The Journal of Socio-Economics 37 (2008) 15541569

Social capital in non-profit organizations:

A multi-disciplinary perspective
Carlos Pestana Barros a, , Francisco Nunes b
a

Instituto de Economia e Gestao, Technical University of Lisbon,

Rua Miguel Lupi 20, 1249-078 Lisbon, Portugal
Instituto Superior de Economia e Gestao, Technical University of Lisbon,
Rua Miguel Lupi 20, 1249-068 Lisbon, Portugal

Received 29 June 2006; received in revised form 19 April 2007; accepted 19 June 2007

Abstract
This paper evaluates human resources of executive managers in Portuguese non-profit organizations
(NPOs), blending two theories of human resources: the human capital model and the social capital model,
based on a questionnaire distributed in Portugal in 2003. We conclude that NPO managers earnings are a
function of both theories. Policy implications are derived.
2007 Elsevier Inc. All rights reserved.
JEL classication: L31; D71; D73; O15
Keywords: Non-profit organizations; Portugal; Human and social capital

1. Introduction
Human resources are a principal factor in the management of non-profit organizations (NPOs).
However, there is a notable absence of research examining human resources in such organizations.
Recent exceptions are Barros and Gomes Santos (2003).
Related to the present research we observe several themes, such as the contribution of social
capital to the quality of life in the workplace (Requena, 2003). The contribution of social capital
to the performance of organizations has been analyzed by Abeysekera (2006). The contribution of
organizational innovations for performance (Mazzanti et al., 2006) and the role of human capital
in poor social contexts (Purkayastha, 2006).

Corresponding author.
E-mail address: cbarros@iseg.utl.pt (C.P. Barros).

1053-5357/$ see front matter 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.socec.2007.06.004

C.P. Barros, F. Nunes / The Journal of Socio-Economics 37 (2008) 15541569

1555

In the present paper, the contribution of human and social capital to the earnings of NPO
managers is analyzed.
Recently, the theory of human capital has been confronted with the social capital theory in
attempting to explain the earnings of specific types of educated individuals. In this paper, we test
these two theories to explain the earnings of NPO managers. The social capital theory asserts that
the market compensates an individuals social ties and networks (Coleman, 1988, 1990; Bourdieu,
1986 and Bourdieu and Wacquant, 1992). Social capital arises because of dense interactions
between social actors who create an intricate web of relational networks around themselves. Such
interactions facilitate the exchange of information and the creation of obligations or expectations,
in addition to the imposition of sanctions on those who fail to meet their obligations. Because
of dense interactions, individuals tend to possess two types of information, one being common
information that enables its validation, thereby possibly enhancing its reliability, the other being
imperfect information, which enables the individual to obtain a competitive advantage. Since
information acquisition and dissemination are time-consuming and costly, it provides significant social capital. This theory has recently come to prominence and is gaining more credence
(Sparrowed et al., 2000) than the human capital theory, which asserted that the market rewards
productivity (Mincer, 1974). The sources of productivity are education, experience and ability.
If the latter theory is correct, individuals should invest in education, whereas if the social capital
theory is right, individuals should place greater value on constructing and expanding social ties
and networks. Almost certainly, both theories explain a part of the reality, but which one is the
more determinant? Since the acquisition of both human and social capital is costly, which should
be the more favoured route upon which to embark by individuals who want to succeed in their
careers? This paper endeavors to provide an answer to this question.
The social capital theory is analyzed and compared with the human capital theory, using data
from a questionnaire distributed in Portugal in 2003. The paper contributes to the literature which
analyses social capital, focusing on NPO managers as a case study.
The paper is organized as follows: Section 2 presents a literature review; Section 3 explains the
theoretical framework; Section 4 presents the contextual setting; Section 5 develops the empirical
study; Section 6 discusses the results; Section 7 presents the contributions, limitations and possible
extensions; and finally, Section 8 concludes.
2. Literature review
Since we are analyzing both the human and social capital theories, the literature survey must
address both of these competing theories. We follow an historic line of exposition, starting with
the older theory (human capital) and moving on to the more recent theory of social capital.
2.1. Human capital theory
According to Mincer (1974), the classic specification of the human capital theory for the
determination of earnings is
log Wi = 0 + 1 Si + 2 EXPi + 3 EXP2i + i

(1)

where log Wi is the log of earnings of individual i, Si a measure of his/her schooling, EXP the
years of experience, i the statistical error term, and are parameters to be estimated, with the
parameter 1 being the return on schooling. Because W is measured in logarithms, the parameter

1556

C.P. Barros, F. Nunes / The Journal of Socio-Economics 37 (2008) 15541569

estimates from this regression can be interpreted as reflecting the proportional change in salary
when the independent variables change.
This model is based in the theoretical contribution of Becker (1993) first published in 1964.
Traditionally, Eq. (1) was estimated by ordinary least squares (OLS), (Vieira, 1999; Hartog et
al., 2001). The estimated i obtained by this approach was not consistent, either because of measurement errors in the schooling variable or because the explanatory variables, such as individual
ability, were correlated with the unobserved disturbances in the equation. Innovative approaches
have been developed in the applied studies of the correlation between schooling and earnings
to deal with this problem (Ashenfelter et al., 1999, p. 452). They fall into three categories: (1)
controls for variables, which may be correlated with schooling and which determine earnings;
(2) comparison of twins or siblings with different schooling levels; and (3) use of instrumental
variables (IV), which affect schooling levels, but do not directly affect earnings.
Among these approaches, the first deals with the issue of ability bias by including explicit
measures that proxy for unobserved ability, such as IQ testing (Grilliches and Mason, 1972). This
approach has been criticised, because it is extremely difficult to develop ability measures that are
not themselves determined by schooling.
The instrumental-variable approach, which is chosen for this paper, employs instruments that
are not correlated with the earnings residual (Card, 1993, 1994). This approach proceeds in
two simultaneous stages: first, it regresses schooling on the instruments chosen and obtains the
predicted values of schooling; it then regresses earnings on the predicted value of schooling,
adopting a 2SLS procedure. By assumption, the instrument is correlated with earnings, because
this influences schooling. The instrument chosen in empirical work may not be truly independent
of earnings residuals.
For example, the instruments used depend on the data available (Angrist and Krueger, 1991).
The latter authors used as schooling instruments the season of birth, finding that, on average, children born earlier in the year have less schooling, because they reach the compulsory school-leaving
age earlier. Levin and Plug (1999), with the same instrument, found the opposite relationship.
Harmon and Walker (1995) used quarter of birth and several dummy variables that were related
to raises of the compulsory school-leaving age. Butcher and Case (1994) found that the gender
composition of a females siblings affected her educational attainment, with females with no sisters receiving, on average, more education. Thus, they used the siblings gender composition as
an instrument.
Other researchers rely on more conventional instruments, such as the respondents social
background. Using Finnish data, Uusitalo (1999) considered schooling by employing dummy
variables for the fathers education and socio-economic status. Blackburn and Neumark (1995)
used instrument variables that were related to the respondents sibling rank, as well as the fact
of living with both parents at the age of 14. Brunello and Miniaci (1999) used as instrumental
variables a set of variables that measure family background, including both the highest educational
level and the fathers and the mothers occupations.
2.2. Social capital theory
The social capital theory (Coleman, 1990) assumes a correlation between social capital and
earnings, since social capital creates value when the relationship among people changes in ways
that facilitate instrumental action. Social capital is defined as the sum of resources that accrue to
an individual by virtue of possessing a durable network of inter-individual relationships (Bourdieu
and Wacquant, 1992, p. 119). It provides a way to characterize an individuals complete set of

C.P. Barros, F. Nunes / The Journal of Socio-Economics 37 (2008) 15541569

1557

relationships and focuses on the access and flow of resources: knowledge, information and other
capital. For a recent survey, see Dasgupta and Serageldin (2003).
At present, there are three formulations of the social capital theory. The first approach is the
weak-tie theory, concentrating on the strength of the social ties used by a person in the process of
finding a job (Granovetter, 1973). In the individual relationship, there are two types of ties, strong
and weak. Strong ties are frequent, emotionally intense ties with friends, advisers and colleagues.
The information possessed by any member of this circle is likely to be quickly shared with the
other members, thus rendering it more common so less valuable and eventually redundant.
Weak ties are infrequent, not emotionally intense and restricted to one narrow type of relationship.
Granovetter (1973) found that weak ties were more likely than strong ties to have been the source
of information on job openings for the sample of job incumbents he interviewed. Subsequent
research drew mixed results (Bridges and Villemez, 1986; McPherson et al., 1992; Seibert et al.,
2001).
The second approach was Burts (1992, 1997) structural hole approach to social capital, based
on a social network of relationships, denoted alters by the author, and the individual, denoted ego
by the author. Burt asserts that a structural hole exists when two alters (network relationship) are
not connected to each other. According to this theory, it is advantageous for an ego (individual) to
be connected to many alters, who are themselves unconnected to the other alters in egos network.
Structural holes provide an individual with three primary benefits: (1) greater bargaining power,
(2) more unique and timely access to information and (3) greater visibility and career opportunities
through the social system. The test of this theory concluded that the structural holes are statistically
significant in explaining the level of social resources (Seibert et al., 2001).
The third approach is the social resources theory (Lin et al., 1981a,b). Social resources are
focused on the nature of the resources embedded within a network. Any alter who possesses
characteristics or controls resources, which are useful for the egos attainment of goals, can be
considered a social resource. Alters who provide career development advice and support are the
relevant social resource when considering an egos pursuit of instrumental career goals. Lin et al.s
research concluded that tie-strength was negatively related to the alters occupational prestige. In
turn, the alters occupational prestige was positively related to the job-secured egos prestige.
Each of the above-mentioned theories claimed to supersede the preceding theory by performing
competitive model-testing. Various authors found that the weak ties seemed to have the stronger
and more robust effect on social resources (Seibert et al., 2001). Moreover, the structural holes
had an independent effect on the level of social resources.
3. Theoretical framework
Previous research has shown that human capital theory and social capital theory play a role
in an individuals social resources. The dependent variable is usually earnings, measured by the
declared monthly wage, including fringe benefits, earned in management activity. The human
capital theory of Becker (1993) and Mincer (1974) is adopted as a theoretical reference of this
paper, alongside the human social capital theory of Coleman (1990) and Bourdieu (1986).
These considerations lead to the following hypotheses:
Hypothesis 1. The greater the NPO managers human capital measured by the education levels,
the higher the earnings.
This is the traditional hypothesis tested in the human capital theory and observed everywhere
(Psacharoupolos, 1985). In order to test this hypothesis, and following the traditional human

1558

C.P. Barros, F. Nunes / The Journal of Socio-Economics 37 (2008) 15541569

capital approach, we consider some of the individuals characteristics as being reliable, theoretical
predictors of his earnings. Among other variables, this could be the case with (i) education
(educ), usually considered the variable most highly correlated with earnings and social status.
This variable was measured by the number of years of education. It could also be the case with
(ii) the individuals number of years worked (worked) to take into account the probability that
NPO managers may enter the field rather late, and (iii) worked squared (worked2 ), as proxies for
experience, following the literature on human capital. Moreover, we included in the model the
age, the traditional measure of experience in human capital models. However, this variable was
highly correlated with worked, introducing multi-colinearity into the model, and for this reason
was deleted from the statistical analysis.
Hypothesis 2. The greater the NPO managers social capital measured by social ties and
networks, the higher the earnings.
This hypothesis, which is based on social capital theory, has already been tested (Seibert et
al., 2001). However, in order to test this second hypothesis, due to its nature, we only consider
objective characteristics. What are the combinations of the measures that actually represent social
capital? We follow the literature review, choosing, inter alia, (iv) the total number of friends and
relatives working in NPOs who have had an impact on the respondents career, as indicated by the
respondent in the questionnaire (Network ties). These social ties generate knowledge spillovers
within the network that eventually translate into value for the individual. This variable is measured
as the sum of the managers NPO ties prior to his/her appointment as an NPO manager. In addition,
(v) there are other ties, such as the number of friends and relatives working in activities other than
NPOs, which similarly can generate knowledge spillovers (Other ties). This variable is measured
as the number of other ties outside of the sector, prior to his/her appointment in the NPO sector. (vi)
High-level ties are similar to the effect of influential godfathers in traditional societies, which
constitute a competitive advantage. The high-level ties are measured as the number of such ties
prior to the appointment as an NPO manager. Godfathers and influential family friends constitute
the high-level ties (High-level ties). (vii) Weak ties are infrequent, not emotionally intense and are
restricted to one narrow type of relationship. They generate knowledge spillovers which are not
shared by the network and therefore, can be the source of a competitive advantage in a market with
imperfect information. Remote relatives who are only occasionally encountered and colleagues
who are not considered to be friends constitute the weak-ties network (Weak ties). (viii) A structural
hole is a term to describe what exists when two relationships are not connected to each other. It is
advantageous for an individual to be connected to many alters, who are themselves unconnected
to the other relationships in the individuals network, in order to generate imperfect information
(Burt, 1997). The number of relationships connected independently of the respondent, as perceived
by the respondent, measures this variable. The variable used in the regression (Structural holes) is
measured as one minus the proportion of declared number of relationships connected. (ix) Career
development advice and support is a resource advantage in the search for employment (Dreher
and Ash, 1990). This is a dummy variable which is equal to one if the respondent had career advice
and zero otherwise (Career development advice).Our empirical strategy consists of estimating a
generalized earnings equation as follows:
ln W = X + Y + XY + i

(2)

In this equation, W is the NPO managers earnings, X the human capital variables, Y the social
capital variables and is a normally distributed error term with zero mean and finite variance.

C.P. Barros, F. Nunes / The Journal of Socio-Economics 37 (2008) 15541569

1559

4. Contextual setting
The education system in Portugal is composed of primary (4 years), secondary (7 years) and
tertiary education (4 or 5 years of college). Compulsory education was just 3 years until as recently
as 1956, when it was extended to 4 years. In 1960, the 4 years of education became compulsory
for girls. In 1964, compulsory education increased to 6 years. In the 1980s, it was extended to 9
years. Based in this legacy the present education system is weak, despite the investment made by
the government in the last years.
It is implicitly assumed that in order to be contracted as a manager in a contemporary Portuguese
private enterprise, a college degree is required, and a Masters degree (MBA) preferred. However,
self-made businessmen and women who manage their own companies do not always have
a degree. The same is expected in the NPO sector, in which a group of individuals organize
themselves in order to set up a social activity. The small dimensions of the Portuguese NPO
establish an analogy between the self-made enterprise and the cooperative form.
Managerial courses which are the basis of NPO managers are considered not at par with the
most developed country. This result is based in the evidence that comes from an exhaustive
survey carried out jointly by Ad-Capita Executive Recruitment and Research and the Cranfield School of Management, UK (see report in pdf: Can Portuguese Managers Compete?
at www.adcapita.com, ACU, 2003). The study highlights areas which are certainly applicable to
the current NPO management activity and lends support to the view that this is a main problem
for national development.
5. Empirical study
The empirical study was carried out by means of a mail questionnaire survey conducted in June
and July 2003 on a sample of NPO managers at their places of work. The universe was made of all
NPO managers obtained in database that was created by the Portuguese Governments Ministry
of Labour of all NPOs. This universe amounted to 625 managers, which is a small number for
statistical purpose, but adequate for a small country. Because of budgetary restrictions and the
limited time available, it was decided to collect data from a stratified, random sample of NPO
managers, amounting to 450 managers, with the central aim of determining their human and
social capital. The sample was first stratified by organization type. The questionnaire contained
two pages, two sections and 24 questions, from which 14 are displayed in Appendix A.
The questionnaires returned totalled 320, from which 252 completed questionnaires were
retained for the present analysis, which represents a response rate of 56% of the sample chosen.
This corresponds to a sampling error of 2.7% with a confidence interval of 95%. The remaining
questionnaires received, but not considered for the present research, were discarded because of
uncompleted fields and incorrectly completed questionnaires.
5.1. Questionnaire
The managers were asked to complete a standard questionnaire, which included questions on
socio-economic and other issues. Table 1 below summarizes the variables used in this paper.
5.1.1. Reliability, validity and generalizability
To ensure the validity and reliability of the data, several steps were taken. First, the point of
departure was a questionnaire already applied in cooperatives (Barros and Gomes Santos, 2003)

1560

C.P. Barros, F. Nunes / The Journal of Socio-Economics 37 (2008) 15541569

Table 1
Characterization of the variables
Variables

Variable description

Log (earnings)

Logarithm of the monthly wage of professional

NPO managers after taxes

22.97

Managers level of education, in years of education

Individuals number of years working as NPO
manager
Individuals squared years worked as NPO manager
Number of friends and relatives in NPO
management who have influenced the respondents
career, indicated by the respondent
Number of friends and relatives in other activities
outside NPO management with influence in the
respondents career, indicated by the respondent
Number of godfathers influential in the
respondents career, indicated by the respondent
Number of infrequent, not emotionally intense ties
of the respondent (remote relatives and colleagues
who are not friends)
One minus the proportion of links among all the
friends, relatives, godfathers and colleagues,
independent of the relationship they have with the
respondent, as perceived by the respondent
Dummy variable which is one if the respondent had
mentoring alter and zero otherwise
Dummy variable which is one if the manager works
in a foundation
Dummy variable which is one if the manager works
in a association

Human capital variables

Education
Experience
Squared experience
Social capital variables
Network ties

Other ties

High-level ties
Weak ties

Structural holes

Career development
advice and support
Foundation
Association
Instrumental variables
Fathers education
Fathers work
Children

Managers fathers level of education, in years of

education
Dummy variable which is one if the father works in
non-profit sector and zero otherwise
Number of children of the manager

Range

Mean

S.D.

2.41

2.285

220
115

16.15
5.205

2.82
4.053

1225

27.992

16.426

05

3.25

2.15

07

2.15

3.45

02

0.82

0.43

110

4.32

1.25

0.20.7

0.42

0.32

01

0.28

01

0.123

01

0.170

410

5.87

01

0.23

05

1.5

3.82

0.45

which was adapted for the present purpose, ensuring that prior research in the field was considered
and face validity established. Second, all relevant literature was taken into consideration. Third,
the questionnaire was pre-tested on students of economics at Instituto Superior de Economia e
Gestao from the Technical University of Lisbon. Following the administration of the final survey,
a stratified random subset of 50 respondents was contacted by phone a second time to check
if any problem persisted, but none were revealed. These procedures ensure the validity of the
questionnaire, meaning that it measures what it was intended to measure. Fourth, the questionnaire
opted for a random sample, with a response rate of 56%, which was considered an acceptable
sample of respondents (Dillman, 1978). This procedure ensures the generalizability of the data,
meaning that the findings are applicable to a more general population. Fifth, the reliability of
the data was examined, analyzing it extensively with alternative methods and reaching the same
conclusions (Barros and Nunes, 2007). The extensive examination of the survey validity, reliability

C.P. Barros, F. Nunes / The Journal of Socio-Economics 37 (2008) 15541569

1561

and generalizability leads to the inference that there is nothing in the evaluation to suggest that it
is either invalid or unreliable.
5.1.2. Testing for non-response
The 56% response rate raises the question of non-response, for which a testing procedure based
on Dillman (1978) was adopted. A first test for this problem involved defining a sub-sample random
choice group of respondent, contacting them by phone again and suggesting testing the answers.
The answers maintained the declared values, ensuring the accuracy of the responses. A second
test involved contacting a random sub-sample of those who had not answered, to understand the
reasons for their non-response. As a result of this, several explanations were given. The first reason
was the individuals declared secrecy policy, which is a common obstacle to questionnaires. The
second reason was a lack of time available to complete the questionnaire during the flight. The
third reason was saturation, associated with completing too many questionnaires. From these three
reasons, it can be asserted that the non-responses have the same characteristics as those who did
respond, establishing the representativeness of the questionnaires that were fully completed.
5.1.3. Data
The general characteristics of these respondents were that they were male (82%), with an
average age of 52. This profile leads to an overall definition of the respondent as male, aged
and middle-class, with a family that includes one child. Other characteristics of the sample are
presented in Table 1. The objective was to evaluate the Portuguese NPO non-profit human and
social capital. To pursue this objective, the questionnaire was structured according non profit
organization types.
5.2. Results
Our empirical strategy consists firstly of estimating the human capital model and the social
capital model with OLS, in order to distinguish between the two models. We verify that the
human capital model has more explanatory power than the social capital model, based on the
R-square and the F-statistics. Second, we estimate the extended model, which blends human and
social capital models together with OLS. However, taking into consideration that according to
the literature, schooling and experience (when measured by age) are potentially endogenous and
measured with error, we re-estimate the extended model with a corrective instrumental variable
(IV) procedure.
Table 2 below shows the estimated results for all three models (human capital model, social
capital model and the extended model that combines human and social variables). Observing the
human capital model, the logarithm of earnings serves as a dependent variable, which is regressed
on the standard controls. We find that this baseline regression (the human capital model) produces
a positive and statistically significant return on schooling, confirming previous research findings that human capital has explanatory power in earnings. The social capital model estimated
by OLS also produces a positive and statistically significant return on social relationship variables, confirming that they also have explanatory power in earnings. However, the social capital
model has less explanatory power. Since we do not know of any published research comparing
both models among NPO managers, we are unable to confirm this result. Finally, adding human
and social capital together in an extended model, the exploratory power increases, confirming
prior research that both human capital and social capital variables have explanatory power in
earnings.

1562

C.P. Barros, F. Nunes / The Journal of Socio-Economics 37 (2008) 15541569

Table 2
Estimates of returns (dependent variable log earnings)
Variables

Constant
Education
Experience
Experience squared
Network ties
Other ties
High-level ties
Weak ties
Structural holes
Career development
Education network
ties
Education weak
ties
Education career
development
Experience network
ties
Foundation
Association
R-Square
Adjusted R-square
F
White het test
BreuschPagan het
test
NOBS

Human capital model

Social capital model

OLS

Human and social

capital model
OLS

Human and social

capital model
IV

OLS
1.002 (0.231)
0.251 (3.128)*
0.389 (3.837)**
0.031 (2.126)*

5.342 (2.982)

0.083 (3.215)
0.528 (2.983)
0.098 (2.754)
0.325 (2.854)
0.358 (3.287)
0.537 (1.028)

1.122 (2.415)*
0.137 (3.245)**
0.321 (3.549)*
0.274 (3.128)*
0.0063 (1.219)
0.316 (3.284)**
0.089 (3.583)**
0.328 (3.472)**
0.145 (2.128)*
0.316 (3.139)**
1.215 (3.127)**

1.136 (3.428)**
0.114 (3.535)**
0.152 (3.219)**
0.142 (3.521)*
0.0043 (2.132)
0.251 (4.321)**
0.421 (3.587)**
0.128 (3.582)**
0.153 (3.127)**
0.227 (3.563)**
1.054 (2.983)**

0.052 (4.218)**

0.051 (3.571)**

1.532 (3.537)**

1.218 (3.023)**

0.732 (4.732)**

0.712 (3.732)**

0.221
0.19
4.328
0.22326
0.6247

0.253
0.20
3.538
1.3147
0.6138

0.128 (3.631)**
0.217 (3.856)**
0.545
0.51
12.568
0.21321
0.8219

0.114 (2.953)**
0.175 (3.121)**
0.637
0.70
15.234
0.27187
1.0254

252

252

252

252

t-Statistics in parentheses are below the parameters, those followed by ** are significant at 1% level and those with * are
significant at 5% level.

Allowing for endogeneity in the equations, we re-estimate the models with instrumental variables. Our instrumental variables are all the variables used in the extended model, together with
three additional instrumental variables to account for schooling and experience. Since the quality
of the proxy variables is important (Madalla, 1992), we looked to previous studies on earnings in
relation to education for guidance as to the instrumental variables for which we opted, as well as
the availability of the data. We chose first, the fathers education (i.e. number of years of education
of the father), assuming this to have had an exogenous influence on the respondents schooling
(Butcher and Case, 1994). Second, if the father worked in the NPO sector, signifying a tradition in
the family (Granovetter, 1973) and third, the number of children of the respondent, assuming this
to have an exogenous influence on the respondents age, since older individuals are more likely
to have children (Brunello and Miniaci, 1999). We did not include instrumental variables for the
human capital variables, since it is not clear from the literature what these instrumental variables
could be.
We considered some test results proposed by Bound et al. (1995) to shed light on the quality
and validity of our instrumental variables. Instrumental quality is ensured if there is a strong
correlation between the instrumental variables and schooling. It is well documented that an IV

C.P. Barros, F. Nunes / The Journal of Socio-Economics 37 (2008) 15541569

1563

procedure using weak instrumental variables may yield more inconsistent point estimates than
those produced by OLS. First, the F-test of joint significance for the respective instrument set is
equal to 0,53. Secondly, instrumental variables are valid provided they affect earnings through
schooling only. We tested whether our instrumental variables had any direct influence on earnings through schooling only with the Sargan test, a test of over-identification restrictions on the
instrumental variables, with an asymptotic X2 distribution and degrees of freedom equal to the
number of over-identifying restrictions. With a value of 4.31 (p-value = 0.53), we cannot reject
the validity of over-identifying restrictions.
The models fit the cross-data well with the extended model, presenting an adjusted R-square
of 54% and an F-statistic of 9.128, denoting that the variables jointly explain the model. The OLS
model, estimated for comparative purposes, is in line with the IV model, but with less explanatory
power (Greene, 2000).
Assuming that the lack of independence appears to be the rule rather than the exception in
survey data, we test heterogeneity with two statistical tests, the BreuschPagan test and Whites
test of heteroscedasticity. Neither of the tests rejects the homoscedasticity in the residuals of all
models.
As expected, earnings are a function of various personal characteristics, configuring human
capital as well as social capital variables. First, relative to human capital variables, we observe
that the higher the education, the higher the earnings, the t-value being statistically significant.
Moreover, we can see that the greater the experience, the higher the earnings, the correlation
being statistically significant. Finally, the greater the square experiences, the higher the earnings,
this coefficient also being statistically significant. The negative sign of this variable, commonly
found in human capital models, signifies that earnings grow until a certain point in time and then
decline.
Second, relative to the social capital theory, we observe that the stronger the network ties, the
higher the earnings, but this coefficient is not statistically significant. Moreover, the stronger the
other ties, the higher the earnings, namely, the higher the high-level ties, the higher the earnings, the
higher the structural holes, the higher the earnings and the higher the career development advice,
the higher the earnings. Finally, relative to the cross-effects between human and social capital,
we have retained only the cross-effects which were statistically significant. It is verified that the
higher the interaction between education and network ties, weak ties and career development, the
higher the earnings. Moreover, the higher the interaction between experience and network ties,
the higher the earnings.
6. Discussion
Our findings point to a significant positive association between earnings and either human
capital or social capital, measured by the variables used in this study, validating the theoretical models of Becker (1993), Mincer (1974), Coleman (1990) and Bourdieu (1986).
These results bring the non-profit sector more into line with the general economic sectors, in
which human capital and social capital are important determinants of earnings. Perhaps these
results stem from the necessity of NPO managers to have formal training in management
schools.
For the interpretation of the parameters in the Mincerian semi-log equation, they are the
proportionate rate of change of earnings for one unit change in the explanatory variables, Greene
(2000). Therefore for the variable years of education, the coefficient is interpreted as the rate
of return for an additional year of education. In the same way, the network ties parameter is

1564

C.P. Barros, F. Nunes / The Journal of Socio-Economics 37 (2008) 15541569

interpreted as the rate of return for an additional friend or relative influent in the CEO career.
For the binary variables, the coefficient is interpreted as the rate of return when the variable is
active.
The return on education for Portuguese non-profit NPO is 25.1% for an additional year of
education when the human capital models are considered, but it decreases to 13.7% whenever the
human and social capital is combined. These results are lower than the average ones observed
the Portuguese economy, Vieira (1999), where an average rate of return of 12% compares with
13.7% in the non-profit organizations. However, when doing this comparison it most been taken
into consideration that the average value obtained for the Portuguese economy is defined for
all labourers, including those with a university degree, and those without a degree. Therefore, a
higher value for managers was expected for a sample that includes only labourers with a university
degree.
However, taking into account endogeneity and including the human capital the return on
education drops to 11%, a result in line with previously published results for Portugal, Vieira
(1999).
Concluding, it can be asserted that the return on education for non-profit managers is slightly
higher than the average value for the average population according to OLS results. Several reasons
can be advanced for such higher return in non-profit organizations. First, non-profit managers
have a university degree and traditional it attracts higher salaries. Therefore, there is some market
rule for university degree that is independent from sector of activity. Second, this sector has
recently attracted public subsidies, which contributed to its profitability. This subsidy policy
corresponds to a trend towards equity in the country. The return on social capital is also high,
namely for high-level ties, signifying that a social tie represents a capital in the Portuguese
non-profit organizations.
What are the implications of these results? First, the NPO manager who is well educated and
is capable, conscientious and effective in his work is more highly rewarded than those who do
not fulfil these criteria. Second, experience plays a positive role in the definition of earnings.
Third, social ties play a major role in the earnings function. Clearly, the situation configures a
joint determination of earnings by human capital theory and social capital theory. What are the
consequences of this result? These findings signify that aspiring NPO managers primarily need
to concentrate on their education and should acquire MBA qualifications, but must not neglect
their cultivation of appropriate, positive social ties. Therefore, it is assumed that that there exists
an order that starts in human capital and finishes in social capital. Without human capital, an
individual appears much less likely to reap the social capital rewards. As NPO management is
usually a profession with intense social ties, we conclude that the professionals who opt for this
activity already have a high level of social ties prior to their involvement in the activity. This
situation can raise the possibility of reverse causation, with social ties inducing more income
than the reverse, but a Hausman test performed on the social variables allows us to discard this
hypothesis for this data set. Moreover, since human capital is more important than social capital,
we should expect to observe a growing market for NPO management education. However, between
two individuals with the same education level, the one who will be selected by an organization is
the one with a degree from one of the leading academic institutions, with the best reputation and
references. These characteristics are positively related with social capital. Therefore, the research
draws the attention of the teachers of NPO management trainers to the need to teach those skills
which enhance social capital, in order to permit their students to interact effectively during faceto-face encounters with employers and clients. Baron and Markman (2000) have identified these
skills as: (i) social perception (the ability to perceive others accurately, including perceptions

C.P. Barros, F. Nunes / The Journal of Socio-Economics 37 (2008) 15541569

1565

of others motives, traits and intentions); (ii) impression-management (proficiency in a wide

range of techniques for producing positive reactions among others, increasing efforts to enhance
ones own image, through agreeing with others, complimenting them and giving them small gifts
during an initial meeting); (iii) persuasion and influencing skills (for changing the attitudes or
behaviour of others in desired directions); and (iv) social adaptability (the ability to adjust to a wide
range of social situations and to feel comfortable with individuals from diverse backgrounds).
Therefore, the teaching of at least some of these skills on NPO management courses should be
considered.
How beneficial is the human capital to non-profit organizations? Should it be stimulated or
avoided? There is no clear answer to this question, since human relationship is unavoidable.
Moreover, friendship is related to personal loyalty spread and builds community by constructing
a solid feeling of recognition and belonging to the group, McGehee and Santos (2004). However,
friendship runs against meritocracy (Arrow et al., 2000). In a non-profit organization the most
relevant people is the one who contributes most to the performance. This criterion emphasises the
role of merit in the management of the organization (Cawley et al., 1999). If the organizational
incentive is to be social and sympathetic then the meritocracy is relegated and the performance
suffers. Recent evidence has emerged to confirm the prevailing perception amongst Portugalbased business managers from overseas that incompetence and inefficiency are rife among their
Portuguese counterparts. This evidence comes from an exhaustive survey carried out jointly by
Ad-Capita Executive Recruitment and Research and the Cranfield School of Management, UK
(see report in pdf: Can Portuguese Managers Compete? at www.adcapita.com). The study
highlights areas which are certainly applicable to the current Portuguese non-profit organizations.
A possible cause of this perception is based in the densification of networks observed in the
Portuguese enterprises which runs against meritocracy (Breen and Goldthorpe, 1999). Based
on this argumentation the social capital should be avoided or at least subordinated to the merit
(Wilkins and Ouchi, 1983).
How is the present research comparable with Mincerian models estimated for the Portuguese economy? Table 3 presents the results of the estimated models for the Portuguese
Table 3
Return on education in Portugal
Author

Period

Data source

Method

Marginal returns

R2

Pscharopoulos

1977

Quadros de Pessoal from ME

OLS

15%

Vieira

1986 and 1999

Quadros de Pessoal from ME

OLS
IV

7.5% men, 8.4%

women
1986-OLS-7.4%
1986-IV-3.03
1992-OLS-7.79%
1992-IV-1.5%
Pooled-1986 and
1992-OLS-7.79%
Pooled-1986 and
1992-IV-1.5%
Standard Mincerian
model-OLS-15%
Standard Mincerian
equation-IV-12.3%
Generalized Mincerian
model-IV-10.8%

Barros and Gomes

Santos (2003)

1999

Questionnaire data on
Cooperative managers

OLS
IV

7%
7%
8%
8%
8.6%
8.5%
0.20
0.55
0.55

1566

C.P. Barros, F. Nunes / The Journal of Socio-Economics 37 (2008) 15541569

economy. It can be verified that the rate of return on education for NPO managers is higher
than the average value obtained in almost all the papers cited. This is explained by the sample
data adopted and is similar to results for homogenous population such as cooperative managers.
7. Contribution, limitations and extensions
At this juncture, it is appropriate to consider the contribution of the paper to NPO management,
as well as its limitations and possible extensions. The key contribution of this paper to the literature
is in its comparison of the human and social capital theories in relation to NPO management.
Moreover, we present an extended literature survey which clarifies the relative contribution of the
present paper.
With regard to the limitations, the first limitation is in relation to the data set and the second, to
the method. With reference to the data set, the questionnaire is somewhat short, which affects the
reliability of the significance tests. However, it can always be claimed that the data set is short,
and therefore, a quality analysis equally could not be carried out. On the basis of the short data set,
the conclusions are limited. In order for the conclusions to be generalized, we would need to have
a larger panel data set. Moreover, it is recognized that time series analysis or panel data is much
richer, allowing causality tests, which cannot be performed with cross-section data. Furthermore,
the instrumental variables for social capital variables were not used, since it is not clear what
they could be. Finally, it is not clear what is the source of exogenous variation embedded in the
instruments used, it being supported by the data available (Angrist and Krueger, 1991) and in
previous research.
The limitation of the method is due to the fact that the questionnaire was conducted only on
Portuguese NPOs, thus not allowing comparison with managers of other types of organization
(i.e. private and public enterprises).
A variety of extensions to this paper can be undertaken. First, annual questionnaires to NPO
managers would permit the construction of panel data series. This would provide a richer data
set. Second, comparison with other case studies on NPO management could permit us to base
conclusions on the results of the paper with more accuracy.
8. Conclusion
This paper has addressed the role of education and social ties in the earnings of NPO managers,
using a questionnaire on a sample of managers of Portuguese NPOs. However, it is also clarified
that human capital is more important than social capital, establishing a ranking order that goes
from human capital to social capital. The conclusion from our results is that NPO managers
must first acquire the appropriate education and then establish social ties in order to maximise
their earnings. The teaching of social skills is also required in order to enhance these managers
earnings. Finally, on the basis of the remarks made earlier in Section 6, more research is needed
to confirm these results.
Acknowledgements
The authors thank two reviewers and the editor for their comments, which contributed to
clarifying and improving the paper.

C.P. Barros, F. Nunes / The Journal of Socio-Economics 37 (2008) 15541569

1567

Appendix A. Questionnaire contains 14 of the 24 questions relative to human and social

capital questions
The questionnaire was designed to support the research question:
Scale items for collecting the data
Human capital variables
(i) What are your monthly earnings after taxes (value in euros)?
(ii) What is your level of education (primary, secondary, university)?
(iii) How many years of education correspond to the level of education declared? (Years of education)
(iv) How many years have you worked as a non-profit organization manager, both in your present and any other
NPO? (Years of working as NPO manager)
(v) How old are you? (Years)
Instrumental variables
(i) What was your fathers level of education (primary, secondary, university)? [The level of education was
translated into years]
(ii) Has your father worked in the non-profit sector? Yes/no.
(iii) Do you have any children? How many?
Social capital variables
(i) How many of your friends and relatives involved in the non-profit sector have been, or are, important to you
professionally because they have supported your entry into this area of activity?
(ii) How many of your friends and relatives not involved in NPOs have been, or are, important to you
professionally because they have supported your entry into this area?
(iii) How many people of influence (godfathers) do you have in your social network who have used their
influence to further your professional career?
(iv) How many remote relatives or colleagues, who are not friends, do you have who have been, or are,
important for your professional advancement?
(v) How many relationships exist between the above-listed individuals (relatives and friends in NPOs, relatives
and friends outside NPOs, godfathers, remote relatives and colleagues)?
(vi) Did you have career development support in your profession? Yes/no.

Further details on the variables collected are available from the authors.
References
Abeysekera, I., 2006. Managing human capital in a privately-owned public hotel chain. International Journal of Hospitality
Management 25 (4), 586601.
AdCapita and Cranfield University, 2003. Can Portuguese Managers Compete? Mimeo. Available at www.adcapita.com.
Angrist, J.D., Krueger, A.B., 1991. Does compulsory school attendance affect schooling and earnings? Quarterly Journal
of Economics 106, 9791014.
Arrow, K., Bowles, S., Vytlacil, E. (Eds.), 2000. Meritocracy and Economic Inequality. Princeton University Press, NJ.
Ashenfelter, O., Harmon, C., Oosterbeek, H., 1999. A review of estimates of the schooling/earnings relationship, with
tests for publication bias. Labour Economics 6, 453470.
Baron, R.A., Markman, G.D., 2000. Beyond social capital: how social skills can enhance an entrepreneurs success. The
Academy of Management Executive 14, 106116.
Barros, C.P., Gomes Santos, J.C., 2003. Earnings and schooling of cooperative managers. Annals of Public and Cooperative
Economics 74 (3), 349364.
Barros, C.P., Nunes, F., 2007. Governance and CEO pay and Performance in Non-Profit Organisations. International
Journal of Social Economics 34, 811827.
Becker, G., 1993. Human Capital: A Theoretical and Empirical Analysis, with Special Reference to Education. University
of Chicago Press.
Blackburn, M., Neumark, D., 1995. Are OLS estimates of the return on schooling biased downward? Another look.
Review of Economics and Statistics 77, 217229.
Bound, J., Jaegger, D., Baker, R., 1995. Problems with instrumental variables estimation when the correlation between
the instruments and the endogenous explanatory variable is weak. Journal of American Statistical Association 90,
443450.

1568

C.P. Barros, F. Nunes / The Journal of Socio-Economics 37 (2008) 15541569

Bourdieu, P., 1986. Forms of capital. In: Richardson, J.G. (Ed.), Handbook of Theory and Research in Sociology of
Education. Greenwood, NY, pp. 241258.
Bourdieu, P., Wacquant, L., 1992. An Invitation to Reflexive Sociology. University of Chicago Press, Chicago, IL.
Breen, R., Goldthorpe, J., 1999. Class inequality and meritocracy: a critique of saunders and an alternative analysis. British
Journal of Sociology vol. 50 (1), 127.
Bridges, W.P., Villemez, W.J., 1986. Informal hiring and income in the labor market. American Sociological Review 51,
482574.
Brunello, G., Miniaci, R., 1999. The economic returns to schooling for Italian men: an evaluation based on instrumental
variables. Labour Economics 6, 509519.
Burt, R.S., 1992. Structural Holes: The Social Structure of Competition. Harvard University Press, Cambridge, MA.
Burt, R.S., 1997. The contingent value of social capital. Administrative Science Quarterly 42, 339
365.
Butcher, K.F., Case, A., 1994. The effects of sibling sex composition on womens education and earnings. Quarterly
Journal of Economics 109, 531563.
Card, D., 1993, Using geographic variation in college proximity to estimate the return to schooling. In: NBER Working
Paper no. w 4483. National Bureau of Economic Research, Cambridge, MA.
Card, D., 1994, Earnings, schooling and ability revisited. In: NBER Working Papers no. w 4832. National Bureau of
Economic Research, Cambridge, MA.
Cawley, J., Heckman, J., Vytlacil, E., 1999. Meritocracy in America: wages with and across occupations. Industrial
Relations 38 (3), 250296.
Coleman, J.S., 1988. Social capital in the creation of human capital. American Journal of Sociology 94, S95
S120.
Coleman, J.S., 1990. Foundation of Social Theory. Harvard Business School Press, Cambridge, MA.
Dasgupta, P., Serageldin, I., 2003. Social Capital: A multifacted Approach. Oxford University Press.
Dillman, D., 1978. In: Dillman, D. (Ed.), Mail and Telephone Surveys: The Total Design Method. Wiley, New York.
Dreher, G.F., Ash, R.A., 1990. A comparative study of mentoring among men and women in administrative, professional,
and technical positions. Journal of Applied Psycology 75, 539546.
Granovetter, M.S., 1973. The strength of weak ties. American Journal of Sociology 78, 13601380.
Greene, W.H., 2000. Econometric Analysis, 4th ed. Prentice Hall, Upper Saddle River, NJ.
Grilliches, Z., Mason, W., 1972. Education, income and ability. Journal of Political Economy 80, 74103.
Harmon, C., Walker, I., 1995. Estimates of the economic return on schooling for the United Kingdom. American Economic
Review 85, 12781286.
Hartog, J., Pereira, P.T., Vieira, J.A.C., 2001. Changing returns to education in Portugal during the 1980s and early 1990s:
OLS and quantile regression estimators. Applied Economics 33, 10211037.
Levin, J., Plug, E.J.S., 1999. Instrumenting education and the return to schooling in the Netherlands. Labour Economics
6, 521534.
Lin, N., Ensel, W.M., Vaughn, J.C., 1981a. Social resources and occupational status attainment. Social Forces 59,
11631181.
Lin, N., Ensel, W.M., Vaughan, J.C., 1981b. Social resources and strength of ties. American Sociological Review 46,
393405.
Madalla, G.S., 1992. Introduction to Econometrics. Macmillan Press, NY.
Mazzanti, M., Pini, P., Tortia, E., 2006. Organisational innovations, human resources and firm performance: the Emilia
Romagna Food Sector. Journal of Socio-Economics 35 (1), 123141.
McGehee, N.G., Santos, C.A., 2004. Social change, discourse and volunteer tourism. Annals of Tourism Research 32 (3),
760779.
McPherson, J.M., Popielarz, P.A., Drobnic, S., 1992. Social networks and organizational dynamics. American Sociological
Review 57, 153170.
Mincer, J., 1974. Schooling, Experience and Earnings. NBER, New York.
Psacharoupolos, G., 1985. Return on education: a further international update and implication. Journal of human Resources
20, 583604.
Purkayastha, D., 2006. Norms of reciprocity and human capital formation in poor patriarchal household. Journal of
Socio-Economics 35 (1), 7282.
Requena, F., 2003. Social capital, satisfaction and quality of life in the workplace. Social Indicators Research 61, 331
360.
Seibert, S.E., Kraimer, M.L., Liden, R.C., 2001. A social capital theory of career success. Academy Management Journal
44, 237291.

C.P. Barros, F. Nunes / The Journal of Socio-Economics 37 (2008) 15541569

1569

Sparrowed, R.T., Linden, R.C., Wayne, S.J., Kraemer, M.C., 2000. Social exchange processes, social structure and
performance. Academy of Management Journal 44, 316325.
Uusitalo, R., 1999. Return to education in Finland. Labour Economics 6, 569580.
Vieira, J.A.C., 1999. Return on education in Portugal. Labour Economics 6, 569580.
Wilkins, A., Ouchi, W., 1983. Efficient cultures: exploring the relationship between culture and organizational performance. Administrative Science Quarterly 28, 468481.