Content uploaded by Jeremy Thornton
Author content
All content in this area was uploaded by Jeremy Thornton on Oct 29, 2015
Content may be subject to copyright.
Electronic copy available at: http://ssrn.com/abstract=1114785
Nonprofit Fundraising in Competitive Donor Markets
Jeremy Thornton*
Assistant Professor of Economics
Samford University
800 Lakeshore Dr.
Birmingham AL, 35229
(205) 726-2128
jpthornt@samford.edu
JEL Codes: L13, L31
Keywords: Nonprofit, fundraising, market structure, efficiency
* Special thanks to John Garen, Frank Scott, Marco Castaneda, along with two anonymous referees for their
helpful comments on this paper.
Electronic copy available at: http://ssrn.com/abstract=1114785
Abstract:
Fundraising expenditures represent an important strategic decision for nonprofit
managers in the face of scarce donor resources. Privately, nonprofit managers weigh the
tradeoff between reaching new donors and increasing the implicit price of output to its
constituents. Socially, competition among nonprofit firms for donations may produce an
excessive level of fundraising. This paper empirically examines nonprofit fundraising
decisions, both privately and socially, under varying market conditions. Analysis of
financial data reveals that as markets become more competitive, nonprofits follow their
private incentives by reducing their fundraising expenditures. However, I find evidence
that, collectively, nonprofits may spend an inefficiently high share of their revenues on
fundraising. As such, I offer alternatives to the common practice of collective fundraising
through institutions like the United Way. Implications of the study include increasing
price transparency to improve market discipline or raising legal and financial barriers to
entry.
Jeremy Thornton, Ph.D., is an assistant professor of economics at Samford University.
His research focuses on the impacts of market competition and governance on nonprofit
firm behavior.
I. Introduction
For the period ranging from 1982 – 1997, the number of 501(c)(3) nonprofits has
increased 128% while private contributions increased only 72% in real terms.1 For the
majority of nonprofit organizations that rely on private contributions, fundraising remains
the primary mechanism of inter-agency competition for scarce donor resources.
Fundraisers mix personal visits to donors, grant writing, telethons, websites, and direct
mail to maximize donation revenues at the lowest cost. It is, however, difficult for
nonprofit managers to determine how much of their own resources to allocate toward
fundraising activities.
Fundraising has positive effects for both donors and nonprofits. For nonprofits,
fundraising messages create awareness and attract charitable gifts to specific programs.
For donors, fundraising messages provide valuable information about specific
characteristics of the nonprofit, reducing the cost of finding their preferred charity. In
theory, the optimization problem for the nonprofit manager is straightforward. Any
nonprofit should keep fundraising until the last dollar spent returns only one dollar in new
donations.2
However, it is a distinct question whether nonprofit managers, collectively,
fundraise too much. Generally, altruistic managers are concerned with the overall
provision of charitable services, not just the success of their own organization. It is
important to recognize that fundraising activities could either increase or reduce overall
charitable output. Any donor that has received multiple solicitations from similar types
of charities can sympathize with the dilemma. As the sophistication of fundraising
1
technologies evolves, the probability of any single donor receiving overlapping
solicitations by nearly identical organizations will increase.
From the perspective of the nonprofit, multiple solicitations from competing
nonprofits represent a lost opportunity. Instead of reaching new constituents, donors may
simply switch between charities without any increase in overall funding. For the donor,
multiple solicitations from similar organizations represent wasted resources. Fundraising
is costly, siphoning resources that could have been spent on charitable output. This
remains a perplexing problem for a nonprofit manager if the private fundraising decisions
lead to a glut of solicitation messages in the market. As a consequence, nonprofit
managers often coordinate their fundraising activities to reduce aggregate fundraising.
This is the broad intention of supporting organizations like the United Way, whose
purpose is to coordinate and suppress wasteful fundraising. It is less clear whether such
actions are warranted or beneficial. This paper approaches the issue systematically by
examining both the private and social impact of market competition on the fundraising
decisions of nonprofit firms. Using an economic model of for-profit advertising, the
paper defines a framework for determining if nonprofits are fundraising excessively.
Historically, academic economic research has portrayed nonprofit firms as passive
recipients of donations from altruistic donors. Foundational papers such as Bergstrom et.
al. (1986) characterize nonprofit institutions as docile mechanisms for the provision of
public goods. In this class of models, donors have some preference for a public good
which can be satisfied by either private or public provision. Nonprofit organizations are
merely a passive mechanism by which those preferences are realized.3
2
This view contrasts with the popular management literature marketed to nonprofit
directors. Titles like Play to Win: A Nonprofit Guide to Strategy or Mission Based
Marketing: How Your Nonprofit can Succeed in a More Competitive World portray an
ominous environment where firms musts compete to survive. More consistent with this
perception are papers that examine strategic decisions made by nonprofits in the face of
scarce donor resources. These papers find that nonprofit firms compete on a variety of
margins including efficiency, quality or fundraising. Yet, modeling these decisions has
remained an underdeveloped area of economic literature. A primary constraint to
empirical analysis remains the obscure nature of the nonprofit firm objective when
operating under a non-distribution constraint. Various papers have approached this issue
differently. In this section, I offer a brief summary of both the theoretical and empirical
literature examining competition among nonprofit firms.
Review of Nonprofit Competition
Relative to their for-profit counterparts, the particular incentives guiding nonprofit
managers are not well understood by economists. The non-distribution constraint implied
by nonprofit status precludes many of the basic incentives and governance mechanisms
relied upon to understand and predict firm behavior. Typically, faith has been placed in
the altruistic preferences of nonprofit managers to operate in the best interest of its
constituents. The following set of papers indicates that assumption may not be well
founded. There is little reason to think that nonprofit organizations pursue socially
optimal goals any more vigorously than public or for-profit institutions.
Glaeser (2003) demonstrates that nonprofit organizations, typically wealthy ones,
will likely conform to objectives of elite workers, rather than donors or other
3
constituents. In his introduction, Glaeser suggests that market competition remains the
primary constraint on nonprofit manager behavior to align firm objectives with those of
its constituents. However, the empirical evidence demonstrating that donors actually have
a significant influence on the decisions of the nonprofit firm has been mixed.
Donors are able to observe a variety of efficiency ratios publicized by watchdog
groups and the nonprofits themselves. Nonprofit organizations often report their
fundraising-expense ratio which represent the fraction of total expenses allocated to
fundraising. A common assumption is that donors have preferences for charitable output.
Consequently, they perceive management or fundraising expenditures as an implicit
“price” of production.4 Weisbrod & Dominguez (1986) originally tested the notion that
donations depend on both the fundraising expenditures and the subsequent price of
charitable output. Using a pooled sample of Form 990 data, they found that higher
solicitation expenditures indeed boost donation revenue to nonprofits. As a secondary
effect, their tests also determine that donors dislike higher fundraising costs and reduce
their contributions when nonprofit organizations are less efficient.
In a follow-up paper, Okten & Weisbrod (2000) confirm that donors are indeed
sensitive to expense ratios, implying a downward sloping demand curve for nonprofit
services. This paper refines earlier work by using panel data from seven industries
covering more than a decade of observations. Their results are similar to the previous
paper, in that the direct effect of fundraising is to increase charitable gifts. However, the
indirect affect of a higher price results in lower donation revenue. Additional findings
from the paper indicate that nonprofits are often not fundraising at an efficient level.
4
Some industries, such as libraries and hospitals, do not fundraise enough to maximize net
revenues, while religious organizations appear to fundraise too much.
Drawing solely from Pennsylvania data, Greenlee & Brown (1999) examine both
fundraising and administrative expenditures. Consistent with prior results, Greenlee &
Brown find a significant and inverse relationship between administration expense ratios
and total contributions. The implication of the study is that donors dislike both
fundraising and administrative expense ratios. In contrast, Frumpkin & Kim (2001) find
no statistically significant relationship between administrative expense ratios and public
donations. The paper draws the conclusion that operational efficiency is not rewarded by
donors. The authors promote the idea that nonprofit organizations should not focus on
reducing non-program expenses, rather firms should optimize their fundraising. Given
this conflicting evidence, it would seem difficult for a nonprofit manager to know what to
do.
A more careful review of the literature reveals that the studies cannot be
compared directly. Frumpkin & Kim (2001) use a national representative panel while
Greenlee & Brown (1999) draw from a much smaller, stately specific sample. The
structure of the empirical model also plays a crucial role. Tinkelman & Mankaney (2005)
examine under what conditions an inverse relationship between price and donation
revenue appears. They find that the empirical specification used, particularly if program
spending is included as a control variable, significantly alters the slope coefficient on
price. On average, relatively large and more established organizations that rely most
heavily on donations appear to face the most significant price elasticity.
5
Even if donors can limit fundraising by reducing gifts in the face of high expense
ratios, this does not guarantee that nonprofits will provide the socially optimal amount of
fundraising. Rose-Ackerman (1987) was first to articulate the dilemma faced by nonprofit
managers. Managers know that increased fundraising encourages new donations.
However, donors may also perceive fundraising expenditures as a cost, diverting
resources from charitable output. The paper develops a set of theoretical models which
demonstrate that, by responding to private incentives, nonprofit organizations pursue
fundraising at levels that may reduce aggregate service provision. Also find that
competition for donations may force fundraising to inefficiently high levels, even when
donors have a strong preference against the use of resources for fundraising.
Empirical research on the interaction of fundraising expenditures and competition
solely among nonprofits is thin.5 The most related paper, Feigenbaum (1987) generates
its own theoretical framework as it examines the competition among medical research
charities for donations. The paper concludes that as market become more concentrated,
implying less competition, charities tend to fundraise less. A ten point decline in the four-
firm concentration ratio resulted in a two cent per dollar increase in solicitation
expenditures. In addition, this rise comes from competition forcing resources away from
other perquisite consumption. However, general conclusions are difficult to draw from
the study. The sample covers only a narrow range of charities that are not necessarily
representative of the nonprofit sector. This paper builds on that original notion of
empirically examining competition among nonprofit organizations.
In the subsequent analysis, nonprofit fundraising serves a similar role as for-profit
advertising. This notion is less controversial than it sounds. The application of for-profit
6
models to a nonprofit context is not new. Diamond & Gooding-Williams (2002) use an
advertising model adapted from consumer research literature to help explain direct
fundraising appeals. Malani & Choi (2004) address the issue more directly by
demonstrating that nonprofit compensation structures in hospitals promote similar goals
as for-profits. In each case, authors successfully leverage the predictive power of for-
profit models to offer insight to the incentives and constraints guiding nonprofit
managers.
II. Theory
Attention surrounding nonprofit fundraising has typically focused on operating
efficiency of the firm. While this term can take on a variety of meanings, it most often
implies the ability of fundraising activities to increase resources available for the
organization, net of its costs. Cordes & Rooney (2004b) differentiate between private
and social fundraising efficiency.6 Foremost, nonprofit firms will respond to their own
private incentives to maximize donation revenue, net of costs. In contrast, social
efficiency seeks to determine if firms are, in the aggregate, offering an amount of
fundraising which maximizes total resources allocated to a charitable cause. These are
precisely the questions addressed in a for-profit context in Grossman and Shapiro (1984).
Rather than reconstruct the derivations of the model, I only note its parameters within the
specific context of a nonprofit firm. For a complete description of the model, see the
original paper.
To be consistent with the original model, fundraising messages offer only
information to donors rather than attempt to alter their preferences. It is easiest to
consider nonprofits mailing a representative brochure that provides information on
7
existence, price (in the form of expense ratios), and service characteristics.7 This
information is crucial because each nonprofit offers a slightly different product. Product
differentiation in the nonprofit context can be driven by several factors. Ideology,
methodology, or targeted beneficiaries can differentiate an organization’s product or
service. In practical terms, this implies that a donor may prefer Baptist churches to
Pentecostal, or domestic anti-poverty initiatives to foreign. Donors will choose the
nonprofit that most closely match their own preferences. Intuitively, the returns to a
fundraising message are lower when more firms are in the market. Because of increased
competition, donors are more likely to find a firm that more closely matches their own
preferences. Therefore, a rational firm will reduce fundraising expenditures with
increased entry.
The model also considers the social welfare implications of solicitation
expenditures given the above conditions. The original model demonstrates that
monopolies will under-provide informative advertising. In the nonprofit context, the
private incentive to fundraise for the firm will fall below the social benefit. Monopoly
providers will cease fundraising before everyone in the market has benefited from the
informative messages from the nonprofit. To understand the social welfare implications
of other market structures, two factors must be considered.
Whether the market will over/under provide fundraising depends on two effects.
The beneficial effect of fundraising is that it matches donors with nonprofits which more
closely fit their own ideological preferences (i.e. lowers transportation costs). Grossman
& Shapiro (1984) describe this as the matching effect. The negative aspect of fundraising
is that organizations generally do not take into account the revenue reduction from other
8
firms when they steal potential donors away from their competitors with their own
solicitation. This is called the customer-capture effect. In the context of the model, the
wasteful capture effects will dominate the beneficial matching effects once the market
expands beyond a small number firms.
If the wasteful effects of advertising dominate then, as new firms enter, aggregate
solicitation expenditures for the entire market should increase. This implies that the
additional solicitation expenditures from new firms rise more quickly than per firm
solicitation expenditures decline. This result would be consistent with the expectation that
the market is providing too much fundraising relative to the social optimum.
Interestingly, the model demonstrates that a social planner would reduce the total number
of firms, yet increase the fundraising intensity of those remaining firms.
In summary, there are two testable implications that can be drawn from the
Grossman & Shapiro model as it has been applied to nonprofits. First, fundraising
intensity should fall as new firms enter the market. The private incentive of the nonprofit
firm is to restrict fundraising expenses in the face of new competition. From a social
welfare perspective, the market will over-provide product diversity with too many firms.
Holding market size constant and per-firm fundraising intensities fall with entry. Yet,
aggregate fundraising expenditures will rise with entry. To the extent that the latter effect
dominates, nonprofit firms will fundraise too much. In the following section, I construct
an empirical model and dataset to examine these two implications.
III. The Data
With two important exceptions, every operating 501(c)(3) charity is required to
file a Form 990 to the IRS.8 This document contains key financial information including
9
revenues, expenses, and balance sheet information. The National Center for Charitable
Statistics (NCCS) takes some of these variables and aggregates them into annual datasets.
These files contain a variety of financial measures for the full population of charities
required to file the Form 990. NCCS Core Files 1990 thru 2000 have been included in the
study, implying an eleven year panel.9
Operating charities have subsequently been organized by the NCCS into a
detailed classification system called the National Taxonomy of Exempt Entities (NTEE).
The taxonomy operates similarly to for-profit SIC codes, where broad classes of
organizations can be broken apart according to the desired level of aggregation. To
examine the impact of competition, it is necessary to restrict our attention to those classes
of charities within the overall population of charities who compete within relatively well-
defined markets. Consequently, three digit NTEE categories were filtered through a
series of selection criteria. Those firms within the three digit classifications which
matched the following criteria were kept in the sample.
1. The charities within the classification are local, in terms of consumption
of output and source of donations.
2. The charities within the classification are reasonably homogeneous across
geographic markets.
3. The charities within the classification provide services which are
substantially distinct from for-profit firms.
4. The charities within the classification should receive a non-trivial fraction
of their revenues from donations.
10
<Insert Table 1 about here>
From these criteria, a list of 16 sub-sectors was distilled and listed in Table 1. The
selection provides a broad cross-section of nonprofit firms ranging from art organizations
to human service providers. Firms from these sub-sectors were then organized into
geographic markets based on Metropolitan Statistical Area (MSA). Only firms located
within a MSA were included in the dataset. A total of 340 MSAs are possible for each
sub-sector. Each sub-sector/MSA paring represents a distinct market. For example, in the
year 2000, the dataset contains 30,392 firms dispersed among 2,440 markets.10
Measures Market Structure
The theoretical model predicts that per-firm fundraising expenditures will decline
with entry. To test this prediction, a standard measure of market structure is used. The
Herfindahl-Hirschman Index (HHI) was calculated based on total revenues and
summarized in Table 2.11 When examining HHI, fourteen of the sixteen sectors trend
toward greater competitiveness (lower HHI) over the sample time periods. Furthermore,
firms in this sample operate in relatively concentrated markets with HHI regularly
exceeding 2000.12
<Insert Table 2 about here>
Measures of Fundraising
The dollar value of fundraising expenditures is the most analogous metric for
fundraising intensity described in the theoretical model, though others are possible.13
This value is reported on the Form 990 and is summarized by sector in Table 3. The
11
distribution of solicitation expenses across nonprofit sectors follow expectations. For
example, museums spend far more on fundraising than do family counseling or senior
citizen centers. Much of this variation is due to the size of the representative organization.
However, the propensity to fundraise also roughly tracks the sectors dependence on
donor contributions relative to their overall revenues.
<Insert Table 3 about here>
To analyze the welfare effects of fundraising under competition, the aggregate
expenditure on solicitation per market is also calculated. This was done by summing the
individual solicitation expenditures for every firm within a market. Aggregate fundraising
expenditures per market also vary widely by sub-sector. The average total solicitation
expenditure in dollars per market is given by Table 4.
<Insert Table 4 about here>
As an additional test, I examine the ratio of fundraising expenditures to public
contributions at the market level. Table 5 summarizes this information by sub-sector.
Note the sizable variation, both within and across sub-sectors, for the average return on
fundraising. Food Pantries have the highest return, earning an average of $162 on each
dollar spent on fundraising. Museums have the lowest return in the sector, averaging
only $19 per fundraising dollar.
12
<Insert Table 5 about here>
Measurement Issues
An important issue demonstrated by the previous tables is the suspicious ability of
many nonprofit firms to receive significant public donations without spending anything
on solicitation to attract those gifts. Hager et. al. (2002) found that 59% of nonprofits
receiving some type of public support report zero fundraising expenses. 14 Furthermore,
nearly one-quarter of those “zero-cost fundraising” organizations received 5 million in
contributions or more. It would seem suspect that a sizable fraction of charities would be
able to raise substantial donation revenue so effortlessly. Despite specific AICPA
guidelines, nonprofit organizations have demonstrated substantial latitude in reporting of
management and fundraising expenditures. Because the IRS does not collect tax revenue
on profits, there has been little effort to enforce strict accounting policies (Government
Accounting Office, 2001).15 Even with the serious potential for measurement error in
expense reporting, it is still possible to characterize the nature of the bias.
To the extent that the measurement error is random, standard errors will rise, but
point estimates of the coefficients will be unbiased. This makes statistical significance
more difficult to detect. It is, however, more likely that management and general
expenses are systematically underreported. To the extent that donors regard management
and fundraising expenditures as a price, nonprofit managers have an incentive to
underreport overhead expenditures. Systematic underreporting of overhead will cause
bias in the intercepts, but slope coefficients will remain unbiased.
Hager et. al. (2002a) hypothesize that many nonprofit firms have some latent (but
underreported) capacity for fundraising. This fundraising capacity may include time the
13
executive director spends talking to supporters, the efforts staff spend managing
incoming funds, or institutional relationships with similar organizations. Regardless,
nonprofit managers may not perceive these duties as ongoing fundraising activities and
therefore, do not record their opportunity cost appropriately. The implication of this
reasoning is that fundraising expenses are censored below a certain threshold because
managers do not recognize the true opportunity costs of their fundraising inputs. To the
extent that reporting zero solicitation is a censoring problem, two alternative techniques
are used.
In addition to using OLS for the full sample, two alternate specifications are used
to account for the impact of underreporting fundraising expenditures. First, only those
firms who have positive solicitation expenditures throughout the panel are kept within the
sample. This limits our attention only to firms who are active fundraisers. However, this
approach may lead to biased and inconsistent estimates for the sample as a whole. As a
secondary option, a Tobit model was used. The Tobit estimation accounts for censored
dependent variables, producing consistent parameter estimates.
IV. Empirical Model
Panel data on fundraising expenditures is used to estimate the empirical
relationship between market structure and fundraising intensity, both within and across
markets. I use the following empirical specifications for firm i in year t for market j:
itjtttit
itit
jt
jtit
MSASECTORTCONT
ASSETSAGEHHIHHISOLICIT
εββββ
βββββ
+++++
++++=
−87615
43
2
210
14
The dependent variable, SOLICIT is the per-firm dollar expenditure on
fundraising as stated by the Form 990. Solicitation expenditures are regressed on
Herfindahl-Hirschman index (HHI). To control potential differences in scale, total assets
(ASSETS) of the firm was included. Furthermore, the age (AGE) of the firm was added
to proxy for the impact of reputation across firms. Lagged total contributions (CONT)
was included to account for the impact of prior government or private donations on
fundraising decisions. Additional controls include a vector of time and sub-sector
dummies. The annual time dummies control for macroeconomic, time varying,
disturbances while the sub-sector dummies control for time invariant differences across
sub-sectors.
Finally, there are potential embedded differences across MSA geographic
markets. To control these possible variations across geographic markets, two approaches
were taken. First, a basic set of demographic controls were chosen based on
characteristics that have been demonstrated to effect donor giving. POPULATION,
PER-CAPITA INCOME, percent of the population with a high school degree
(EDUCATION), percent of the population which is African-American (BLACK), and
percentage of the population which is Hispanic (HISPANIC) have each been shown to
impact overall giving (ARRFC, 2002). A full set of indicators is included for each
Metropolitan Statistical Area. Alternatively, dummy variables were included to control
possible systematic differences across markets. However, this approach substantially
reduced variation, potentially masking important effects. The models are estimated first
by OLS.
15
To address the potential measurement error, two separate tests for robustness are
considered. Following Andreoni (2003), the model was estimated again after dropping all
the observations where SOLICIT = 0. Removing these firms focuses attention on the
relationship between fundraising and market structure for only those firms which spend
positive amounts on solicitation. Alternatively, a Tobit estimation technique is used on
the full sample. This approach compensates for the substantial censoring problem with
nonprofits reporting zero fundraising expenditure.
To test the welfare implications of the theoretical model, I replace aggregate
market solicitation expenditures as the dependent variable. Instead of individual firm
observations, each market is now an individual observation. The empirical model is
specified as follows for market j in time t:
jttt
jtj
jt
jtjt
SECTORT
MRKTCONTMSAHHIHHIS
εββ
βββββ
+++
+++++=
65
43
2
210
Where Sj = , or aggregate solicitation expenditures for each
market. The dependent variable is regressed on the Herfindahl-Hirschman Index (HHI).
In order to control fundamental differences across markets, MSA demographic
characteristics are again used. MRKTCONT represents aggregate contributions within an
entire market in the time period to control for market size. Finally, as before, a set of time
(T) and 16 sub-sector dummies are included (SECTOR). As an additional measure of
market structure the actual number of firms (n) is also tested. This has the advantage of
being closely related to the theoretical model, yet doesn’t account for the distribution of
revenues within a market. The use of n as the market structure variable may have the
∑
=
n
ii
SOLICIT
1
)(
16
effect of over-emphasizing the competitive impact of new firms entering the market. The
model is estimated using OLS.
V. Results
Table 6 offers a summary of regression results for per-firm solicitation
expenditures. Estimation results for the impact of market concentration were similar
across specifications. An increase in market concentration results in a positive and
statistically significant increase in per-firm fundraising (solicitation) expenditures.
Regressions (1) and (2) retain the full sample of nonprofit firms totaling nearly 12,000
unique firms. Regression (1) uses market level fixed effects, while regression (2) proxies
for important MSA variation uses demographic data. Results indicate that for low levels
of HHI (atomistic markets), a one point increase in HHI results in a seven dollar increase
in solicitation expenditures. As markets become increasingly concentrated (higher HHI),
solicitation continues to grow at a declining rate until there are only two firms in the
market. Per-firm solicitation expenditures decrease slightly for monopoly firms. These
results are consistent with for-profit literature examining the interaction between market
structure and advertising.16
< Insert Table 6 about here >
Regressions 4 and 5 offer results when “zero cost fundraisers” are dropped,
reducing sample size to around 8,000 unique firms over the panel. The relationship
between market structure and fundraising remains positive and statistically significant
over the relevant range. Coefficients on market structure increase in magnitude slightly
using MSA dummies relative to demographic controls. The final test for robustness uses
17
a Tobit model to compensate for the potentially censored dependent variable. The
relationship between market concentration and fundraising expenditures is again positive
and statistically over the relevant range. The magnitude of the coefficients on HHI are
nearly identical to the OLS estimation.
Results in this case confirm theoretical predictions that fundraising intensity
should increase with market power. The magnitude of the estimated coefficients would
indicate that market structure plays an economically important role in determining
fundraising intensity. For example, moving from a relatively competitive market of ten
firms to an oligopoly market of four firms implies an increase in solicitation expenditures
of $7,487 per firm. Moving from this oligopoly to a duopoly implies an increase of over
$7,189. To put this in context, the average fundraising expenditure for the entire sample
is slightly over $50,000.
A variety of alternative specifications were tested. HHI was lagged by one period
to account for potential delays in reacting to new entrants. This had negligible effects on
the regression coefficients. To be more consistent with the theoretical model, the number
of firms in the market n was used instead of the more popular measure of market
structure, HHI. Signs remained consistent with theoretical predictions, yet statistical
significance suffered. It is likely that much of the variation in market structure is
manifested in shifting market share rather than new entry. This makes the impact of new
entry difficult to detect. A random effects model was also tested, however subsequent
Hausman specification tests indicated that that a fixed effects estimation was preferable.
Finally, using a simple Chow test, I fail to reject the hypothesis that the slope coefficients
18
are similar across the sixteen nonprofit sub-sectors. This implies that the sectors can be
pooled with reasonable confidence.
The second issue addressed by the model is whether additional fundraising
expenditures, in the aggregate, lead to additional donation revenues for the market.
Because marginal fundraising expenditures are not observed, it is difficult to determine if
nonprofit markets are over-providing fundraising messages. Instead, it is possible to
observe if these nonprofit markets have characteristics which follow the directional
predictions of the model. Table 7 gives the results for the impact of market concentration
on fundraising expenditures. Regression 6 demonstrates there is an inverse and
statistically significant relationship between market concentration and fundraising. This
result indicates that the additional fundraising expenditures of a new firm entering the
market overwhelms the effect of the reduction in per-firm fundraising. This is the
necessary condition outlined in the theoretical for excessive fundraising in a market.
Specifically, increasing market concentration by one point on the HHI would result in
approximately 2,123 dollar decrease in aggregate fundraising. Mirroring the results from
per-firm fundraising, the relationship becomes slightly positive once monopoly is
reached. Recalling that HHI and n move in opposite directions, regression 7 reveals a
positive and statistically significant relationship between the number of firms in a market
(n) and fundraising expenditures.
<Insert Table 7 about here>
19
As a final check, it is helpful to look at the ratio of private donations to
solicitation dollars used to raise that amount. The ratio approximates the return on one
dollar allocated to fundraising. Here, approximate, implies that these are average values
rather than marginal. However, Tinkelman (2002) offers a set of reasonable assumptions,
whereby average expense ratios offer good approximations for marginal values. Contrary
to regressions 6 and 7, it appears that HHI has little impact on the average returns to
fundraising. The coefficient on HHI retains the expected positive sign, yet there is little
statistical significance. In contrast, using n as the measure for market structure does give
an inverse and statistically significant coefficient. This result implies, for low levels of n,
that adding an additional firm into a market reduces the return to fundraising by roughly
two dollars. Given that a typical market in the sample has a return of seventy five dollars
for every dollar allocated to fundraising, this is not an overwhelming impact.
VI. Conclusion
Altruistic nonprofit managers face both private and social incentives when
making fundraising decisions. The analysis presented here confirms the prediction
nonprofit organizations will follow private market incentives by reducing per-firm
fundraising with increased competition. This trend is important for two reasons. First,
nonprofit firms appear to behave similarly to their for-profit counterparts. Nonprofits,
despite the non-distribution constraint, appear responsive to market forces when making
their fundraising decisions. As their private benefit to solicitation declines in a saturated
market, nonprofit firms will voluntarily reduce their fundraising outlays. This lends
credibility to the notion that market competition plays an underappreciated role in
guiding nonprofit firm behavior.
20
Second, even though per-firm fundraising declines with entry, aggregate
fundraising rises. This is true even when the size of the donor market is held constant.
The implication of this finding is that additional fundraising resources are likely stealing
donors away from other nonprofits rather then generating new resources for a particular
cause. This is a genuine concern for altruistic nonprofit managers who care not only for
the performance of their own organization, but with the overall provision of a charitable
service. Ironically, average fundraising expense ratios do not indicate that this sample of
nonprofits are fundraising beyond what is privately optimal. In fact, it appears that most
firms within the sample have the private incentive to fundraise even more. These
managers face the difficult choice of serving their own organization by fundraising
aggressively, or devising some method to reduce aggregate solicitation expenditures.
Previous research has focused on the use of collective fundraising (like the United
Way) or legal caps on fundraising ratios as a solution for discouraging excessive
fundraising expenditures. This mechanism, however, has substantial disadvantages. The
findings of this paper would indicate two alternatives to this mechanism. First, market
competition appears to be an effective private restraint on fundraising. To the extent that
donors can observe relative expense ratios, competitive donor markets will continue to
restrain fundraising activities. As better information is made available to a wider donor
audience, price competition among nonprofit firms will likely become even more
effective.
Second, if nonprofit managers are indeed worried about high aggregate levels of
fundraising, the crucial issue is the overall number of nonprofit firms in the market, not
their individual fundraising decisions. The solution presented by the model is that
21
managers should advocate for increased barriers to entry, thereby reducing the number of
firms. Currently, legal and financial barriers for the formation of a nonprofit are low.
Typically a few thousand dollars in legal fees and four months time is all that is necessary
to receive charitable tax status. The most direct method to reduce excess fundraising is
simply raising the cost of entering the nonprofit market. This may include a range of
policies from raising the fees associated with the granting of charitable tax status to a
more draconian measure such as Certificates of Need. While this would correct the
problem of excessive aggregate fundraising, ironically, it would stimulate greater
fundraising activity per-firm.
Despite its results, the analysis does have limitations. Any paper of this type is
subject to criticism of the choice of market construction. In this case, nonprofit markets
were organized based on Metropolitan Statistical Areas (MSA). Great care was taken to
ensure that the nonprofit organizations included in the sample could reasonably be
expected to compete locally for donations within those geographic boundaries. However,
there will be cases where the actual market boundaries of firms do not match up with
available proxies. Furthermore, the paper is not able to address the bulk of nonprofit
organizations who compete in larger regional or national markets for donations. This
eliminates many interesting classes of organizations including universities and large
hospitals. Another useful extension would be closely examining the donor market
boundaries of different types of nonprofits. As more research is dedicated to examining
the impacts of market competition among nonprofits, it will take on increasing
importance to define the precise boundaries of those markets.
22
End Notes
1 Based on authors calculations and data taken from the New Nonprofit Almanac,
Weitzman& Jalandoni (2002)
2 This presumes that nonprofit are maximizing revenues net of costs (net-revenue). If the
organization is, instead, a budget maximzer then the firm should continue fundraising
until the last dollar spend returns no more donation revenue. See Steinberg (1986) for
more detailed discussion of this issue.
3 The formation of a nonprofit organization usually falls under various state laws. The
defining characteristic of the nonprofit firm is the legal restriction against private
inurement, whereby excess earnings of the organization cannot contribute to the private
benefit of a steak-holder (i.e. the non-distribution constraint). This is a distinct issue from
charitable tax status and means that the organization cannot distribute net earnings to its
stakeholders. Stated succinctly, no entity may be a residual claimant. This removes the
profit maximization objective assumed for most for-profit firms. Among organizations
characterized as nonprofit entities, only a fraction of them are classified under section
501(c)(3) of the Federal tax code as tax-exempt charitable organizations. IRC § 501(c)(3)
defines charitable organizations as those who are operated exclusively for religious,
charitable, scientific, testing for public safety, literary, or educational purposes, or to
foster national or international amateur sports competition, or the prevention of cruelty to
children or animals. These charities are unique in that they are relieved of specific tax
burdens and can offer donors the ability to deduct contributions from their federal income
tax obligations. This study restricts its attention to this population of nonprofits.
23
4 The concept of price is not immediately clear in the case of donor markets. The critical
issue is what donors perceive as the cost facing them when making a donation to a
nonprofit firm. One common measure is the fraction of the donation remaining net of
fundraising costs and administration overhead (overhead expense ratio). This would
imply a price for one dollar’s worth of output equal to 1/(1-F-A), where F and A are the
fraction of donations allocated to fundraising and administration. Watchdog
organizations such as the Wise Giving Alliance by the Better Business Bureau aggregate
expense ratios so that donors can compare across similar nonprofits more easily. The
BBB found that 79% of Americans said that fundraising expense ratios was important to
them in deciding how to allocate their donations (Cordes & Rooney, 2004a)
5 There has been a sizable literature concerning competition between for-profits and
nonprofits. See Duggan (2002) for a recent example. This literature almost exclusively
focuses on the hospitals and the healthcare industry. These firms, however, do not rely
heavily on donations are often much larger than a typical charity. As such, that literature
is not very helpful in analyzing competition solely among charitable nonprofits. There
has also been considerable work on the impact of government grants on fundraising
efforts, the so called “crowding out hypothesis”. Andreoni & Payne (2003) and show
that nonprofits reduce fundraising effort when a grant is received. This strain of research,
however, does not address competition among nonprofits and so is not directly relevant to
the current study.
6 The theoretical model conceives of the social optimum as the sum of consumer surplus
plus revenue (net of production costs) to the nonprofit.
24
7 Readers may object to this dramatic simplification. Indeed, it should be acknowledged
that nonprofit organizations have a wide variety of tools for solicitation. Solicitation
technologies could include: direct mail, broadcast media, telemarketing, corporate
fundraising, grant writing, along with personal contact. Sargeant & Kahler (1999) offer
estimations on the differing rates of return to each type of fundraising for UK charities.
For this analysis, however, only a generic solicitation technology is considered. First, it is
not possible to differentiate fundraising methodologies from Form 990 data. The implicit
assumption of the model is that nonprofit managers are already choosing the optimal
bundle of fundraising technologies for their circumstances. A more detailed dataset on a
specific sub-sector could yield additional insight on the substitution effect between
different types of fundraising methodologies. This, however, is beyond the scope of the
study. At present, we are only concerned with the aggregate dollar value of fundraising,
not how it was spent.
8 Nonprofit organizations with less than $25,000 in revenue as well as religious
congregations are not required to file the Form 990. The IRS keeps no data on these
firms.
9 Nominal values for expenditures were converted to real using GDP deflator
information.
10 Because the model focuses on the fundraising activities of firms, those organizations
which reported no fundraising over the entire 11 year panel were removed from the
dataset. It is likely that these organizations received charitable funds from specialized
25
support organizations instead of operating their fundraising activities “in-house”. These
are special cases that should be considered separately.
11 The Herfindahl-Hirschman Index is calculated by the summing the squared market
shares of each firm in a market.
12 As a point of reference, the Department of Justice, considers HHI indexes above 1800
as “highly concentrated” markets in the case of financial institutions (Laderman, 2003).
13 Other measures are possible such as the fraction of total revenues spent on fundraising
or the fraction of donation revenue spent on fundraising. The Herfindahl-Hirschman
Index is calculated by the summing the squared market shares of each firm in a market.
14 The Nonprofit Overhead Cost Project (sponsored by the Urban Institute and Indiana
University) has released a series of research briefs focused on analyzing the issue of
overhead costs in nonprofits. While No. 4 offers anecdotal evidence that nonprofit staff
tend to underreport their fundraising expenditures, accuracy does appear positively
correlated with firm expenditures and size (Yetman and Yetman, 2004). There are viable
reasons for organizations to report little or no solicitation expenditures while, at the same
time, receiving substantial donations. Organizations may rely on volunteer staff or board
members to solicit funds. Or the nonprofit may belong to a partnership organization such
as the United Way which collects funds on behalf of the nonprofit. Yet recent research
indicated that it is more likely the case that managers and staff do not fully account for
their time spent on fundraising. For the particular sample drawn for this empirical
analysis, approximately 72% of organizations receiving donations report spending
nothing on fundraising.
26
15 IRS reviewed only .029 percent of all non-profit returns in 2001 (Government
Accounting Office, 2002)
16 Ackerberg (2003) and Lee (2002) find at least weak support for the inverted U
relationship between market concentration and advertising in a wide variety of for-profit
firms. Advertising expenditures increase as markets become more concentrated. When
markets become highly concentrated (such as in a duopoly) advertising then appears to
decline.
27
References
Ackerberg, D. (2003). ” Advertising, Learning, and Consumer Choice in Experience
Good Markets: An Empirical Examination,” International Economic Review,
44(3), pp.1007-40.
Andreoni, J. & Payne, A. (2003). “Do Government Grants to Private Charities Crowd
Out Giving or Fundraising?” American Economic Review, 93(3), pp. 792-812.
(ARRFC) The Center on Philanthropy (2002). Giving USA 2002. Indiana University &
the American Association of Fundraising Counsel.
Bergstrom T., Blume L., Varian H. (1986). “On the Private Provision of Public Goods,”
Journal of Public Economics, 29, pp. 25-49.
Cordes, J. & Rooney, P. (2004a). “Task Force Report: Fundraising Costs,” National
Council on Nonprofit Enterprise. www.nationalcne.org.
Cordes, J. & Rooney, P. (2004b). “Fundraising Costs,” in Effective Economic Decision-
Making by Nonprofit Firms. Young, Dennis ed., The Foundation Center.
Diamond, W. & Goodings-Williams, S. (2002). “Using Advertising Constructs and
Methods to Understand Direct Mail Fundraising Appeals,” Nonprofit
Management and Leadership, 12(3), pp. 225-242.
Duggan, M. (2002). ” Hospital Market Structure and the Behavior of Not-for-Profit
Hospitals,” RAND Journal of Economics, 33(3), pp. 433-46.
Glaeser, E. (2003). “The Governance of Not-for-Profit Firms,” in Edward Glaeser, ed.,
The Governance of Nonprofit Firms. Chicago, University of Chicago Press.
Greenlee, J. S., & Brown, K. L. (1999). “The Impact of Accounting Information on
28
Contributions to Charitable Organizations,” Research in Accounting Regulation,
13: pp.111–125.
Grossman, G. & Shapiro, C (1984). “Informative Advertising with Differentiated
Products,” Review of Economic Studies, 51: pp. 63-81.
Government Accounting Office (2002). “Tax-Exempt Organizations: Improvements
Possible in Public, IRS, and State Oversight of Charities,” GAO-02-526.
Hager, M., Rooney, P., Pollak, T. (2002). “How Fundraising Gets Carried Out in US
Nonprofit Organizations,” International Journal of Nonprofit and Voluntary
Sector Marketing, 7(4): pp. 311-325.
Ladermann, Liz. (2003). “Good News on Twelfth District Banking Market
Concentration,” FRBSF Economic Letter, 2003-31.
Lee, C. (2002). “Advertising, Its Determinants, and Market Structure,” Review of
Industrial Organization, 21(1) : pp. 89-101
Malani, A. & Choi, A. (2004) “Are Non-Profit Firms Simply For-Profits in Disguise?”
University of Virginia Law School, Working Paper.
Okten, C. & Weisbrod, B. (2000). “Determinant of donations in private nonprofit
markets,” Journal of Public Economics, 75, pp. 255-272.
Rose-Ackerman, Susan. (1982). “Charitable Giving and Excessive Fundraising,”
Quarterly Journal of Economics, 97(2). pp. 193-212.
Sargeant, A. & Kahler, J. (1999). “Returns on Fundraising Expenditures in the Voluntary
Sector,” Nonprofit Management and Leadership, 10(1). pp. 5-19.
Steinberg, R. (1986). “The Revealed Objective Functions of Nonprofit Firms,” The
RAND Journal of Economics, 17(4): pp. 508-526.
29
Shapiro, C. (1980). “Advertising and Welfare: Comment,” The Bell Journal of
Economics, 12(2): 749-751.
Tinkelman, D. & Mankaney, K. (2005) “When is administrative efficiency associated
with charitable donations,” Pace University, working paper.
Tuckman, H. (1998) “Competition, Commercialization, and the Evolution of Nonprofit
Organizational Structures,” Journal of Policy Analysis and Management, 17(2):
pp. 175-194.
Weitzman, M. & Jalandoni, N. (2002) The New Nonprofit Almanac and Desk Reference
Jossey-Bass, San Francisco.
Yetman , M. and Yetman, R. (2004) “The Effects of Governance on the Financial
Reporting Quality of Nonprofit Organizations,” Working Paper.
30
Table 1: Selected Markets for year 2000
Sector Description & NTEE Code Number of
firms Number of
Markets
1 Museums A50-A57 2,090 291
2 Performing Arts A62-A6C 6,487 317
3 Community Health Treatment E30-E42 2,350 308
4 Abuse Prevention I70-I73 650 220
5 Employment & Vocational Training J20-J33 2,373 298
6 Nursing, home health care E90-E92 1,973 252
7 Substance Abuse Prev. & Treat. F20-F22 2,239 283
8 Hot Lines & Crisis Prevention F40-F42 283 165
9 Crime Prevention and Rehabilitation I20-I44 1,223 237
10 Food Pantries & Programs K30-K36 1,117 270
11 Public Housing & Rehabilitation L21-L25 3,290 285
12 Homeless Shelters L40-L41 & P85 976 215
13 Community Centers P28 1,048 240
14 Family Counseling P46 621 197
15 Senior Centers P81 1741 290
16 Residential Care & Group Homes P73 2,337 293
Totals 30,978 2,440
Source: NCCS year 2000 Core Files and Author’s Calculations
31
Table 2: Mean HHI by sector by year
Sector 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000
1 2016 2118 2269 2290 2313 2133 2126 2122 2096 2052 1947
std. dev. 1806 1802 1858 1824 1829 1868 1750 1780 1776 1682 1685
2 3007 3235 3188 3212 3122 3247 3118 3085 3001 2942 2940
std. de. 2509 2556 2515 2550 2515 2509 2487 2501 2495 2481 2416
3 3137 3128 3092 2960 2898 2801 2824 2721 2738 2742 2720
std. dev. 2570 2523 2466 2287 2228 2208 2163 2102 2072 2034 2059
4 2768 2797 2768 2793 2748 2737 2664 2603 2600 2577 2548
std. dev. 2068 2118 2121 2188 2116 2136 2081 2054 2036 1982 2013
5 1744 1811 1781 1747 1754 1759 1743 1746 1702 1700 1716
std. dev. 1878 1894 1871 1849 1838 1867 1841 1866 1833 1838 1845
6 2406 2419 2402 2424 2398 2370 2565 2614 2597 2598 2377
std. dev. 2292 2294 2226 2217 2127 2173 2194 2217 2199 2123 2123
7 3314 3461 3447 3334 3244 3294 3417 3327 3391 3483 3543
std. dev. 2683 2772 2789 2694 2686 2697 2686 2599 2555 2590 2634
8 3243 3203 3300 3431 3326 3241 3222 3221 3160 3135 3135
std. dev. 2561 2606 2586 2631 2556 2565 2558 2605 2521 2526 2517
9 3321 3357 3241 3146 3139 3114 3185 3372 3514 3483 3505
std. dev. 2513 2514 2431 2393 2358 2317 2257 2304 2305 2364 2416
10 2064 2032 1867 1841 1806 1845 1798 1896 1852 1812 1870
std. dev. 2203 2194 2061 2004 1966 1953 1952 2003 1922 1843 1869
11 2670 2670 2600 2527 2493 2547 2468 2369 2438 2549 2457
std. dev. 2378 2358 2303 2184 2188 2099 2122 2031 2048 2034 2025
12 3711 3651 3645 3708 3805 3729 3611 3551 3536 3526 3469
std. dev. 2686 2666 2626 2599 2641 2573 2592 2547 2529 2490 2500
13 3387 3328 3202 3130 3231 3212 3238 3279 3275 3259 3308
std. dev. 2709 2701 2622 2531 2604 2568 2558 2598 2552 2558 2623
14 3235 3276 3235 3277 3264 3213 3255 3198 3245 3236 3292
std. dev. 2521 2552 2546 2573 2527 2466 2449 2459 2490 2501 2478
15 2531 2445 2337 2323 2307 2293 2374 2331 2296 2330 2372
std. dev. 2096 2043 1983 1946 1945 1960 1993 1969 1933 1985 2016
16 2016 2118 2269 2290 2313 2133 2126 2122 2096 2052 1947
std. dev. 1806 1802 1858 1824 1829 1868 1750 1780 1776 1682 1685
Source: NCCS Core Files and author’s calculations.
Standard Deviations are in italics.
See Table 1 for a list of the 16 nonprofit sectors.
32
Table 3: Average Solicitation Expenditure Per-Firm in 2000 (dollars)
Fundraising Expenditures (Solicit)
Sector Mean Std. Dev. Min. 75 percent 90 percent Max.
1 $107,288 543096 0 34,778 203,129 13,300,000
2 $26,840 202089 0 1,585 27,852 7,497,000
3 $22,269 151755 0 0 40,761 5,957,075
4 $19,933 65029 0 11,315 49,973 867,716
5 $19,031 384060 0 0 17,324 18,400,000
6 $22,192 294719 0 0 7,002 8,920,499
7 $19,961 331504 0 0 11,620 12,100,000
8 $18,039 65954 0 10,952 38,094 813,812
9 $28,864 291086 0 0 18,190 6,112,900
10 $39,565 258738 0 6,350 67,704 6,804,877
11 $3,786 91303 0 0 0 5,056,438
12 $35,691 181646 0 16,019 82,130 4,842,037
13 $42,272 188686 0 12,680 93,234 4,565,529
14 $12,142 49146 0 2,044 28,811 808,508
15 $12,621 71120 0 909 25,676 2,219,581
16 $37,816 374603 0 0 41,846 13,100,000
Source: NCCS Core Files year 2000 and author’s calculations
Table 4: Average Aggregate fundraising per market (S) in Year 2000
Aggregate Market Fundraising Expenditures (S)
Sector Mean Std. Dev. Min. 75 percent 90 percent Max.
1 33,600,000 59,500,000 0 28,100,000 113,000,000 234,000,000
2 53,300,000 99,800,000 0 47,700,000 335,000,000 335,000,000
3 12,000,000 17,700,000 0 13,900,000 47,600,000 67,600,000
4 11,100,000 33,500,000 0 9,775,111 25,700,000 335,000,000
5 15,200,000 43,800,000 0 11,600,000 25,900,000 335,000,000
6 10,500,000 14,000,000 0 15,100,000 34,600,000 67,600,000
7 17,700,000 58,900,000 0 11,800,000 25,900,000 335,000,000
8 7,794,626 15,500,000 0 6,545,554 20,100,000 99,100,000
9 6,649,950 15,300,000 0 5,286,378 15,200,000 99,100,000
10 14,100,000 50,300,000 0 7,005,190 26,400,000 335,000,000
11 3,115,666 5,989,442 0 3,031,660 9,406,051 57,600,000
12 8,904,060 13,500,000 0 9,775,111 25,600,000 99,100,000
13 4,851,861 8,892,884 0 5,855,101 10,800,000 184,000,000
14 6,801,946 15,100,000 0 4,526,414 26,400,000 147,000,000
15 5,644,499 9,991,641 0 5,286,378 18,000,000 47,600,000
16 12,400,000 25,400,000 0 10,500,000 26,800,000 147,000,000
Source: NCCS Core Files 2000 and author’s calculations
33
Table 5: Donations received per dollar of solicitation, by market in Year 2000
Public Donations/Solicitation Expenditures
Sector mean Std. Dev. Min. 75 percent 90 percent Max.
1 19.12 35.50 0.02 15.75 27.09 648.84
2 15.49 51.92 0.06 14.75 22.63 2,455.45
3 45.32 245.61 1.26 30.26 49.33 5,837.89
4 41.29 112.85 1.69 28.73 60.60 879.86
5 185.03 2,679.79 0.00 57.79 80.93 66,514.52
6 24.71 43.52 0.06 22.02 51.45 619.29
7 102.22 522.76 0.01 57.76 96.76 9,195.35
8 56.65 227.84 1.04 41.73 76.60 2,919.27
9 62.12 387.53 0.00 35.04 69.70 6,793.35
10 162.71 1,297.91 4.34 52.84 98.64 20,202.55
11 89.66 501.18 0.00 28.38 83.98 6,193.68
12 41.79 139.42 1.04 30.52 57.84 2,574.78
13 45.59 153.69 0.06 40.62 71.90 2,388.05
14 125.70 919.64 0.02 43.98 109.68 15,025.23
15 99.12 446.05 0.00 52.84 127.00 8,636.91
16 44.97 258.64 0.11 30.65 54.24 4,833.84
Source: NCCS Core Files 2000 and author’s calculations
34
Table 6: Per-Firm Solicitation Expenditure Regression Results
Solicit
OLS OLS solicit>0 Tobit
(1) (2) (3) (4) (5)
HHI 6.84** 6.36** 6.89** 9.75** 7.90**
(2.75) (2.69) (2.90) (2.42) (2.00)
HHI2 -0.0005** -0.0006** -0.0008** -0.0007** -0.0007*
(2.67) (2.48) (3.24) (2.21) (1.94)
AGE 883** 635** 1690** 1369** 985**
(2.45) (3.97) (14.70) (2.53) (3.33)
ASSETS 0.005** 0.007** 0.007** 0.005** 0.006**
(3.79) (5.37) (55.50) (3.15) (4.34)
Lag CONTRIBUTIONS 0.02** 0.02* 0.02** 0.03** 0.03**
(3.08) (1.84) (139.60) (2.96) (1.97)
EDUCATION 209 36676** -10829
(0.01) (1.35) (0.39)
Per-Capita Income 0.69* 3.56** 1.04
(1.66) (7.84) (1.52)
MSA Population 0.004 0.009** 0.005
(1.45) (4.05) (0.85)
% BLACK 4386 -20730 18559
(0.24) (1.21) (0.60)
% HISPANIC 18397 -6460 33186
(1.07) (0.39) (1.04)
R2 0.25 0.17 0. 20 0.27
Regressions are clustered by market
t-statistics are in parentheses
Sector and time dummies have been suppressed for each regression
Market dummies have been suppressed for regressions (1) and (3)
** and * denote statistical significance at the 5 and 10 percent level; respectively
35
Table 7: Aggregate Market Solicitation Expenditure Regression Results
Market Solicitation (S) Donations/Solicitation
(6) (7) (8) (9)
HHI -2,123.65** 0.03
(3.91) (0.32)
HHI2 0.14** 0.000004
(3.52) (0.58)
N 62,876** -2.02**
(7.60) (2.54)
N2 -4.43 0.0019**
(0.50) (2.52)
Donations 3.60** 3.13**
(3.70) (3.38)
Education -11,489,680 -9,884,986 1,158 1,044
(1.36) (1.35) (1.51) (1.40)
Per-Capita Income 282.56* 141.48 0.04 0.04
(1.88) (1.11) (1.07) (1.09)
MSA Population 1.09** -0.45 0.00 0.00*
(2.05) (1.17) (1.53) (1.79)
%Black -2,166,864 -1,384,771 -123 -232
(1.12) (0.97) (0.47) (0.83)
% Hispanic -1,601,074 -977,839 -144 -183
(0.98) (0.87) (0.55) (0.67)
R2 0.27 0.54 0.04 0.03
Robust standard errors used
t-statistics are in parentheses
Sector and time dummies have been suppressed for each regression
** and * denote statistical significance at the 5 and 10 percent level; respectively
36
