogers[1993]).
Panel B: Relative Explanatory Power using Vuong (1989)The variables are grouped according to the following hypotheses:
H1: Total Fixed Assets, Total Program Expenses H2: Administrative Efficiency, Dollar Growth in Contributions H3: Commercial Revenue Share, Liquid Assets/Expenses, Investment Portfolio/Total Assets
H1
vs. H2H1
vs. H3H2
vs. H3H1 vs.
H2 & H3
Total CEO Compensation 4.31*** 3.12*** -12.09*** 2.97***CEO Salary 5.06*** 3.48*** -15.40*** 3.33***CEO Benefits 2.09** 1.82** -2.78*** 1.68*
* p-value (two-sided) < .10 ** p-value (two-sided) < .05 *** p-value (two-sided) < .01
z-statistics resulting from Vuong Test (1989). Positive/(negative) values indicate that the first/(second) group of variables explains significantly more of the variance in compensation.
Table 4Compensation Analysis by Major Industry Classifications
Panel A: Regression Analysis Arts Education Health Human
ServicesReligious Other
Constant 33,187.02*** 42,037.14*** 25,434.33** 35,792.80*** 54,251.60*** 34,459.72***Total Fixed Assets (per $1,000) 0.70 0.38*** 1.36*** -0.76 -0.92 -0.14
Total Program Expenses (per $1,000) 28.04*** 0.87 3.85** 11.02*** 19.85 10.21***
Administrative Efficiency 8,706.56* 1,464.20 22,092.82* 9,247.86** -22,082.46* 13,084.30**
Dollar Growth in Contributions (per $1,000)
-0.06 1.27*** 0.18 0.91 1.41 0.71**
Commercial Revenue Share -9,884.12*** 37,869.40*** 56,175.75*** 12,117.66*** 24,018.88*** 17,340.44***
Liquid Assets/ Expenses 499.35* 272.54** -918.90 -18.35*** -767.51 -413.04Investment Portfolio/ Total Assets 32,107.96*** 85,356.32*** 66,527.01*** 48,018.78*** 25,834.45** 48,255.37***
Plus: Primary Metropolitan Statistical AreasAdjusted R2 0.451 0.245 0.370 0.145 0.134 0.252Observations 2,857 4,595 7,047 12,842 1,763 4,123Number of Firms 2,307 3,453 5,707 10,395 1,579 3,467
* p-value (two-sided) < .10 ** p-value (two-sided) < .05 *** p-value (two-sided) < .01The p-values are computed using White’s robust standard errors (White [1980]). In addition, the estimator of variance used assumes the observations are not
independent but that they are divided in M groups (i.e., firms) G1, G2, ..., GM that are independent. Specifically, the estimator is , where =
(2ln L/2)-1 and uk(G) is the contribution of the kth group to the scores ln L/ (Huber [1967] and Rogers[1993]).
Table 4 (Continued)
Panel B: Relative Explanatory Power using Vuong (1989)
The variables are grouped according to the following hypotheses: H1: Total Fixed Assets, Total Program Expenses H2: Administrative Efficiency, Dollar Growth in Contributions H3: Commercial Revenue Share, Net Income/Total Revenues, Investment Portfolio/Total Assets
H1 vs. H2
H1 vs. H3
H2 vs. H3
H1 vs. H2 & H3
Arts 9.37*** 8.34*** -2.56*** 8.19***Education 4.17*** 1.26 -5.13*** 0.61Health 7.31*** 5.60*** -10.08*** 5.48***Human Services 4.26*** 1.49* -5.11*** 1.32Religious 2.30* 1.65* -1.26 1.55*Other 4.26*** 3.42*** -2.52** 2.94***
* p-value (two-sided) < .10** p-value (two-sided) < .05*** p-value (two-sided) < .01
z-statistics resulting from Vuong Test (1989). Positive/(negative) values indicate that the first/(second) group of variables explains significantly more of the variance in compensation.
