Вы находитесь на странице: 1из 27

FACULTY OF ECONOMICS

AND BUSINESS
CAMPUS BRUSSELS
CAMPUS BRUSSEL
Not Confidential
FACULTEIT ECONOMIE EN
Internally Organised Masters Thesis
BEDRIJFSWETENSCHAPPEN

Financial vulnerability
of nonprofit
organizations
The health care sector in Belgium

Simon GEVCEN
Masters Thesis Submitted for the Degree of
Master in International Business Economics and Management
Supervisor: Sandra VERBRUGGEN
Academic Year: 2014-2015
Defended in June 2015

FACULTY OF ECONOMICS AND BUSINESS - CAMPUS BRUSSELS


WARMOESBERG 26 B 1000 BRUSSELS BELGIUM

FACULTY ECONOMICS
AND BUSINESS
CAMPUS BRUSSELS

MASTERS THESIS
Financial vulnerability of nonprofit organizations
The health care sector in Belgium
Simon GEVCEN

Abstract
Whether an organization is for-profit or nonprofit, its financial vulnerability is a concern for its
management. The common denominator between both types of organizations is the harm that a
financial distress can have on their ability to provide constant services. However, nonprofits are not
for-profit, which implies that the models describing the financial vulnerability of the for-profit sector
cannot be literally transposed to nonprofit organizations. Indeed, new models have been designed
for the nonprofit sector. Therefore, this article intends to summarize all the models assessing the
financial vulnerability of the nonprofit sector.
Throughout the past ten years, scholars pointed out the weaknesses of the Belgian health care
nonprofit organizations, generally described as financially vulnerable. However, the literature backing
this statement is weak. Inexistent, perhaps. In fact, no statistical research nor any empirical study
supporting this statement has been conducted yet. Therefore, this article describes the first empirical
investigation on the nonprofit sector in Belgium in order to assess the variables affecting its financial
vulnerability in the period 2010-2013. Relying on longitudinal data about human health
organizations, residential care organizations, and social work organizations, this study determines
the significance of continuous and categorical predictors as determinants of the financial vulnerability
of nonprofit organizations in this health care segment of the Belgian nonprofit sector.
Acknowledgments: I express my faithful thanks to Pr. Sandra Verbruggen for her support and
guidance throughout the achievement of this Masters thesis.
Keywords: financial, vulnerability, performance, exposure, distress, ratios, nonprofits, nonprofit
organizations, health care, Belgium.

Introduction
Financial vulnerability is the potential exposure of an organization facing financial distress, which is
a common situation for many nonprofit organizations (Wang & Liu, 2010, p. 463). According to
Tuckman & Chang (1991), the nonprofit sector has been in the shadows for a very long time (p.
459). Although the academic community started to address the topic only recently, researchers have
given increased attention to the topic of vulnerability and organizational demise (Hager, 2001, p.
376) in the past fifteen years.

We can identify four reasons explaining the lack of research on this topic, which can then be seen as
a priori limitations that bounded the previous investigations. First, a generally accepted definition of
a nonprofit organization is lacking. Second, the economic weight of the nonprofit sector remains
marginal although it is getting more importance over time at an increasing rate (Banque Nationale
de Belgique, 2004, p. 9). Third, researches do not always have all necessary data because little
information is available to inform the general public (Tuckman & Chang, 1991, p. 449). Fourth, the
literature did not end up on a one-fit-all methodology regarding the study of nonprofit organizations.
However, the lack of research does not mean a lack of interest. In his article, Trussel (2002) noted
that whether or not a nonprofit organization is susceptible to financial problems is a concern of all
stakeholders of the organization, because financial problems might not allow an organization to
continue to meet its objectives and provide services (p. 17). Since some organizations close as a
result of financial distress (Tevel, Katz, & Brock, 2014, p. 4), this concern is of interest. In fact, our
study shall attract the attention of many types of agents including managers during the process of
strategic planning, government bodies setting policies, auditors conducting review studies, donors in
the decision-process for their contribution, and potential creditors in determining credit-worthiness
of charitable organizations (Greenlee & Trussel, 2000, p. 209).
Thus, the research question to which we intend to answer at the end of this research is the following:
What are the variables affecting the financial vulnerability of the Belgian health care nonprofit
organizations between 2010 and 2013? Throughout the process of this study, the core research
objective is to determine to what extend certain parameters of an NPO1 can determine its financial
vulnerability2, since the latter can harm any NPO in its ability to deliver its social objective (Arshad,
Bakar, Razali, & Omar, 2013, p. 408). In their pioneer study, Tuckman & Chang (1991) stressed
three reasons to investigate on this field.
First, the nonprofit sector is growing rapidly over time (Tuckman & Chang, p. 446). As it continues
to grow, the demand for accounting services by these organizations will increase (Trussel, Greenlee,
& Brady, 2002, p. 66). In fact, NPOs remain key players in the Belgian economic landscape (Vander
Donckt & Rigo, 2013, p. 11) with respect to their contribution of more than 5% to the wealth of the
country. Given the importance of this sector, it is of interest to study its financial vulnerability since
the Belgian nonprofit sector is not only one of the largest in Europe but is also a product of the []
welfare state model (Mertens, et al., 1999, p. 43).
Second, the impacts of the nonprofit sector on employment is substantial as it affects a large number
of people (Tuckman & Chang, 1991, p. 446). Its contribution to the countrys employment is about
11.9% in 2010 and reflects the labor-intensive nature of production (Vander Donckt & Rigo, 2013,
p. 11). This support improved during the period 2010-2013 with an increasing number of people
employed in that sector, as shown in Appendix 1. Given the importance of this sector, it is of interest
to study its financial vulnerability since it has considerable effects both on employment [] and on
the institutions that train people for employment (Tuckman & Chang, 1991, p. 446).
Third, the revenue sources of the nonprofit sector is a key factor to analyze the ability of a nonprofit
organization to respond to financial pressures (Tuckman & Chang, 1991, p. 446). NPOs generate
most of their revenue from sales of goods and services as well as from current transfers (Vander
Donckt & Rigo, 2013, p. 12). However, public funding is the major source of revenue for NPOs. In
fact, more than half of the public funding going to the nonprofit sector goes to the healthcare sector
(Vander Donckt & Rigo, 2013, p. 12). Table 1 shows the percentage distribution of the public funding
to the nonprofit sectors in Belgium by sector of activity in 2010.

1
2

Abbreviation of Nonprofit Organization.


Abbreviated as FV.

Table 1. IWPP3 distribution of Belgian public funding, by sector of activity (2010)


2,6

11,5

1,5

28,5

Agriculture and Industry


Services
Education
Human health
Social action
Arts, spectacles, and recreational activities

1,7

54,2

Source : (Vander Donckt & Rigo, 2013, p. 28)


A specific focus on this third element can be a starting point to our study. Actually, the diversity of
revenue sources can show some instability, and this instability in the donor base can emanate from
several sources (Tuckman & Chang, 1991, p. 447). Three changes can occur and cause distress in
the sources of revenues for NPOs. The first change concerns the donors preferences: the taste of
donators can vary when deciding which nonprofit to fund over certain periods. The second change
concerns the economy: if the economic situation of donators income is getting worse, donations
may be among the first discretionary items that donors cut (Tuckman & Chang, 1991, p. 447). The
third change concerns the tax law: tax regulation affects the net contribution of a donor.
Consequently, NPOs experiencing revenue instability are more likely to remain financially
vulnerable compared to those not experiencing it.
With respect to this research background, the contribution of our study is twofold. The first
contribution has regard with the application to the Belgian nonprofit sector of some models presented
in this article. In fact, a thorough review of the literature on the financial vulnerability of Belgian
NPOs leads to the conclusion of a complete lack of research in this area. Many roads are then open
for our investigation. Despite the publications of some authors with regard to the financial
vulnerability of NPOs in general, further research needs to expand the knowledge of this field in
Belgium. In fact, the Belgian health care nonprofit sector is generally said to be financially vulnerable,
without any statistical basis. Therefore, with the development of this Masters thesis, our contribution
intend to be the basement of the empirical literature on the Belgian health care nonprofit sector.
The second contribution has regard with the methods. Although the models developed in the
business sector are often inappropriate for organizations whose purpose is to maximize service rather
than profit (Trussel, Greenlee, & Brady, 2002, p. 66), Wang & Liu (2010) noted that the consensus
of how to measure the financial performance for non-profit organization has not been reached among
scholars (p. 463). This view is shared by McIndoe & Sullivan (2014) who noted that meanwhile
there is no consensus on a single indicator of nonprofit financial health, examining several financial
ratios simultaneously has emerged as a common practice to assess the relationship between
measures of nonprofit financial vulnerability and important outcomes such as survival (p. 3).
Therefore, with the development of this Masters thesis, our contribution intend to adapt models
presented in the literature review to the health care sector of the Belgian nonprofit sector.
In spite of the attention given by experts to the nonprofit sector, this segment of the economy is
relatively unknown. The shortage of statistical data did not allow researchers to optimize its study.
However, the Johns Hopkins University located in Baltimore, United States, coordinated the biggest
project ever launched regarding the nonprofit sector at the world stage, known as the Comparative
Nonprofit Sector Project (Mertens, 2004, p. 44). This project lead to the publication of a fundamental
handbook by the United Nations (2003) giving to all institutions in charge of national accounts the
incentive to create a Satellite Account of Institutions without Profit Purpose 4 (Mare, 2010a).

3
4

Abbreviation of Institutions Without Profit Purpose.


In French, Compte satellite des institutions sans but lucratif. In Dutch, Bijkomende rekening van de instellingen zonder winstoogmerk.

A satellite account is a set of structured and coherent data, regarding a particular field, and
established in coherence with the central framework of the national accounts of the country (Banque
Nationale de Belgique, 2010, p. 3) to offer statistical data and macroeconomic information (Mare,
2010b). Indeed, this handbook recommends statistical standards and guidelines for the
development of data on non-profit institutions (United Nations, 2003, p. iii).
Following the recommendation of Eurostat (Mare, 2010c), Belgium has been one of the first
countries (Acx, Rigo, & Vander Donckt, 2011, p. 5) to publish in early 2004 his satellite account. For
the case of Belgium, this satellite account is managed by the Institute of National Accounts 5, which
is the body in charge of the annual release of data regarding the Belgian nonprofit sector (Banque
Nationale de Belgique, 2012, p. 3). Actually, in order to be considered in Belgium as an institution
without profit purpose, entities must comply with all the following criteria:

Be nonprofit : prohibit the distribution of profit to its members or administrators,


Be recognized : as organization with an institutional existence,
Be private : institutionally separate from the State,
Be independent : have its own rules and decision-making bodies,
Be open : membership is open and the entity is able to collect voluntary resources in the form
of donations and volunteering (Mare, 2010a).

The process of gathering all institutions without profit purpose does not require considering all of
them as a homogeneous group. (Banque Nationale de Belgique, 2004, p. 12). The set of these
institutions are divided in two modes: institutional segmentation, and sectorial segmentation.
First, regarding the institutional segmentation shown in Table 2, the Belgian legislator recognizes
many types of organizations. However, our research focuses on certain specific types of
organizations. In fact, the most important category is the association without profit purpose, which
consists of at least three persons, either nationals or foreigners, with the purpose of realizing a nonlucrative objective (Portail Belgium, 2012). It is entitled to obtain the legal personality if it releases
its articles in the Belgian Official Journal 6 and registers its activities in the Crossroads Bank for
Enterprises7.
Table 2. Institutional segmentation of NPOs
A. Without legal personality
- De facto association8 refers to any gathering of people, which does not require any legal
recognition by the legislator.
B. With legal personality (27 June 1921 Law, modified by 2 May 2002 Law)
- Public utility institution9 refers to any association known both as foundations and as
professional unions (Banque Nationale de Belgique, 2004, p. 14).
- Association without profit purpose10 refers to any association that is not devoted to
industrial or commercial operations and which does not seek to obtain material
gains for its members (Mertens & Lefbvre, 2001, p. 8).
- International association without profit purpose11 refers to any association that fits with the
requirements of the association without profit purpose, but that is furthermore
pursuing a nonprofit purpose of international utility.
Source : (Mertens & Lefbvre, 2001, p. 8; Mare, et al., 2005, p. 11)

In French, Institut des comptes nationaux. In Dutch, Instituut voor de Nationale Rekeningen.
In French, Moniteur belge. In Dutch, Belgisch staatsblad.
7
In French, Banque-carrefour des entreprises. In Dutch, Kruispuntbank van Ondernemingen.
8
In French, Association de fait. In Dutch, Feitelijke vereniging.
9
In French, Etablissement dutilit publique. In Dutch, Instelling van openbaar.
10
In French, Association Sans But Lucratif (ASBL). In Dutch, Vereniging Zonder Winstoogmerk (VZW).
11
In French, Association internationale. In Dutch, Internationale associatie.
6

Indeed, the legislation regulates the boundaries of this non-lucrative objective. It prevents it from
pursuing an objective of enrichment. Hence, the organization is not mandated to retain any minimum
capital, nor allowed to redistribute benefits made from the organization activities to its members,
neither allowed to aim to make profit although fees membership can be charged (Service Public
Fdral Justice, 2013). Since it operates as a nonprofit entity, it is subject only to the income tax
on legal entities (Business Belgium, 2010). However, the application of the corporation tax is
immediate if subsequent investigations reveal a profit seek of the organization.
Second, regarding the activity sector segmentation shown in Table 3, its classification rely on the
European Nomenclature for Economic Activities12 13 that is a reference framework for the production
and release of statistical data regarding economic activities in Europe. Furthermore, we can also look
at the value added of each sector to the overall sector of IWPP.
Table 3. Distribution of the IWPP added value, by sector of activity (2010)
11

13

Agriculture and Industry


2

Services
Education
Human health
Social action

33

Arts, spectacles, and recreational activities

36

Other service activities


Source : (Vander Donckt & Rigo, 2013, p. 20)
Furthermore, Appendix 2 shows the importance of these activities regarding their level of
employment in 2010. From the total number of employed units, about half is allocated to only two
activities social action, and human health. In fact, these two categories are those that account for
most of the job creation with respectively 9% and 5.8%. Although both of them are of interest for
this study, a specific research focus has to be selected. Therefore, the research focus our study will
be on the human health segment of the nonprofit organization sector.

Literature review
This literature review intends to give an overview of the most important models developed both in
the for-profit perspective and in the nonprofit perspective. The former has seen the emergence of
two models: Altman (1968) model, and Ohlson (1980) model. The latter has seen the rise of three
models: Program Expenditure Model with two extensions, Net Assets Model, and Net Earnings Model.
The for-profit sector is of interest because all models rely on accounting measures. In fact, most of
them focus on one factor: corporate bankruptcy. Actually, they mainly use profitability, cash flow,
and leverage ratios as predictor variables (Keating, Fischer, Gordon, & Greenlee, 2005, p. 4). In the
1960s, this area of research became popular. At that time, Altman (1968)s model appeared to be
the most influential study in the field of predicting corporate failure (Bukhori, Othman, Aris, &
Omar, 2013, p. 375) based on the event of bankruptcy (Ohlson, 1980, p. 109), which relies on a
Z-Score based on five variables that had the highest predictive power (Keating, Fischer, Gordon, &
Greenlee, 2005, p. 4). Actually, five financial ratios determine the Z-Score. In practice, the lower the
Z-Score, the higher the probability of bankruptcy. Nowadays, academics and practitioners still widely
use the Z-Score as its accuracy reached 95% (Karamzadeh, 2013, p. 2008).

12
13

In French, Nomenclature europenne des activits conomiques. In Dutch, Europese activiteitennomenclatuur.


Abbreviated as NACE-Bel.

The next generation of studies on corporate bankruptcy prediction include Santomero & Vinso (1977),
Ohlson (1980) and Zmijewski (1984). However, Ohlson (1980) is the most widely used and cited.
His model is made for the probabilistic prediction of bankruptcy among industrial firms (Tevel, Katz,
& Brock, 2014, p. 6). Actually, his one-year prediction model relies on an O-Score, determined by
the combination of results to these six factors. His study concluded that it could indicate at what
point in time a companys financial statements were released to the public, and one could therefore
check whether the company entered bankruptcy prior to, or after the date of release of the financials
(Pongsatat, Ramage, & Lawrence, 2004, p. 3). Since his model could accurately predict corporate
failures, his model was used since then in thousands of studies, with or without variations, including
in the context of nonprofit organizations (Tevel, Katz, & Brock, 2014, p. 6). Consequentely, with
regard to the accuracy of these two models, their popularity are reflected in the frequent use of the
two models as empirical proxies for bankruptcy risk in accounting research (Keating, Fischer,
Gordon, & Greenlee, 2005, p. 5).
Before the nineties, most of the studies focused on the for-profit perspective. However, the lack of
research regarding the financial vulnerability of the nonprofit sector was enormous. For the purpose
of this study, we do not take into account the for-profit models as basis since, despite some authors
assurances, those models came out to be not suitable for the analysis of nonprofit organizations
(Tevel, Katz, & Brock, 2014, p. 11).

1. Program Expenditure Model


In 1991, a first model appeared. In fact, Tuckman & Chang (1991) were the first to focus on the
interest of this research, in which they defined an NPO as financially vulnerable if it [is] likely to cut
back its service offerings immediately when it experiences a financial shock (p. 445). A financial
shock can be for instance an economic recession with the potential consequences it includes, e.g.
the withdrawal of a major donor from the list of donators. Those NPOs that are not financially
vulnerable may not directly cut their service offerings.
Their study stated that the factors defining an NPOs financial situation are the generosity of its
supporters; the adequacy, stability, and diversity of its revenues; the quality of its management; its
capacity to withstand revenue fluctuations; and the size of its capital (Tuckman & Chang, 1991, p.
445). Based on these factors, they generate a model to measure an NPOs financial vulnerability.
This model proposes four criteria for judging this financial flexibility (Thomas & Trafford, 2013, p.
632) as shown in Table 4: revenue concentration, low or negative operating margins, inadequate
equity balances, and low administrative costs (Keating, Fischer, Gordon, & Greenlee, 2005, p. 6).
Revenue concentration ratio. An NPO with revenues coming from few resources sources may have
more difficulties to count on alternative sources in case of financial distress, hence being classified
as financially vulnerable. Conversely, those NPOs with large variety of sources of income will
overcome the distress with fewer difficulties.
Operating margin ratio. An NPO with a relatively low or negative level of margins may face more
difficulties to face a financial distress compared to an NPO with relatively higher level of operating
margins. The latter will be able to operate with a reduced operating margin rather than [reducing]
its program offerings (Greenlee & Trussel, 2000, p. 200), hence avoiding its classification as
financially vulnerable.
Equity ratio. An NPO with a low volume of equity relative to the size of its operations (revenues) may
face more difficulties to replace the withdrawal of a major donor from its resources sources after a
financial distress compared an NPO with a relatively higher volume. When the distress occurs, the
latter has the capacity to leverage its assets rather than reduce its program offerings (Greenlee &
Trussel, 2000, p. 200). Hence, the higher the equity balance, the bigger the likelihood the NPO will
not be classified financially vulnerable.

Administrative cost ratio. An NPO with relatively high administrative costs may have fewer difficulties
to face a financial distress compared to an NPO with relatively low administrative costs. If a shock
occurs, the former can cut more discretionary costs of its administration than the letter. Therefore,
a higher level of administrative costs reduce the probability of decreasing an NPOs service programs,
hence avoiding its classification as financially vulnerable.
Their methodology relied on a random selection of NPOs usng the U.S. Internal Revenue Service 14.
The four ratios were applied to a 1983 national sample of tax returns filed by 4,730 charitable
nonprofits (Tuckman & Chang, 1991, p. 445). Only organizations that filed this annual Form 990
are included in the study. They divided the results obtained into quintiles, and two categories of NPOs
appeared. On the one hand, an NPO is severally-at-risk if it falls in the lowest quintile for all four
criteria. On the other hand, an NPO is at-risk if it falls in the bottom quintile with respect to any one
of the four criteria (Tuckman & Chang, 1991, p. 451).
Table 4. Financial Vulnerability Ratios by Tuckman & Chang (1991)
Ratios

Formula

Sign

Revenue concentration ratio (CONCEN)

Revenue source j 2
(
)
Total revenues

Operating margin ratio (MARGIN)

Total revenues Total expense


Total revenues

Equity ratio (EQUITY)

Total equity
Total revenue

Administrative cost ratio (ADMIN)

Administrative expenses
Total expenses

Source: (Greenlee & Trussel, 2000, p. 203)


Tuckman & Chang (1991) derived some conclusions from their results, including the suggestion that
health care NPOs are among the ones most likely to be impacted by a financial shock (p. 458).
Nevertheless, their study is subject to an important bias linked to the IRS database usage. On the
one hand, as small organizations are underrepresented, it is likely that our study does not fully
capture the financial problems of social services agencies, which tend to be smaller than other
nonprofit organizations (Tuckman & Chang, 1991, p. 458). On the other hand, the study excludes
all those organizations that do not fill the Form 990, which implies the need to design an alternative
model to study the financial vulnerability of those NPOs that do not fil this form. However, their study
did not aim at literally predict future financial vulnerability but to measure an NPOs actual financial
vulnerability. Actually, they did not attempt to see if these variables could predict future financial
distress (Tevel, Katz, & Brock, 2014, p. 4). This desire of prediction is the core of the extensions
made by Greenlee & Trussel (2000) and by Trussel & Greenlee (2001).
1.1.

First Extension

Based on the groundwork of Tuckman & Chang (1991), Greenlee & Trussel (2000) were the first to
use the previously described four accounting ratios as well as methodologies common in the forprofit sector (Trussel, 2002, p. 18) to design a new model able to predict the financial vulnerability
of nonprofit organizations (Wang & Liu, 2010, p. 464). The core of this new model is to establish
measures predicting future financial health rather than identification of past problems (Cordery &
Baskerville, 2010, p. 2).

14

U.S. government agency responsible for tax collection and tax law enforcement (Internal Revenue Service, 2015).

Since the viewpoint changed from measuring actual distress to predicting future distress, a change
of definition became necessary. As noted earlier, Tuckman & Chang (1991) linked the financial
vulnerability to the reduction in program services after a financial shock. However, Greenlee & Trussel
(2000) ignores the immediacy of [this] definition (Thomas & Trafford, 2013, p. 634) and modified
it so that any NPO is financially vulnerable if it saw an overall decline in program expenses during a
three-year period (Keating, Fischer, Gordon, & Greenlee, 2005, p. 6) between 1986 and1995.
From program services to program expenses, two reasons explain the abandon of Tuckman & Chang
(1991)s definition. First, the accounting system does not fully capture program services (Greenlee
& Trussel, 2000, p. 202). Second, it is difficult to determine which charities experienced a financial
shock (Greenlee & Trussel, 2000, p. 202). To simplify the model of Greenlee & Trussel (2000), a
financially vulnerable NPO is one that shows a diminution in its program expenditures for instance in
1993, 1994, and 1995. They measure then the accounting ratios in 1992 to evaluate whether the
model could predict the NPOs actual financial situation in 1995. Therefore, the study uses the four
ratios to determine an NPOs probability to become financially vulnerable.
Table 5. Finanvial Vulnerability Model by Greenlee & Trussel (2000)
Financial Vulnerability Probability

1
(1+ )

Z = 3.0610 + 0.1153(EQUITY) + 1.2528(CONCEN) 2.2639(ADMIN) 3.4289(MARGIN)


Interpretation rules:
Strong indication of financial vulnerability if the probability is higher than 10%.
Strong indication of no financial vulnerability if the probability is lower than 7%.
Caution recommended in classifying NPOs if the probability is between 7% and 10%.
Source: (Cordery & Baskerville, 2010, p. 4)
Table 5 shows the formula defined by Greenlee & Trussel (2000) to compute the financial vulnerability
probability. They designed this formula, based on a sample of 3,151 NPOs from a multi-year database
of the IRS delivered by the National Centre for Charitable Statistics. This sample includes all NPOs
with assets higher than USD 10 million as well as 4,000 random NPOs of small size. Actually, to test
their model, they followed five steps.
First, the model applies the definition of financial vulnerability and divides the set of NPOs in two
groups: one financially vulnerable, and another one not financially vulnerable. Second, it randomly
selects about half of the NPOs of both groups. One will be the model sample to generate a logistic
regression model, and the remaining is used as a holdout sample. The model will test the significance
of the independent variables, with the predefined ratios. Third, it computes the financial vulnerability
probabilities for each NPO by using the actual ratios for each organization in the initial sample
(Greenlee & Trussel, 2000, p. 205). The authors used a binary logistic regression model to predict
the probability whether an NPO is financially vulnerable or not, based on the ratio. Fourth, it classifies
these probabilities and chooses cut-off points, i.e. 7% and 10%, to categorize each NPO as financially
vulnerable or not. Fifth, it computes the same probabilities of each NPO not selected in the second
step, to see if the model developed [] could accurately predict which organizations would become
financially vulnerable (Greenlee & Trussel, 2000, p. 206).
The study noted that CONCEN, MARGIN, and ADMIN are significant in predicting financial
vulnerability, but EQUITY is not. In other words, there is a significant relationship between financial
distress and three of Tuckman & Chang variables (Tevel, Katz, & Brock, 2014, p. 4). Actually, the
model was able to predict correctly 65 percent of the financially vulnerable organizations and 58
percent of all organizations as financially vulnerable or not (Trussel, 2002, p. 19). In addition, this
result was corroborated by the application of the same model to another sample, which gave similar
results the correct prediction of 61 percent of the financially vulnerable organizations and 61
percent of all organizations (Trussel, 2002, p. 19).

Nevertheless, the model assumes any NPO as financially vulnerable if it faces a program expenditures
reduction on a period of three years, but this precludes any determination of financial vulnerability
during the first four years of an organizations existence (Greenlee & Trussel, 2000, p. 207).
Moreover, Greenlee & Trussel (2000) noted that further developments would experiment new
definitions of financial vulnerability, increase the time scale of the model, and increase the sample
population. A revision of the model published by Trussel (2002) covers these further developments.
1.2.

Second Extension

Meanwhile Greenlee & Trussel (2000) designed their first development of Tuckman & Chang (1991)
model, Hager (2000) pointed the difference between the failure of arts NPOs and of the non-arts
NPOs. Indeed, arts NPOs were less likely to survive the study period (p. 4) compared to any other
types of NPOs. Actually, he applied the Tuckman & Chang (1991) model to this arts sector. Therefore,
he relied on the IRS database containing all registered arts organizations to obtain a sample of 7,226
organizations, subdivided in 7 subsectors: Visual arts organizations, Arts museums, Performing arts
centers and schools, Dance organizations, Theatre organizations, Instrumental and choral music, and
Generic performing arts (Hager, 2001, p. 383). Those organizations filed the annual Form 990 tax
return from the IRS for a period of three tax years as all variables are constructed from
organizational finances reported in 1990, 1991, and 1992 (Hager, 2001, p. 380).
Hager (2001) noted that although they showed promise, the financial vulnerability measures derived
by Tuckman and Chang (1991) could not be used across the board to predict failure in performing
arts [] organizations (Cordery & Baskerville, 2010, p. 4). In fact, to various degrees, the ratios
can explain the demise or longevity of a nonprofit organization (Hodge & Piccolo, 2011, p. 523).
He concluded that the four ratios of Tuckman & Chang (1991) are not applicable to all NPOs since
the applicability of these measures may vary across nonprofit sub-groups (McIndoe & Sullivan,
2014, p. 3). Furthermore, the conclusions from Hager (2001) are limited to only arts NPOs. Since
the focus of his study was too narrow, his findings do not appear to have been applied to other
specific areas of the non-profit sector (Cordery & Baskerville, 2010, p. 4). Therefore, although his
study was important to the development of the literature on NPOs financial vulnerability, we do not
rely on his findings in our research.

2. Net Assets Model


Considering the limitations of the Program Expenditure Model, this second model was developed by
Trussel (2002), Trussel et al. (2002), Trussel & Greenlee (2004) and is known as the Net Assets
Model. It provided responses to the previous critics. On the one hand, the ratios CONCEN and
MARGIN do not change at all. On the other hand, the author expanded the population of
organizations. Instead of using the IRS Statistics of Income database that is biased toward very
large organizations (p. 21), this model uses the IRS Core Files database, which includes all NPOs
that filled the 990 Form. Therefore, this new database gives a better representation of NPOs.
However, this broader database contains less data, which leads to some remarks. First, missing data
indicates the impossibility to apply some ratios: EQUITY was amended, and ADMIN was abrogated.
Second, the author added new indicators, namely SIZE and SECTOR, as shown in Table 6.
Debt ratio. The former equity ratio is replaced by a debt ratio, measuring the relative amount of
debt that an organization uses to finance its programs (Trussel, 2002, p. 20). An NPO with relatively
more debt may face more difficulties to finance its programs compared to an NPO with relatively less
debt. Therefore, an NPO is more vulnerable to financial distress if it shows a high debt ratio as it will
be limited in its ability to replace these lost revenues (Trussel, Greenlee, & Brady, 2002, p. 66).
Size. An NPOs size is an important factor to determine its likelihood to overcome financial distress.
Indeed, it is more likely to be financially vulnerable if it is small due to factors such as age,
reputation, economies of scale related to costs (Trussel, 2002, p. 20).

10

Sector. An NPOs sector links financial distress to financial capacities. Many macroeconomic factors,
such as inflation or recession, affect NPOs differently. These differences lead the NPO to be more
exposed or less exposed to financial distress, according to the sector it belongs. Trussel (2002)
expanded the model to 10 subsector variables: Arts, culture, and humanities; Education;
Environment and animals; Health; Human services; International, foreign affairs; Public, societal
benefit; Religion-related; Mutual-membership benefit; and Unknown (Trussel, 2002, p. 23).
Table 6. Financial Vulnerability Ratios by Trussel (2002)
Ratios

Formula

Sign

Revenue concentration ratio (CONCEN)

2
Revenue j
(
)
Total revenues

Surplus margin ratio (MARGIN)

Total revenues Total expense


Total revenues

Debt ratio (EQUITY)

Total liabilities
Total assets

Size (SIZE)

Natural log of total assets

Sector (SECTOR)

Dummy (or False) variable

N/A

Source: (Trussel, 2002, p. 21)


To execute it, the model follows four steps. First, it computes the financial indicators, mentioned in
Table 6, for the last five years. Second, it selects the subsector to which the NPO belongs (in this
case, health care). Third, it compares the financial indicators of the most recent year to the subsector
benchmark. Fourth, it computes the FVI index using the Z equation provided in Table 7 : the higher
the FVI, the greater the likelihood that the organization will have financial problems (Trussel,
Greenlee, & Brady, 2002, p. 68).
Table 7. Financial Vulnerability Model by Trussel (2002)
Financial Vulnerability Index

1
(1+ )

= 0.2475 + 0.8402(CONCEN) 1.3527(MARGIN) + 1.1080(DEBT) 0.1396(SIZE)


Interpretation rules:
Strong indication of financial vulnerability if the probability is higher than 20%.
Strong indication of no financial vulnerability if the probability is lower than 10%.
Caution recommended in classifying NPOs if the probability is between 10% and 20%.
Source: (Cordery & Baskerville, 2010, p. 4; Trussel, 2002, p. 29)
Trussel (2002) attests of three limitations that constraint the results. First, the model slightly
modifies the definition of a financially vulnerable NPO to be one that must have had more than a 20
percent decrease (Trussel, 2002, p. 20) in its net assets over three years. However, this percentage
is arbitrarily set as an assumption, hence limiting the research itself. Second, following the same
procedure as in the first model, Table 7 shows the formula to compute the financial vulnerability
probability. Nevertheless, this formula was designed to study NPOs only within the United States.
Third, the study is based on the period from 1996 to 1999, which is a short timeframe.

11

3. Net Earnings Model


Having regard to the two previous model, the Net Earnings model is an adaptation from the forprofit sectors bankruptcy and financial distress literature as well as from the nonprofit sector
concentration literature (Cordery, Sim, & Baskerville, 2013, p. 187). Although many authors wrote
on this model, we will cover only the most important papers.
First, Gilbert et al. (1990), in their attempt to predict the failure of NPOs, changed the view regarding
financial vulnerability as financial distress. In fact, the model describes that financial distress occurs
when facing negative cumulative earnings, income from continuing operations, over any consecutive
three-year period (Gilbert, Menon, & Schwartz, 1990, p. 163). Their study concluded that the
significant variables in predicting financial vulnerability are low earnings/total assets, low cash flows
from operations/current liabilities, and low stakeholders equity/total liabilities (Cordery, Sim, &
Baskerville, 2013, p. 190).
Second, Hodge & Piccolo (2005) used the resource dependence theory to study NPOs governance.
This theory analyses how an organizations external resources affect its behavior. The authors found
that the diversification of revenue sources (by opposition to the concentration of revenue sources) is
crucial for an organizations viability. Regarding revenue concentration, Carroll and Stater (2009)
confirmed that a diversified portfolio of revenue is more likely to yield organizational stability
(Cordery & Baskerville, 2010, p. 6). They have shown that the most stable NPOs are larger, have
greater growth potential and are located in urban areas (Cordery, Sim, & Baskerville, 2013, p. 190).
Moreover, Keating et al. (2005) noted that organizations should first give the priority to commercial
revenue rather than to donations, and second ensure the sufficiency of these donations and
endowments. They suggested an expanded model, which performs relatively better to predict
financial vulnerability. It includes two more variables in order to represent reliance on commercialtype activities to generate revenues and endowment sufficiency (Tevel, Katz, & Brock, 2014, p. 4).
To summarize, organizations are financially vulnerable if they do not receive sufficient revenue, more
specifically, if its cumulative earnings over four years were negative (Cordery, Sim, & Baskerville,
2013, p. 190). Finally, Table 8 shows the expected signs and significance of the models presented in
this literature review and Table 9 shows the evolution of the definition of financial vulnerability.
Table 8. Summary of estimated signs of the coefficients in a logistic model and
significance when predicting financial vulnerability

Measure

Expected
sign

Greenlee &
Trussel (2000)

Hager
(2001)

Sign

Sign

Sig

Sign

Sig

Trussel &
Greenlee
(2004)

Trussel
(2002)
Sig

Sign

Sig

Sign

EQUITY

CONCEN

ADMIN

Omitted

MARGIN

Legend: means significance, no means no significance (at

Keating et
al. (2005)

Sig

= 0.05)

Source: (Thomas & Trafford, 2013, p. 635)

12

Table 9. Summary of definitions for a financially vulnerable NPO


PROGRAM EXPENDITURE MODEL
Tuckman & Chang (1991)

the one that is likely to cut back its service offerings


immediately when it experiences a financial shock.

Hager (2001)

the one that fails to report to the I.R.S. over a period of


succeeding years.

Greenlee & Trussel (2000)

the one that had a decrease in its program expenditures over


three consecutive years.

NET ASSETS MODEL


Trussel et al. (2002)
Trussel (2002)
Trussel & Greenlee (2004)

the one that a decrease in its net assets over three consecutive
years.
the one that had more than a 20 percent decrease in its net
assets over three consecutive years.
the one that had more than a 50 percent decrease in its net
assets over three consecutive years.

NET EARNINGS MODEL


Gilbert et al. (1990)

the one that had negative cumulative earnings over three


consecutive years.
Source: (Cordery, Sim, & Baskerville, 2013, p. 189)

Methods
Since this study aims at describing the financial strength of the Belgian healthcare nonprofit sector,
our research question is stated as follow: What are the variables affecting the financial vulnerability
of the Belgian health care nonprofit organizations between 2010 and 2013? In consequence, we
performed a binominal logistic regression to determine these variables. We considered this type of
regression as the most appropriate tool for our analysis, first, because it leads to a binary variable
(financially vulnerable or not) and, second, because the independent predictor variables are
continuous.
In order to describe the financial vulnerability of this sector, we need a database to study it. Since
most of the models presented in our literature review were designed for the nonprofit sector within
the United States, these models used data from the IRS Statistics of Income database. In our case,
our study investigates the nonprofit sector of Belgium. Therefore, we rely on the Bel-First15 database
from the Bureau van Dijk (Bel-first, 2015), which is one of the most comprehensive available
database as it contains information about 1.2 million companies in Belgium (Vlerick Library, 2015).
Furthermore, this database also relies on the NACE-Bel nomenclature mentioned in our introduction.
Consequently, we built a sample of health care NPOs. In fact, we restrained the number of companies
from the database by selecting all active organizations with a legal status recognized as
Associations with the codes 86, 87, and 88. More specifically, Appendix 3 presents the exhaustive
listing of all types of activities included in these three health care NACE-Bel codes. In fact, these
Associations cover all types of institutions with legal personality as shown in Table 2. The total
number of nonprofit falling into these three categories amounts to 706 NPOs. However, although the
database is broad, all necessary information were not always available for all the studied years, that
is, 2010-2013. Therefore, 11 nonprofits had then to be removed from the sample since they had too
many missing values and the program could not impute them. As a result, the number of
observations fell to exactly 695 NPOs, as shown in Table 10.

15

Belgium Financial Reports and Statistics on Belgian and Luxembourg Companies

13

Table 10. Composition of sample


Total number of Associations in the database

1.754.614

Selected number of Associations in the sample


86 Human health activities
87 Residential care activities
88 Social work activities without accommodation

100
400
206

Less number of Associations without data for all studied years

11

Total number of Associations

695
Source: (Bel-first, 2015)

We adapted the methodology of Trussel et al. (2002) to our study. We used a definition of a financially
vulnerable NPO as the one showing a negative percentage change in net assets. Therefore, each
health care NPO is classified as financially vulnerable or not (Cordery, Sim, & Baskerville, 2013, p.
191). The dependent variable of the binominal logistic regression is financial vulnerability, with two
categories: it is not financially vulnerable (0) and it is financially vulnerable (1) as shown in Table
11. The independent variables are the predictors as defined in Table 6 : MARGIN, EQUITY, CONCEN,
SIZE, SECTOR87 (a binomial variable with values of 1 if the sector if 87, and 0 otherwise), and
SECTOR88 (a binomial variable with values of 1 if the sector is 88, and 0 otherwise).
Table 11. Financial Vulnerability Equations
Financially Vulnerable
4 1
1

Not Financially Vulnerable


4 1

>0

<0

As shown in Table 12, we state four hypothesis. The first hypothesis is expected to occur based on
Trussel & Greenlee (2000). In fact, an NPO with relatively low margins is more likely to be financially
vulnerable compared to an NPO with relatively higher margins. On the one hand, high margins lead
the NPO to able to operate without reducing its program services. On the other hand, if margins are
low, we expect the Belgian health care NPOs to reduce their program offerings. Consequently, the
NPO is likely to be classified as financially vulnerable. Therefore, we expect the independent variable
MARGIN to have a negative influence on the financial vulnerability of NPOs.
The second hypothesis is expected to occur based on Trussel (2002). In fact, an NPO with relatively
more debt is more likely to be financially vulnerable compared to an NPO with relatively less debt.
On the one hand, relatively low amounts of debt lead the NPO to be able to finance its program
services. On the other hand, if the ratio of liabilities over assets is high, we expect the Belgian health
care NPOs to not be able to finance their program services. Consequently, the NPO is likely to be
classified as financially vulnerable. Therefore, we expect the independent variable EQUITY to have a
significant positive influence on the financial vulnerability of NPOs.
The third hypothesis is expected to occur based on Cordery & Baskerville (2010) and Thomas &
Traford (2013). In fact, an NPO with relatively diverse sources of revenues is less likely to be
financially vulnerable compared to an NPO with relatively narrow sources of revenues. On the one
hand, if the portfolio of funding is diverse, it will lead to the stability of the NPO. On the other hand,
if the concentration ratio is high, we expect the Belgian health care NPOs to not be able to easily find
alternative sources of finance if one of its donors cease to fund it. Consequently, the NPO is likely to
be classified as financially vulnerable. Therefore, we expect the independent variable CONCEN to
have a significant positive influence on the financial vulnerability of NPOs.

14

The fourth hypothesis is expected to occur based on Trussel et al. (2002). In fact, a relatively bigger
NPO is less likely to be financially vulnerable compared to a relatively smaller NPO. On the one hand,
if the size of the organization is large, it will lead to the NPO to overcome more easily any financial
distress. On the other hand, if the size is smaller, we expect the Belgian health care NPOs to not
easily overcome any financial distress. Consequently, the NPO is likely to be classified as financially
vulnerable. Therefore, we expect the independent variable SIZE to have a significant negative
influence on the financial vulnerability of NPOs.
The fifth hypothesis is expected to occur. In fact, we decomposed the variable SECTOR by creating
the dummy variables SECTOR87 and SECTOR88, with the sector 86 as reference. In fact, these
dummy variables indicate an expected shift in the dependent variable due to the influence of
categories, i.e. due to the belonging of the NPO to specific subsectors of activities in the health care
nomenclature. Therefore, we expect the independent variable SECTOR to have a significant influence
on the financial vulnerability of NPOs.
Table 12. Financial Vulnerability Hypothesis
H1

The surplus margin has a significant negative influence on financial vulnerability.

H2

The equity has a significant positive influence on financial vulnerability.

H3

The revenue concentration index has a significant positive influence on financial vulnerability.

H4

The organization size has a significant negative influence on financial vulnerability.

H5

The sector has a significant influence on financial vulnerability.

We performed two preliminary actions on this sample. First, we imputed the missing values in the
data set, which is a practice of 'filling in' missing data with plausible values (PennState Department
of Statistics, 2015). In our case, we performed this practice via the estimation of a regression model
that is predicting values based on other variables. It actually use the variables that we already have
in order to substitute the predicted value to be an actual data. In fact, this is the best technique for
the amount (about 15%) of missing data that we have, and their type (missing not-at-random).
Second, we aggregated the independent variables MARGIN, EQUITY, CONCEN and SIZE for the period
2010-2013 by averaging them, in order to use the resulting variables as predictors (Greenlee &
Trussel, 2000, p. 4; Cordery & Baskerville, 2010, p. 9; Tevel, Katz, & Brock, 2014, p. 8; Hager, 2001,
pp. 380, 384).
Moreover, we checked two assumptions. First, we assumed that there is no significant
multicollinearity16 (i.e. the independent variables are not strongly correlated with one another). To
verify this assumption, we run a linear regression analysis with the same variables as for the
binominal logistic regression. As shown in Table 13, the multicollinearity is not a threat to our model
under the following condition: for all the variables, the variance inflation factor17 must be lower than
3. Indeed, it is effectively the case. In consequence, there is no significant multicollinearity in our
model. Therefore, the first assumption is verified. Second, we assumed that there are no important
outliers in the data series. As shown in Table 14, the note a states there are no outliers in the data
set. Therefore, the second assumption is verified as well.

16

If we have important multicollinearity in our model, the variances of the coefficients increase very much, making those
coefficients unstable and unreliable and reduces the prediction power of the model. The VIFs show us whether the coefficients
variances are inflated due to high multicollinearity. The lower this indicator the better. A value of 1 means that the corresponding
predictor is not at all correlated with the others (that is a rather theoretical situation). A value between 1 and 5 shows a low
correlation. A value between 5 and 10 shows a moderate correlation. A value over 10 indicates a strong correlation.
17
Abbreviated as VIF (VIFs for the plural).

15

Table 13. Collinearity Statistics

Table 14. Outliers table

Table 15 shows a summary statistics with the relevant information retrieved from the dataset. As
expected, we obtained differences between the 224 financially vulnerable NPOs and the 471 not
financially vulnerable NPOs. First, mean MARGIN is lower for FV NPOs, which implies that financially
vulnerable NPOs have a lower surplus margin (-0.27%) than those that are not financially vulnerable
(1.7%). Second, mean EQUITY is lower for not financially vulnerable NPOs, which implies that
financially vulnerable NPOs have more debt (45.55%) than those that are not financially vulnerable
(41.3%). Third, mean CONCEN is lower for financially vulnerable NPOs, which implies that financially
vulnerable NPOs have lower concentration of revenues (73%) than not financially vulnerable
(74.9%). Fourth, mean SIZE is higher for not financially vulnerable NPOs, which implies that
financially vulnerable NPOs are smaller in average (15.283) than those that are not financially
vulnerable (15.649). Furthermore, Appendixes 4, 5, and 6 show these statistics per sector of activity,
i.e. sectors 86, 87 and 88.
Table 15. Summary Statistics Financially Vulnerable and Not Financially Vulnerable
Status

Number

FV

224

NFV

471

Variable

Mean

Standard Deviation

MARGIN
EQUITY
CONCEN
SIZE
MARGIN
EQUITY
CONCEN
SIZE

-0.027
0.455
0.730
15.283*
0.017
0.413
0.749
15.649*

0.503
0.274
0.158
1.479
0.092
0.257
0.152
1.648

* The difference between FV and NFV mean values is significant at = 0.05, by using t-statistic test.

In addition, Table 16 provides descriptive statistics, namely t-statistic and Pearson correlations. First,
the one-sample or univariate t test compares the mean of a sample with a given value, often zero
(IDRE, 2015). Second, only two Pearson correlation coefficients are statistically significant from this
matrix. On the one hand, the correlation between MARGIN and SIZE is -0.175 and it is statistically
significant. On the other hand, the correlation between CONCEN and SIZE is -0.109 and it is
statistically significant as well.
Table 16. Descriptive Statistics for the Independent Variables
Univariate test
t-statistic
Pearson correlations
EQUITY
CONCEN
SIZE

MARGIN

EQUITY

CONCEN

SIZE

-.832

24.864*

69.123*

154.799*

-.011
-.041
-.175*

.096
-.080

-.109*

* Significance at = 0.01 (two-tailed).

16

Results
Before interpreting the binominal logistic regression, a two-fold analysis of the independent variables
is necessary. First, SPSS generates the omnibus tests of model coefficients, analyzing whether the
independent variables are good predictors, overall, for the outcome variable of financial vulnerability.
In our case, as shown in Table 17, the chi-square test value with 6 degrees of freedom is 19.242,
p=0.004. As we added all explanatory variables in one block, there is only one-step and all three
values are the same. Since the significance value (or p value) is lower than 0.05, our independent
variables satisfactorily explain the dependent variable. Second, SPSS generates the HosmerLemeshow test, analyzing how bad the independent variables predict the dependent variable. For
our model to be valid, this test should not be significant. In our case, as shown in Table 18, the
dependent variables adequately predict the outcome variable since p=0.067 > 0.05.
Subsequently, SPSS generates the pseudo-R values with both the Cox-Snell and Nagelkerke
indicators, telling us what percentage of the variation in the dependent variable is explained by the
independent variables. On the one hand, the Cox-Snell statistics can take any value between 0
inclusive and 1 exclusive. On the other hand, the Nagelkerke indicator is a corrected Cox-Snell
indicator, normalized so that it can take the value 1. For this reason, this last indicator is often the
only one to be reported and interpreted. In our case, as shown in Table 19, the independent variables
explain about 3.8% of the variation in the outcome variable, as assessed by the Nagelkerke statistics.
Table 17. Omnibus Tests of Model Coefficients

Table 18. Hosmer-Lemeshow Test

Table 19. Good-of-fit indicators

With regard to these preliminary analyses, the binomial logistic regression generates the coefficients
table as shown in Table 20. It presents the output resulting from including all of the candidate
predictor variables in the equation (Stat Tutorials, 2015) to show the individual impact of each
predictor on the financial vulnerability of NPOs. Actually, we will report three elements for each
variable. First, we describe the regression coefficient B18 used to determine which of the independent
variables have the biggest effect on the dependent variable. Second, we describe the results of the
Wald test19 used to know whether the coefficient is statistically significant20, i.e. whether the
corresponding variable has a good predicting power. Third, we describe the antilogarithm21 of the
coefficient Exp(B) used to know the chance that an NPO is financially vulnerable, depending on the
corresponding predictor. Our results are presented in two parts. We will present first the coefficients
for the continuous predictors (MARGIN, EQUITY, CONCEN, and SIZE), and second the coefficients for
the categorical predictors (SECTOR87 and SECTOR88).
18

The regression coefficient B represents the relationship between a given explanatory variable and the outcome variable. It is
precisely how much the outcome variable changes after a change of one unit in the explanatory variable (ReStore, 2015).
19
The Wald test is a way of testing the significance of particular explanatory variables in a statistical model (Crichton, 2001).
20
For the statistically significant coefficients, we are going to report the 95% confidence interval.
21
If x is the logarithm of y, then y is the antilogarithm of x.

17

Table 20. Variables in the Equation

1. Continuous Predictors Analysis


Regarding the continuous predictors, the four ratios show the following results. First, the regression
coefficient for MARGIN is -0.971, and it is not statistically significant as assessed by the Wald test
(p=0.250 > 0.05). The antilogarithm of the regression coefficient is 0.379. In consequence, the
hypothesis H1 is not verified. Second, the regression coefficient for EQUITY is 0.558, and it is nearly
significant as assessed by the Wald test (p=0.077 < 0.10). The antilogarithm of the regression
coefficient is 1.748. In consequence, the hypothesis H2 is partially verified. Third, the regression
coefficient for CONCEN is -1.283, and it is statistically significant as assessed by the Wald test
(p=0.024 < 0.05). The antilogarithm of the regression coefficient is 0.277. In consequence, the
hypothesis H3 is verified. Fourth, the regression coefficient for SIZE is -0.160, and it is statistically
significant as assessed by the Wald test (p=0.007 < 0.05). The antilogarithm of the regression
coefficient is 0.852. In consequence, the hypothesis H4 is verified.
Therefore, the variables with the strongest influence on financial vulnerability are SIZE, CONCEN,
and EQUITY. MARGIN does not seem to have a significant influence. Based on the antilogarithms, we
can determine the chance (odds) of getting financially vulnerable and compute the probability to be

financially vulnerable. The formula for the probability is =


, where o are the odds. The reference
1+

values for the continuous predictors are considered the lower values: NPOs with higher values are
compared with those with lower values in terms of odds to become financially vulnerable. Table 21
presents, for each continuous predictor, the odds of becoming financially vulnerable as well as the
probabilities to be and to not be financially vulnerable, for the cases with high values of the predictors.
For the cases with the low values, the probabilities will be reversed.
Table 21. Odds of financial vulnerability for continuous predictors
Variable

Odds Exp(B) of becoming


financially vulnerable

Probability of being
financially vulnerable

Probability of not being


financially vulnerable

MARGIN
EQUITY
CONCEN
SIZE

0.379
1.748
0.277
0.852

0.275
0.636
0.217
0.460

0.725
0.364
0.783
0.540

Four consequences are retrieved from this analysis. First, the odds that an NPO with big surplus
margin is financially vulnerable are about 62% (1-0.379) lower than the same odds for an NPO with
small surplus margin or 2.6 times lower (1/0.379). Thus, the probability that an NPO with big
surplus margin is financially vulnerable is 27.5%, and the probability that an NPO with big surplus
margin is not financially vulnerable is 72.5%. For an NPO with small surplus margin, these
probabilities are reversed.

18

Second, the odds that an NPO with high equity is financially vulnerable are about 75% (1-1.748)
higher than the same odds for an NPO with low equity or 1.7 times higher. Thus, the probability
that an NPO with high equity is financially vulnerable is 63.6%, and the probability that an NPO with
high equity is not financially vulnerable is 36.4%. For an NPO with low equity, these probabilities are
reversed.
Third, the odds that an NPO with high revenue concentration index is financially vulnerable are about
72% (1-0.277) lower than the same odds for an NPO with low revenue concentration index or 3.6
times lower (1/0.277). Thus, the probability that an NPO with high revenue concentration index is
financially vulnerable is 21.7%, and the probability that an NPO with high revenue concentration
index is not financially vulnerable is 78.3%. For an NPO with low revenue concentration index, these
probabilities are reversed.
Fourth, the odds that an NPO with big size is financially vulnerable are about 15% (1-0.852) lower
than the odds for an NPO with small size or 1.2 times lower (1/0.852). Therefore, the chance of
financial vulnerability decreases by 15% when the NPO size increases. Thus, the probability that an
NPO with big size is financially vulnerable is 46%, and the probability that an NPO with big size is
not financially vulnerable is 54%. For an NPO with small size, these probabilities are reversed.

2. Categorical Predictors Analysis


Regarding the categorical predictors, the coefficients are only provided for the category 1 for both
variables SECTOR87 and SECTOR88. As a reminder, these variables take the value 1 if the NPO
belongs respectively the sector 87 or the sector 88, and 0 otherwise. Therefore, for the reference
category 0, the coefficients are easy to compute, given that the sum of the coefficients for each
categorical variable is equal to zero.
Subsequently, the two dummy variables SECTOR87 and SECTOR88 show the following results. First,
the regression coefficient for SECTOR87 (category 1) is -0.195, and it is not statistically significant
as assessed by the Wald test (p=0.448 > 0.05). The antilogarithm of the regression coefficient is
0.822. In consequence, the hypothesis H5 is not verified. Second, the regression coefficient for
SECTOR88 (category 1) is -0.357, and it is not statistically significant as assessed by the Wald test
(p=0.231 > 0.05). The antilogarithm of the regression coefficient is 0.700. In consequence, the
hypothesis H5 is not verified as again.
Therefore, the dummy variables SECTOR87 and SECTOR88 do not seem to have a substantial
influence on financial vulnerability. Table 22 presents, for each categorical predictor, the odds of
being financially vulnerable as well as the probabilities to be and not to be financially vulnerable, for
the category 1. For the reference category 0, these probabilities will be reversed.
Table 22. Odds of financial vulnerability for categorical predictors
Variable

Odds of becoming financially


vulnerable Exp(B)

Probability of being
financially vulnerable

Probability of not being


financially vulnerable

SECTOR 87
SECTOR 88

0.822
0.700

0.451
0.412

0.549
0.588

Two consequences are retrieved from this analysis. First, the odds that an NPO belonging to sector
87 is financially vulnerable are about 18% lower than the same odds for an NPO belonging to sector
86 (reference) (1-0.822) or 1.2 times lower (1/0.822). Thus, the probability that an NPO belonging
to sector 87 is financially vulnerable is 45.1%, and the probability that an NPO with big surplus
margin is not financially vulnerable is 54.9%. Second, the odds that an NPO belonging to sector 88
is financially vulnerable are about 30% lower than the same odds for an NPO belonging to sector 86
(reference) (1-0.700) or 1.4 times lower (1/0.700). Thus, the probability that an NPO belonging to
sector 88 is financially vulnerable is 41.2%, and the probability that an NPO belonging to sector 88
is not financially vulnerable is 58.8%.

19

Out of the NPOS that are not financially vulnerable, 98.1% were correctly classified, while out of
those that are financially vulnerable, 4% were correctly classified. Overall, 67.8% of the NPOs were
correctly classified (471 out of 695) as shown in Table 23.
Table 23. Classification table

Discussions
With regard to these results, we will discuss five elements. We will firstly discuss whether our results
demonstrate our hypotheses. We will secondly explain how our results give an answer to our research
question. We will thirdly develop some arguments to convince the reader that our study has changed
the fields knowledge. We will fourthly deal with the international component of our study. We will
finally raise some recommendations for further studies.
First, our study show diverse results regarding the hypothesis. The first hypothesis stating a
significant influence of surplus margin on financial vulnerability is not verified. This finding contradicts
both Trussel and Greenlee (2001) and Trussel (2002) concluding that MARGIN significantly influence
financial vulnerability. The second hypothesis stating a significant influence of equity on financial
vulnerability is partly verified. Although a predictor that has a low p-value is likely to be a meaningful
addition to [the] model because changes in the predictor's value are related to changes in the
response variable (Frost, 2013) and despite very low p-value generated by the Wald test, the
regression coefficient mitigates the verification of this hypothesis. This finding is relatively in line
both with Trussel (2001) concluding that EQUITY significantly influence financial vulnerability, and
Trussel (2002) concluding that EQUITY does not significantly influence financial vulnerability.
The third hypothesis stating a significant influence of revenue concentration index on financial
vulnerability is verified. This finding contradicts Trussel and Greenlee (2001) concluding that CONCEN
does not significantly influence financial vulnerability, and is in line with Trussel (2002) concluding
that CONCEN significantly influence financial vulnerability. The fourth hypothesis stating a significant
influence of organization size on financial vulnerability is verified. This finding is in line with Trussel
and Greenlee (2001) concluding that SIZE significantly influence financial vulnerability, and
contradicts Trussel (2002) concluding that SIZE does not significantly influence financial
vulnerability. The fifth hypothesis stating a significant influence of sector on financial vulnerability is
not verified. This contradicts Trussel and Greenlee (2001) concluding that SECTOR significantly
influence financial vulnerability.
Second, our research question examined What are the variables affecting the financial vulnerability
of the Belgian health care nonprofit organizations between 2010 and 2013? The predictors generally
have a substantial influence on financial vulnerability, as assessed by the omnibus tests and the
Hosmer-Lemeshow test. The predictors with the most important contributions are size, revenue
concentration index and equity, in this order. For these variables, the regression coefficients are
statistically significant (or borderline significant). The predictor surplus margin has a meaningful
effect on the outcome variable, though the regression coefficient is not statistically significant. As for
the variable sector, it does not seem to have an important effect on financial vulnerability.

20

Third, the contribution of our study entail a two-fold change in the fields knowledge. On the one
hand, this study has changed the fields knowledge as it intended to be the very first empirical
groundwork for the literature on the Belgian nonprofit sector. In fact, most experts generally
discussed the financial vulnerability of the nonprofit sector in Belgium whereas no statistical research
has ever been conducted yet. On the other hand, this study has changed the fields knowledge also
because we adapted Trussel (2002) in an entirely other environment than the United States, for
which it has been designed. Indeed, it is particularly relevant since most US-developed models have
been found to exhibit different levels of explanatory power in different segments (Cordery, Sim, &
Baskerville, 2013, p. 190) of the nonprofit sector.
Fourth, and consequently to our third discussion, we can reveal the potential international component
of our study by its geographical context. Although our study focused on a particular segment of the
Belgian nonprofit sector, it may be relevant to compare our results in a broader geographical context,
weather it is to European results, to O.E.C.D. results, or merely to international results. In fact, such
comparisons allow us to evaluate the strength of the nonprofit sectors across countries. Furthermore,
our study can be generalized and applied to other context than the Belgian one since any national
nonprofit sector can be the focus the analysis model that we applied.
Finally, two recommendations can be raised for further studies. On the one hand, a first limitation of
our study has regard with the relatively small database. Although the database contained a wide
number of associations, our study did not cover all of them first because of the limited availability of
data, and second because there were missing values. In fact, we have used the data imputation as
a technique to overcome the missing data but this can also be considered as a weakness of our
study. However, a pairwise deletion technique could not be conceivable because the amount and the
type of missing data, as explained in our methodology, since the number of valid cases in our sample
would decrease substantially. Therefore, further studies should collect comprehensive data over a
wider timeframe than only four years. On the other hand, a second limitation of our study has regard
with the research focus, restrained to only on one segment of the nonprofit sector, i.e. the health
care segment. Therefore, further researches should expand it to include all the nonprofit subsectors.

Conclusion
Throughout the development of this Masters thesis, the core idea behind our investigation was to
lay the empirical foundation of the literature on the Belgian nonprofit sector. In fact, this lack of
empirical research has been a constant motive for pursuing our study. As explained in the
introduction, the health care subsector accounts for more than half of the public funding allocated to
the nonprofit sector and represents about a third of the value added of the overall nonprofit sector
(Vander Donckt & Rigo, 2013, pp. 21-22). Despite the predominance of this sector, a thorough
research regarding the financial vulnerability of health care organizations was necessary to
understand the impact of certain parameters on their ability to overcome any financial distress.
Financial distress can occur at any time for any organization, whether for-profit or nonprofit. Margins
can fall, donators can withdraw their contributions, debt can increase, and the size of the organization
can be impacted as well. Therefore, an organizations management should be concerned by its
vulnerability. This study, overall, concludes that among the five parameters defined by Trussel et al.
(2002), only three of them organizational size, revenue concentration, and debt ratio are
significant parameters that affected the financial vulnerability of Belgian health care nonprofit
organizations between 2010 and 2013. This may raise some questions for further researches. Since
these three variables are significant, how did these parameters vary over the 2008-2009 financial
crisis period? To what extend the crisis has impacted those organizations in their financial
vulnerability? How did they overcome this shock? Could their situation before the crisis be able to
tackle such distress? New doors are open for further investigations.

21

References
Acx, R., Rigo, C., & Vander Donckt, M. (2011). Le poids conomique des associations en Belgique.
Bruxelles: Fondation Roi Baudoin.
Arshad, R., Bakar, N. A., Razali, W. A., & Omar, N. (2013). Financial Vulnerability, Risk Management
and Accountability of Non-Profit Organisations. Journal of Energy Technologies and Policy,
3(11), 7.
Banque Nationale de Belgique. (2004). Le compte satellite des institutions sans but lucratif : 20002001. Bruxelles.
Banque Nationale de Belgique. (2010). Le compte satellite des institutions sans but lucratif : 20002008. Bruxelles.
Banque Nationale de Belgique. (2012). Le compte satellite des institutions sans but lucratif : 20092010. Bruxelles.
Bel-first. (2015). Retrieved March 29, 2015, from https://belfirst.bvdinfo.com
Bukhori, M., Othman, R., Aris, N., & Omar, N. (2013). Assessing Financial Vulnerability of
Cooperative. Journal of Energy Technologies and Policy, 3(11), 374-381.
Business

Belgium.

(2010).

Retrieved

January

28,

2015,

from

http://business.belgium.be/en/managing_your_business/setting_up_your_business
Cordery, C. J., Sim, D., & Baskerville, R. F. (2013). Three models, one goal: Assessing nancial
vulnerability in New Zealand amateur sports clubs. Sport Management Review, 16(2), 186199. doi:10.1016/j.smr.2012.08.002
Cordery, C., & Baskerville, R. (2010). Assessing financial vulnerability in nonprofit sports
organizations. Wellington: Victoria University.
Crichton, N. (2001). Information Point: Wald Test. Journal of Clinical Nursing, 774. Retrieved from
www.blackwellpublishing.com/specialarticles/jcn_10_774.pdf
Davagle, M. (1997). Memento des ASBL. Kluewer Edition Juridiques.
Europa.

(2015).

Retrieved

April

3rd,

2015,

from

http://ec.europa.eu/competition/mergers/cases/index/nace_all.html
Frost, J. (2013). How to Interpret Regression Analysis Results: P-values and Coefficients. Retrieved
April

21,

2015,

from

http://blog.minitab.com/blog/adventures-in-statistics/how-to-

interpret-regression-analysis-results-p-values-and-coefficients
Gilbert, L. R., Menon, K., & Schwartz, K. B. (1990). Predicting bankruptcy for rms in nancial
distress. Journal of Business Finance & Accounting, 17(1), 161-171.
Greenlee, J. S., & Trussel, J. M. (2000). Predicting the Financial Vulnerability of Charitable
Organizations.

Nonprofit

Management

&

Leadership,

11(2),

199-210.

doi:10.1002/nml.11205
Hager, M. A. (2000). The Survivability Factor: Research on the Closure of Nonprofit Arts
Organizations. Americans for the Arts Monograph, 4(1), pp. 1-12.

22

Hager, M. A. (2001). Financial Vulnerability Among Arts Organizations: A Test of the Tuckman-Chang
Measures.

Nonprofit

and

Voluntary

Sector

Quarterly,

30(2),

pp.

376-392.

doi:10.1177/0899764001302010
Hodge, M. M., & Piccolo, R. F. (2011). Nonprofit Board Effectiveness, Private Philanthropy, and
Financial Vulnerability. Public Administration Quarterly, 35(4), 520-550.
IDRE.

(2015).

Retrieved

April

29,

2015,

from

http://www.ats.ucla.edu/stat/spss/output/Spss_ttest.htm
Internal Revenue Service. (2015). Retrieved March 1, 2015, from www.irs.gov
Karamzadeh, M. S. (2013). Application and Comparison of Altman and Ohlson Models to Predict.
Research Journal of Applied Sciences, Engineering and Technology, 5(6), 2007-2011.
Keating, E. K., Fischer, M., Gordon, T. P., & Greenlee, J. (2005). Assessing Financial Vulnerability in
the Nonprofit Sector. Cambridge: Harvard University.
Mare,

M.

(2010a).

Compte

satellite

de

ISBL.

Retrieved

March

25,

2015,

from

www.ces.ulg.ac.be/fr_FR/services/cles/dictionnaire/c-d/compte-satellite-de-isbl
Mare, M. (2010b). Qu'est ce qu'un compte satellite? Retrieved March 25, 2015, from
www.ces.ulg.ac.be/fr_FR/services/cles/notes-de-synthese/comprendre-le-compte-satellitedes-isbl-en-belgique/qu-est-ce-qu-un-compte-satellite
Mare, M. (2010c). CES ULg - Comprendre le compte satellite des ISBL en Belgique. Retrieved March
25, 2015, from www.ces.ulg.ac.be/fr_FR/services/cles/notes-de-synthese/comprendre-lecompte-satellite-des-isbl-en-belgique
Mare, M., Gijselinckx, C., Loose, M., Rijpens, J., & Franchois, E. (2008). Les associations en
Belgique. Bruxelles: Fondation Roi Baudoin.
Mare, M., Mertens, S., Defourny, J., Develtere, P., Raymaekers, P., & Meireman, K. (2005). Le
secteur associatif en Belgique : Une analyse quantitative et qualitative. Bruxelles: Fondation
Roi Baudoin.
McIndoe, H., & Sullivan, F. (2014). Nonprofit Responses to Financial Uncertainty: How Does Financial
Vulnerability Shape Nonprofit Collaboration? Journal of Management and Sustainability, 4(3),
1-15. doi:10.5539/jms.v4n3p1
Mertens, S. (2004). Le compte satellite des associations en Belgique. Revue internationale(292), pp.
42-58.
Mertens, S., & Lefbvre, M. (2001). Testing the handbook of nonprofit institutions in the system of
national accounts: Belgian report. Center for Social Economy - ULg.
Mertens, S., Adam, S., Defourny, J., Mare, M., Pacolet, J., & Van de Putte, I. (1999). Chapter 2:
Belgium. In L. M. Salamon, H. K. Anheler, R. List, S. Toepler, & S. Wojciech Sokolowski and
Associates, Global Civil Society: Dimensions of the Nonprofit Sector (pp. 43-61). Baltimore:
John Hopkins Center for Civil Society Studies.
Ohlson, J. A. (1980). Financial Ratios and the Probabilistic Prediction of Bankruptcy. Journal of
Accounting Research, 18(1), 109-131.
PennState Department of Statistics. (2015). Retrieved April 19, 2015, from Multiple Imputation:
http://sites.stat.psu.edu/~jls/mifaq.html

23

Pongsatat, S., Ramage, J., & Lawrence, H. (2004). Bankruptcy Prediction for Large and Small Firms
in Asia: A Comparison of Ohlson and Altman. Journal of Accounting and Croporate
Governance, 1(2), 1-13.
Portail

Belgium.

(2012).

Retrieved

January

30,

2015,

from

www.belgium.be/fr/economie/entreprise/creation/types_de_societe
ReStore. (2015). Retrieved April 19, 2015, from Using Statistical Regression Methods in Education
Research:

http://www.restore.ac.uk/srme/www/fac/soc/wie/research-

new/srme/glossary/index2f4f.html?selectedLetter=R#regression-coefficient
Service

Public

Fdral

Justice.

(2013).

Retrieved

January

26,

2015,

from

http://justice.belgium.be/fr/binaries/ASBL-FR_tcm421-194125.pdf
Stat Tutorials. (2015). Retrieved April 19, 2015, from

Logistic Regression using SPSS:

http://www.stattutorials.com/SPSS/TUTORIAL-SPSS-Logistic-Regression.htm
Statistics

Solutions.

(2015).

Retrieved

April

19,

2015,

from

Multicollinearity:

http://www.statisticssolutions.com/multicollinearity/
Tevel, E., Katz, H., & Brock, D. M. (2014). Nonprofit Financial Vulnerability: Testing Competing
Models, Recommended Improvements, and Implications. VOLUNTAS: International Journal
of Voluntary and Nonprofit Organizations, 17.
Thomas, R., & Trafford, R. (2013). Were UK Culture, Sport and Recreation Charities Prepared for the
2008 Economic Downturn? An Application of Tuckman and Chang's Measures of Financial
Vulnerability. VOLUNTAS: International Journal of Voluntary and Nonprofit Organizations,
24(3), 630-648. doi:10.1007/s11266-012-9273-1
Trussel, J. M. (2002). Revisiting the Prediction of Financial Vulnerability. Nonprofit Management &
Leadership, 13(1), 17-31.
Trussel, J. M., Greenlee, J. S., & Brady, T. (2002). Predicting Financial Vulnerability in Charitable
Organizations. The CPA Journal, 72(6), 66-69.
Tuckman, H. P., & Chang, C. F. (1991). A Methodology for Measuring the Financial Vulnerability of
Charitable Nonprofit Organizations. Nonprofit and Voluntary Sector Quarterly, 20(4), 445460. doi:10.1177/089976409102000407
United Nations. (2003). Handbook on Non-Profit Institutions in the System of National Accounts. New
York.
Vander Donckt, M., & Rigo, C. (2013). Le poids conomique des institutions sans but lucratif en
Belgique. Bruxelles: Fondation Roi Baudoin.
Vlerick Library. (2015). Retrieved March 29, 2015, from http://www.vlerick.com/en/research-andfaculty/vlerick-library/databases/belfirst
Wang, H.-C., & Liu, A. Y. (2010). Performance and Vulnerability of Non-Profit Organization in Taiwan.
(IEEE, Ed.) Computer, Communication, Control and Automation (3CA), 2, 463-465.
doi:10.1109/3CA.2010.5533355

24

Appendixes
Appendix 1. Employment in IWPP (2004-2010)
2004

2005

2006

2007

2008

2009

2010

Number of IWPP

16,913

17,015

17,400

17,624

17,794

18,630

18,731

Wage employment
(absolute value)

368,300

382,900

401,500

416.500

431,700

430,600

446,500

Wage employment
(as % of the
10.5%
10.8%
11.1%
11.3%
11.5%
11.5%
11.9%
countrys wage
employment)
Source : (Mare, Gijselinckx, Loose, Rijpens, & Franchois, 2008, p. 13; Mare, 2010a; Banque
Nationale de Belgique, 2010, p. 7; Banque Nationale de Belgique, 2012, p. 8; Vander Donckt &
Rigo, 2013, p. 17)

Appendix 2. Employment in IWPP, by sector of activity (2010)


Wage employment in 2010 (in
thousands, units)
Agriculture and industry
10.3
Services
61.7
Education
8.9
Human health
146.7
Social action
163.6
Arts, spectacles, and other recreational activities
16.0
Other service activities
39.5
Total of IWPP
446.5
Rest of the economy
3,310.8
Source : (Vander Donckt & Rigo, 2013, p. 30)

25

Appendix 3. NACE-Bel 2008 Nomenclature Code Q


Q86

Human health activities

Q86.1

Hospital activities

Q86.1.0

Hospital activities

Q86.2

Medical and dental practice activities

Q86.2.1

General medical practice activities

Q86.2.2

Specialist medical practice activities

Q86.2.3

Dental practice activities

Q86.9

Other human health activities

Q86.9.0

Other human health activities

Q87

Residential care activities

Q87.1

Residential nursing care activities

Q87.1.0

Residential nursing care activities

Q87.2

Residential care activities for mental retardation, mental health and substance abuse

Q87.2.0

Residential care activities for mental retardation, mental health and substance abuse

Q87.3

Residential care activities for the elderly and disabled

Q87.3.0

Residential care activities for the elderly and disabled

Q87.9

Other residential care activities

Q87.9.0

Other residential care activities

Q88

Social work activities without accommodation

Q88.1

Social work activities without accommodation for the elderly and disabled

Q88.1.0

Social work activities without accommodation for the elderly and disabled

Q88.9

Other social work activities without accommodation

Q88.9.1 -

Child day-care activities

Q88.9.9 -

Other social work activities without accommodation n.e.c.


Source : (Europa, 2015)

26

Appendix 4. Summary Statistics Financially Vulnerable and Not Financially Vulnerable


Sector 86
Status
Number
Variable
Mean
Standard Deviation
FV
31
MARGIN
0.021
0.089
EQUITY
0.492
0.242
CONCEN
0.752
0.171
a
SIZE
15.713
1.955
NFV
67
MARGIN
0.017
0.057
EQUITY
0.450
0.215
CONCEN
0.803
0.131
SIZE
17.631a
1.699
* The difference between FV and NFV mean values is significant at = 0.05, by using t-statistic test.

Appendix 5. Summary Statistics Financially Vulnerable and Not Financially Vulnerable


Sector 87
Status
Number
Variable
Mean
Standard Deviation
FV
127
MARGIN
-0.039
0.087
EQUITY
0.451
0.240
CONCEN
0.754
0.840
SIZE
15.423
1.196
NFV
267
MARGIN
0.019
0.087
EQUITY
0.426
0.280
CONCEN
0.762
0.127
SIZE
15.542
1.143
* The difference between FV and NFV mean values is significant at = 0.05, by using t-statistic test.

Appendix 6. Summary Statistics Financially Vulnerable and Not Financially Vulnerable


Sector 88
Status
Number
Variable
Mean
Standard Deviation
FV
66
MARGIN
-0.030
0.085
EQUITY
0.446
0.342
CONCEN
0.672
0.172
SIZE
14.811
1.813
NFV
137
MARGIN
0.013
0.114
EQUITY
0.371
0.222
CONCEN
0.698
0.188
SIZE
15.019
1.887
* The difference between FV and NFV mean values is significant at = 0.05, by using t-statistic test.

27