1

Kofinti, R. E., Koomson, I., & Peprah, J. A. (2023).

Borrower discouragement and multidimensional child
deprivation in Ghana. The Journal of Economic
Inequality, 1-19.
https://link.springer.com/article/10.1007/s10888-023-
09578-6

*Corresponding Author
(Scopus/Q1/ Social Sciences Citation Index)
The Journal of Economic Inequality
https://doi.org/10.1007/s10888-023-09578-6

Borrower discouragement and multidimensional child

deprivation in Ghana

Raymond Elikplim Kofinti1 · Isaac Koomson2,3 · James Atta Peprah4

Received: 4 January 2022 / Accepted: 16 March 2023

© The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2023

Abstract
With increasing scholarly attention on child deprivation to understand its drivers and poten-
tial policies needed for its alleviation, the discouraged borrower syndrome has received lit-
tle attention despite its potential role in stifling household resources needed to cater for the
needs of children. Using the seventh round of the Ghana Living Standards Survey data,
this study examines the effect of borrower discouragement on multidimensional child dep-
rivation. Endogeneity associated with borrower discouragement is instrumented with the
number of neighbours that are discouraged. We found that the share of children who are
multidimensionally deprived in Ghana is 40.8 percent. Our endogeneity-corrected esti-
mates show that borrower discouragement is associated with 4.0 percentage point increase
in multidimensional child deprivation. This outcome is consistent across different quasi-
experimental methods, and alternative cut-offs used in identifying the multidimensionally
deprived child. The results also suggest that the effect of borrower discouragement on child
deprivation is more pronounced among rural-located children (in general) and girls (in par-
ticular). We identify inability of non-farm business start-ups and reduction in household
per-capita income as potential channels through which borrower discouragement affects
multidimensional child deprivation.

Keywords Child deprivation · Discouraged borrowers · Ghana

* Raymond Elikplim Kofinti

rkofinti@ucc.edu.gh
Isaac Koomson
i.koomson@uq.edu.au; koomsonisaac@gmail.com
James Atta Peprah
jpeprah@ucc.edu.gh
1
Department of Data Science and Economic Policy, School of Economics, University of Cape
Coast, Cape Coast, Ghana
2
Centre for the Business and Economics of Health, The University of Queensland, St Lucia, QLD,
Australia
3
Network for Socioeconomic Research and Advancement (NESRA), Accra, Ghana
4
Department of Applied Economics, School of Economics, University of Cape Coast, Cape Coast,
Ghana

13
Vol.:(0123456789)
 R. E. Kofinti et al.

1 Introduction

Approximately one out of three children in developing countries was facing multiple dep-
rivations in early 2000s and more than 80% of children were experiencing deprivations in
either shelter or sanitation (Gordon et al. 2003). Post 2008 estimates indicate that about
300 million children in sub-Saharan Africa are multidimensionally deprived in two to five
dimensions crucial for their survival and development (de Milliano and Plavgo 2018). Dep-
rivations exert negative effects on children by inhibiting their development in early child-
hood, constraining their participation in society, and limiting children’s access to proper
education, healthcare, and other basic services (Bima et al. 2017; Gordon et al. 2003;
Kofinti and Annim 2016; Minujin 2011; Minujin and Delamonica 2012). The ramifica-
tions of deprivations in early childhood are dire as they are often interrelated and multi-
faceted (Ortiz et al. 2012) and are most likely to perpetuate the cycle of poverty (Duncan
et al. 1998; Oshio et al. 2010; Ratcliffe and McKernan 2010). Not surprisingly, global and
scholarly attention to alleviate deprivations among children, by identifying its drivers, has
assumed prominence since early 2000s. This attention is evident in the prominence given
to child deprivation in the sustainable development goals (SDGs) and the globally con-
certed efforts being made to reduce child deprivation in its multidimensional forms (BAP-
PENAS and UNICEF 2017; Dornan 2017; Koomson et al. 2023; Pemberton et al. 2007,
2012; Sanfilippo et al. 2012; UNICEF and Global Coalition against Child Poverty 2017).
Among the many factors identified as drivers of multidimensional child deprivation,
household income remains one of the most cited (Agyire-Tettey et al. 2021; Fonta et al.
2020; Koomson et al. 2023). A recent study by Agyire-Tettey et al. (2021) shows that liv-
ing standards is the main contributor of multidimensional deprivation in early childhood in
Ghana, of which income is one of the key factors. Household income provides the basis of
physical conditions which have implications for health, physical, emotional, and cognitive
development of children. Equally, these conditions indicate the living standards of chil-
dren as well as monetary deprivation of households (Agyire-Tettey et al. 2021). The core
argument of this paper is that making finance accessible to households has a long-term
positive impact on the welfare of children thus, enhancing the human capital of a country.
Access to credit, and financial inclusion in general, has been identified as having the poten-
tial to address most of the wellbeing indicators of households and children (Koomson et al.
2023). The benefits from access to credit can only be realised by children whose parents
apply for and are granted the loans/credit.
Well-being is not produced in a vacuum. It is financed with financial resources such as
taking money from the bank. In this regard well-being or welfare is assumed to be derived
from income and consumption (Kingdon and Knight 2004). According to the United
Nations (1989), children thrive when their survival, developmental and protection needs
are provided. However, for poor households that find it difficult to borrow or are refused
loan facilities, it means no investment and limited income for the provision of these needs.
The limited income also means low consumption of goods and services that would trans-
late into the non-monetary welfare indicators. The argument is that being denied a loan
hinders the ability and resourcefulness to undertake productive investment which can influ-
ence the wellbeing of household members, especially children.
While there is a consensus that access to credit has a positive impact on house-
hold welfare, there is a paucity of studies that have evaluated the effect of credit con-
straints on household members and that of children, in particular (Djoumessi et al.
2018). In a treatment effect study, Peprah (2018) showed that non-discouraged clients

13
Borrower discouragement and multidimensional child deprivation in Ghana

of micro-credit were able to send their children to school on regular basis compared to
those who did not receive credit, and clients who had received loans had their children
being healthier.
Despite the welfare gains from credit access, there is ample evidence of ‘poten-
tially good’ borrowers who choose not to apply for bank loans because they feel their
application will be rejected and are therefore called discouraged borrowers (Kon and
Storey 2003). According to Kon and Storey (2003), high borrowing cost from for-
mal financial institutions is a primary source of the discouraged borrower syndrome.
They may therefore lack capital to start a business or expand their businesses thus
affecting their income generating capacity. This presumes that discouraged borrow-
ers may not be able to contribute to the improvement of their children’s welfare in
terms of education, health, and general living conditions (Ansong et al. 2021; Koom-
son and Afoakwah 2022). The mechanism can also be seen from the perspective that
when households are discouraged to borrow, investments in health and education of
their children will not happen even when, from a purely financial perspective, they
would be profitable. However, in case the family is not discouraged in terms of bor-
rowing, they will have access to credit to invest in businesses, improve household
income and family well-being (see Ansong et al. 2021; Koomson and Afoakwah
2022; Swamy 2014). Financial resources from such lucrative ventures become reli-
able source of funds for investment in children’s education, health, and the provision
of domestic amenities that promote quality standard of living for children and other
members of the household.
There is myriad of empirical papers that have focused on the effect of borrower
discouragement on several firm outcomes (see e.g., Chakravarty and Xiang 2013; Freel
et al. 2012; Kon and Storey 2003) but its implication on household welfare outcomes
have received little scholarly attention. Due to this, we do not yet know much about the
channels through which borrower discouragement affect household welfare outcomes
in general and that of children in particular. However, it is important to emphasize that
understanding the implications of borrower discouragement on welfare outcomes will
influence the needed policy design required to decrease its prevalence.
We contribute to the literature by being the first to provide empirical evidence of
how borrower discouragement affects children’s welfare outcomes. Specifically, we
examine the link between borrower discouragement and multidimensional child dep-
rivation from a developing country setting, using a comprehensive nationally repre-
sentative data. We further explore the mediating role of non-farm business venturing
as a potential channel through which borrower discouragement transmits to child dep-
rivation. The study also adds to the literature by addressing the endogeneity problem
associated with borrower discouragement and ensuring consistency in findings through
the application of alternative quasi-experimental methods. We find that children in dis-
couraged borrower households are more likely to be deprived in terms of education,
health, and general living conditions. Our findings are relevant for child development
policies that support households with deprived children. Moreover, our findings will
contribute to the inclusive finance agenda by shedding light on the child welfare impli-
cations of credit constraint.
The remaining sections of the paper are arranged as follows: the next section exam-
ines the conceptual link between borrower discouragement and child deprivation; sec-
tion three discusses data and methodology; section four presents the results and discus-
sion, and the final section concludes the paper with some policy recommendations.

13
 R. E. Kofinti et al.

2 Conceptual link between borrower discouragement and child

deprivation

Credit constraint has been empirically found to have a negative influence on rural
household welfare. For example, Baiyegunhi and Fraser (2010) examined the impact of
credit constraint on household welfare and found that unconstrained households have a
higher consumption expenditure than constrained homes. Similarly, Tran et al. (2014)
found that households with increased access to credit can improve their consumption
pattern and thus improve the welfare of their members. Rui and Xi (2010) found in rural
China that welfare losses are larger if the household’s credit request is rejected, and
this exerts a negative impact on food consumption per day and the most affected people
being children in the household. In this section we explore how borrower discourage-
ment translates into child deprivation through income, entrepreneurship, and growth of
family businesses.

2.1 Income

Theoretically, considering access to finance as an input into the household production

function means that lack of finance to undertake productive activity will reduce output
which will in turn affect the level of household income, especially those in the informal
sector. This is based on the premise that the final objective of borrowing for investment
is to generate income for expending on household food, health, clothing, and educa-
tion. Extending the effect to cover household members means that some members of
the household (especially children) may be deprived if credit is not accessed. In line
with Honohan and King (2012), the use of formal banking services is associated with
an increase in individual monthly income, thus, discouraged borrower households, may
experience lower income levels, with its attendant deprivations on household members.
The level of deprivation may be severe for younger children than adult children. Chil-
dren, depending on their age, have special needs concerning nutrition, care, and educa-
tion. In most cases malnutrition, diseases and infections have more adverse and poten-
tially severe consequences for children under 5 years of age compared to children of
6–17 years. The conditions of such children may be worsened in situations where there
is lack of parental capacity to generate income to cater for their basic needs.

2.2 Entrepreneurship

The remediation perspective of poverty-entrepreneurship nexus argues that entrepre-

neurship offers resources (such as skills, training, finance) that enable markets to flour-
ish so that households can make use of these resources to lift themselves out of poverty
for the benefit of household members and the larger economy. It has been well estab-
lished that access to finance is key for complementing entrepreneurial trait to boost self-
employment (Peprah et al. 2015; Peprah and Koomson 2015). It is important to note
that the presence of discouraged borrowers could be evidence of a market imperfection;
however, informal turndowns represent an efficient mechanism in SME debt markets
explaining a type of borrower discouragement. Households that are successful in their

13
Borrower discouragement and multidimensional child deprivation in Ghana

entrepreneurial drive are more likely to generate income to cater for their dependents.
Thus, in an enterprising household, children are less likely to be deprived.

2.3 Growth of family business

The concept of family business is not new in Ghana. Like all other companies, family
businesses in Ghana need finance. Unfortunately, family loan applications are most often
rejected thus discouraging people with business ideas to venture and expand. The main
sources of capital for small and growing number of enterprises are retained earnings,
investments from own savings (Gentry and Hubbard 2004) and assistance from family and
friends. While majority of family businesses use family loans to finance their operations,
others use private equity. However, once these sources are fully exhausted or are lacking,
they face the challenge of sourcing funds from traditional banks, and this is the point bor-
rower discouragement begins to manifest because they might not qualify in terms of lack of
collateral or improper business documentations. This phenomenon may negatively impact
growth of family businesses and in turn affect the welfare of household members, espe-
cially children who are the most dependent on household resources. Thus, a well-func-
tioning financial system that encourages borrowing of financial resources may contribute
significantly to family business outcomes in the short-term as well as welfare-enhancing
effects on household members in the long-term.

2.4 Living standards and education

Access to credit induces households’ ability to spend on household durable consumption.

An example of this is home finance loans which expedites households’ transition towards
home ownership (Akuffo 2006). Currently, nine commercial banks in Ghana officially offer
three forms of mortgage loans, namely conventional mortgage loans, home improvement
and home completion loans (Prime-Stat 2018). Home ownership and its related ancillar-
ies are directly linked to the living standards dimension of the child deprivation construct.
Opportunity International (2022), also points to child education loans which aid parents
to access the resources they need to enhance human capital development of their chil-
dren. Additionally, affordable private schools also access child education loans to expand
their infrastructure base to engender enrolment. This shows that any constraint in access
to credit, such as borrower discouragement can have a direct negative effect on child
deprivation.

Decrease Living condition/education/health Increase

Income
Entrepreneurship/family business

Borrower Discouragement Increase Child Deprivation

Fig. 1  Conceptual link between borrower discouragement, income, entrepreneurship, and child deprivation
(Source: Authors’ Construct)

13
 R. E. Kofinti et al.

Integrating the narratives above within the context of the theoretical and empirical liter-
ature, Fig. 1 summarises and presents the link between borrower discouragement, income,
entrepreneurship, and child deprivation.

3 Data and variables

The data used for this study is sourced from the seventh round of the Ghana Living Stand-
ards Survey (GLSS7) (GSS 2019) which were collected in 2016/17 by the Ghana Statistical
Service (GSS) across the then 10 (but now 16) regions of Ghana. The survey used a prob-
ability sampling approach (two-stage), and included sections on demography, housing con-
ditions, energy and fuel use, employment, education, water and sanitation, health, access
to financial and insurance services, remittance and household assets, disability, migration,
agriculture, non-farm activities and governance, among others. The rationale for choosing
the GLSS7 is that it is the only wave of the survey which includes anthropometric meas-
ures which are required for the health dimension of the multidimensional child deprivation
measure. The GLSS7 covered 15,000 households with response rates being 93.4% leading
to a final sample size of 14,009 households. After merging the files/sections containing our
variables of interest, we had a working sample of 8,996 households with 24,524 children
aged 0 to 17 years. After adjusting for missing data in the regression analysis, the model
with the most observations comprised 14,631 children.

3.1 Multidimensional child deprivation

Our measure of multidimensional child deprivation draws from the multidimensional pov-
erty measure from the Oxford Poverty and Human Development Initiative (Alkire and Foster
2007) grounded on the relative deprivation theory and capabilities concepts (Gordon 2006;
Sen 1976; Townsend 1979). The method consists of two steps: identification and aggregation.
For the purposes of our study, we focused on the identification step to identify children expe-
riencing multidimensional deprivations in Ghana. The identification step consists of a dual
cut-off approach. First, the deprivation cut-off identifies if the child is deprived or not in each
indicator. Second, the poverty cut-off k identifies if the child is multidimensionally deprived
by assessing their vector of weighted deprivations. Our study focused on three broad equally
weighted dimensions (living standards, education and health) represented by twelve indicators
(see Table A1) with well-defined deprivation cut-offs and summary of the deprivation indica-
tors across age groups (see Table A2). At the second stage, we identified children experiencing
multidimensional deprivations with a conventional poverty cut-off of 0.333.
Mathematically, Eq. (1) is the deprivation profile of children with respective deprivation
cut-offs.
𝛼i = wi l1 + w2 l2 ⋯ + wn ln (1)
where 𝛼i is the child deprivation scores, li = 1 if the child is deprived in indicator i and
li = 0 if otherwise, wi is the weight attached to indicator i with di wi = 1. The second cut-
∑
off of 0.333 is applied to the weighted deprivation scores to identify children who are mul-
tidimensional deprived. It is worth stating that our measurement of multidimensional child
deprivation incorporated a life-cycle approach in the selection of dimensions and indicators

13
Borrower discouragement and multidimensional child deprivation in Ghana

for children under 5 years and those between 6 to 17 years to capture the different depriva-
tions children experience at different stages of their life (see Table A1).

3.2 Borrower discouragement

Following the conceptualisation and measurement of borrower discouragement in the

existing literature, we define and identify discouraged borrowers as "potentially good" bor-
rowers who opt not to apply for formal bank loans owing to fear of rejection and other
specified factors (Chakravarty and Xiang 2013; Kon and Storey 2003). In this study, we
use the GLSS7 question that asked respondents: “what was the main reason for not trying
to obtain a loan?”. Responses to the question are 1 “No need”; 2 “Interest rate too high”;
3 Demand for collateral”; 4 “Already has too much debt”; 5 “Cannot obtain the amount
needed” and 6 “Other”. To produce a binary borrower discouragement variable, responses
such as “interest rate too high”; “demand for collateral” and “cannot obtain the amount
needed” and “Other” were coded as 1 to depict discouraged borrowers, and 0 if otherwise.
Summary statistics for all variables included in the analysis are presented in Table A3.

4 Estimation technique

We estimate a linear probability model (LPM) in which discouraged borrower influences

multidimensional child deprivation while controlling for other covariates at the child,
household, and regional levels. Since child deprivation and borrower discouragement are
both binary variables, the LPM remains an appropriate model because it is not prone to
the potential biases that are associated with estimates of continuous variables when used
in an LPM. In line with existing studies, we estimate Eq. (1) using the Linear Probability
Model (LPM) because it lends the coefficients to easy interpretation as marginal effects
(Afoakwah et al. 2020a, b; Caudill 1988). Moreover, the coefficients of the LPM directly
align with those of the two-stage least squares (2SLS) which is the remedy employed for
the endogeneity problem. The empirical equation is specified as Eq. (2):
′
Childepih = 𝛼 + 𝛽DBh + Γf emalechdih + 𝛀HHh + 𝝋i + 𝛖h + 𝜖i (2)

where Childepih represents multidimensional child deprivation status of child i in

household h, DB denotes discouraged borrower status of the household (i.e., the head),
f emalechdih is a binary variable for gender of child, where 1 represents female and 0 is
male. HHi is a vector of household and household head characteristics identified in extant
studies as being determinants of multidimensional child deprivation which include age of
household head, female-headed home, educated head, account ownership by head, rural
household, married, separated/divorced/widowed, and household size. 𝝋 captures Child
year-of-birth cohort fixed effects while 𝛖 represents the regional fixed effects.

4.1 Potential endogeneity

We acknowledge that the estimates are likely to be endogenous. The source of endogeneity
could be traced to the possible reverse causality inherent in the relationship between borrower
discouragement and child poverty. We suspect the endogeneity to result in a downward bias in
the estimated relationship between borrower discouragement and child poverty or to produce

13
 R. E. Kofinti et al.

inconsistent estimates (Churchill and Marisetty 2020; Koomson and Danquah 2021). We
argue that, first, children in discouraged borrower homes are more likely to be multidimen-
sionally deprived. Secondly, households that are determined to provide basic needs for their
children in health or education can decide to apply for loans from the bank irrespective of
the barriers (Ansong et al. 2021; Koomson and Afoakwah 2022). Such barriers can result in
rejections in loan applications which breed discouraged borrower tendencies. We address the
endogeneity problem with the instrumental variable (IV) technique. We exploit the number
of discouraged neighbours within a cluster as an instrument for discouraged borrower homes.
The cluster variable in the GLSS7 has the number of households ranging from eight (8) to
fifteen (15), which we use to capture neighbours. This instrument captures the density of dis-
couraged borrowers within a respondent’s neighbourhood and can best describe the respond-
ent’s proclivity towards discouragement in the credit market or the level of borrower-related
despair in the cluster.
This instrument is consistent with the widely used leave-out mean approach which is
employed to ensure instrument validity. Although we consider our instrument to be valid
based on its satisfaction of the relevance and exclusion restriction conditions, there are
arguments for and against the leave-out mean approach. While some authors believe that
instruments based on leave-out means are likely to not satisfy the exclusion condition due
to spillover effects and simultaneity problems (Betz et al. 2018), others also suggest that
such instruments address endogeneity effectively because they operate as treatment vari-
ables (Hossain et al. 2019). To this end, leave-out mean instruments have been used across
a range of research areas such as gambling (Koomson et al. 2022), women’s empowerment
(Hossain et al. 2019), labour force participation (Lenze and Klasen 2017) and in religious
and political studies (Fruehwirth et al. 2019; Persson and Tabellini 2009). With respect
to our instrument, we argue that whether a household will capitulate to discouragement
tendencies towards borrowing depends on the information available to such a household,
which depends on the number of neighbours who are discouraged from borrowing. Hence,
we argue that the more discouraged neighbours one has, the higher the likelihood of being
discouraged which satisfies the relevance condition. We also do not expect the number of
discouraged neighbours to be directly associated with child deprivation without first influ-
encing the parents’ own borrower discouragement. Having explained this, we then estimate
the effect of borrower discouragement on multidimensional child deprivation using two-
stage least squares (2SLS).
To ensure robustness of the 2SLS results, we also employed the Lewbel (2012) hetero-
scedastic-adjusted instrument method and the control function approach (CFA) which have
been used widely to resolve endogeneity in existing studies (Koomson et al. 2021; Frempong
et al. 2021; Wooldridge 2015). The Lewbel method is proven to provide consistent estimates.
Consistent with previous studies, we use the Lewbel approach which combines both internal
and external instruments to address endogeneity (Frempong et al. 2021; Koomson et al. 2021;
Martey et al. 2022; Wooldridge 2015). The CFA method is essentially a 2SLS but unlike the
standard 2SLS, the CFA predicts the first stage residual and incorporates it as a control vari-
able in the second stage model (Wooldridge 2015). In rare circumstances, the CFA provides
more efficient second-stage estimators than 2SLS or can provide same estimates as the stand-
ard 2SLS (Tchetgen 2014).

13
Borrower discouragement and multidimensional child deprivation in Ghana

Table 1  Borrower discouragement and child deprivation (OLS results)

(1) (2) (3)
Variables All children Children under 5 years Children 6–17 years

Discouraged borrower 0.016** 0.024* 0.009

(0.007) (0.014) (0.009)
Female child –0.041*** –0.004 –0.053***
(0.007) (0.013) (0.008)
Age of household head –0.010 –0.037* 0.009
(0.011) (0.020) (0.014)
Female–headed home –0.014* 0.004 –0.025***
(0.008) (0.014) (0.009)
Educated head –0.126*** –0.140*** –0.162***
(0.008) (0.014) (0.009)
account ownership by head –0.089*** –0.060*** –0.103***
(0.010) (0.020) (0.011)
Rural household 0.211*** 0.307*** 0.164***
(0.008) (0.018) (0.009)
Married 0.008 –0.004 0.004
(0.012) (0.018) (0.015)
Separated/Divorced/Widowed 0.014 0.005 0.002
(0.018) (0.034) (0.021)
Household size 0.001 –0.003* 0.003**
(0.001) (0.002) (0.001)
Child year-of-birth cohort effects
2002–2005 –0.032*** –0.031***
(0.011) (0.011)
2006–2009 –0.047*** –0.047***
(0.012) (0.012)
2010–2013 0.203*** 0.004
(0.013) (0.016)
2014–2017 0.435*** 0.054*** 0.507***
(0.012) (0.014) (0.191)
Region fixed effects Yes Yes Yes
Observations 14,623 3,682 10,989
R-squared 0.209 0.182 0.095
Mean of child deprivation 0.415 0.302
Chow test: LR chi2 (16): (2) = (3) 669.39***

Standard errors in parentheses

*
p < 0.05, ** p < 0.01, *** p < 0.001

13
 R. E. Kofinti et al.

5 Results

5.1 Baseline results

Table 1 reports baseline estimates for the link between borrower discouragement and mul-
tidimensional child deprivation. We present estimates for all children in Column 1 whereas
Columns 2 and 3 display estimates for children under 5 years and children aged between
6–17 years respectively. To ensure reliable grounds for comparison of estimates from sub-
sampled models, we apply the widely used Chow test of differences to the results reported
in Columns 2 and 3 (see e.g., Chow 1960; Koomson et al. 2022; Nunoo et al. 2018). As
reported in Table 1 and for subsequent subsampled results, the chi-square value is signifi-
cant at 1% alpha level, implying that the estimated coefficients for the effect of borrower
discouragement on multidimensional child deprivation are statistically different between
under 5 children and those between 6–17 years.
In Column 1, we observe that borrower discouragement is associated with an increase
in multidimensional child deprivation by 1.6 percentage points. In Columns 2 and 3, bor-
rower discouragement is associated with an increase of 2.4 and 0.9 percentage points in
multidimensional child deprivation respectively, albeit not statistically significant in Col-
umn 3. Overall, these results indicate that the discouraged borrower syndrome worsens
multidimensional child deprivation especially for children under five years. The relative
analyses can be contextualised by comparing the ratio of the coefficient to the mean of the
dependent variable for the two subsamples. Since the coefficient is not statistically sig-
nificant in Column 3, we maintain that the effect of borrower discouragement in worsening
child deprivation is mainly evident among children under five years. The possible infer-
ence to be drawn for this outcome is that households experiencing discouraged borrower
syndrome miss out on the potential financial gains from credit access which is required
to directly increase household income and standards of living on one hand, and to invest
in non-farm business out of which the profit could increase the resources available to the
household. Hence, borrower discouraged households are more likely to lack the resources
to invest in basic needs of their children such as health and education (Agyire-Tettey et al.
2021; Fonta et al. 2020).

5.2 Endogeneity‑corrected results

In this section, we report 2SLS results because OLS produces biased estimates in the
existence of endogeneity. We estimate our 2SLS regressions using number of discour-
aged neighbours as instrument (see Table 2). All the Chow test results are significant at
1% alpha level, indicating that the coefficients of borrower discouragement are statistically
different between children under 5 and those aged 6 to 17 years. In line with expectations,
the first stage results indicate that an increase in the number of discouraged neighbours
positively relates with discouraged borrower homes (Columns 1 to 3). Consistent with
Stock and Yogo (2002), the F statistics across all models are greater than the threshold of
10 which indicates that our instrument does not have a weak relationship with borrower
discouragement. We can infer that the endogeneity of borrower discouragement causes a
downward bias in our baseline estimates. This conclusion is reached on the basis that our
instrumented estimates in Table 2 are bigger compared to the baseline results. Specifically,
in Table 2, borrower discouragement is associated with 4.0 percentage point increase in

13
Borrower discouragement and multidimensional child deprivation in Ghana

Table 2  Borrower discouragement and child deprivation (IV results)

(1) (2) (3)
Variables All children Children under Children 6–17 years
5 years

Discouraged borrower 0.040** 0.052** 0.037*

(0.016) (0.022) (0.021)
Child-specific variables Yes Yes Yes
Household head variables Yes Yes Yes
Household-level variables Yes Yes Yes
Region fixed effect Yes Yes Yes
Child year-of-birth cohort effects Yes Yes Yes
First stage
No. of discouraged neighbours 0.014*** 0.014*** 0.014***
(0.001) (0.001) (0.001)
F-statistic 2799.98 739.23 2066.58
Observations 14,623 3,682 10,989
R-squared 0.209 0.181 0.094
Mean of child deprivation 0.415 0.302
Chow test: LR chi2 (16): (2) = (3) 663.58***

Standard errors in parentheses

*
p < 0.05, ** p < 0.01, *** p < 0.001

multidimensional child deprivation (see Column 1). In Columns 2 and 3, we see that bor-
rower discouragement is associated with 5.2 and 3.7 percentage points increase in multidi-
mensional child deprivation for children under 5 and those aged 6 to 17 years respectively,
indicating that borrower discouragement worsens multidimensional child deprivation more
for those under five years. The contextualised ratio is 0.125 and 0.123 for children under
five and those aged 6 to 17 years which confirms our findings that the discouraged bor-
rower–child deprivation nexus is more pronounced for children under five years.

5.2.1 Gender and location of child

In this subsection, we report results for the rural–urban subsamples and for male-and
female-children to unravel any unobservable location and gender heterogeneities associ-
ated with borrower discouragement and child deprivations. The analyses in this section
also intersect with the SDG’s core objective of “leaving no one behind” by encouraging
researchers to engage in disaggregated analysis. Table 3 presents 2SLS estimates for male
and female children in Columns 1 and 2 as well as for rural and urban sub-samples in Col-
umns 3 and 4 respectively. The results of the Chow test are all significant at the alpha level
of 1%, indicating that the coefficients for borrower discouragement in the male–female and
rural–urban models are statistically different. In Column 2, we observe that borrower dis-
couragement is associated with 5.5 percentage point increase in multidimensional child
deprivation, but the outcome is not statistically significant in Column 1. Scrutinising the
estimates in Columns 1 and 2 reveal that increase in child deprivations emanating from
borrower discouragement is mainly observed among girls compared to boys. In Column 3,

13
 R. E. Kofinti et al.

Table 3  Borrower discouragement and child deprivation (Gender & Location)

Variables (1) (2) (3) (4)
Gender of child Location of child
Male Female Rural Urban

Discouraged borrower 0.027 0.055** 0.047** 0.025

(0.023) (0.023) (0.020) (0.027)
Sex of child No No Yes Yes
Household head variables Yes Yes Yes Yes
Household-level variables Yes Yes Yes Yes
Rural household Yes Yes No No
Region fixed effect Yes Yes Yes Yes
Child year-of-birth cohort effects Yes Yes Yes Yes
First stage
No. of discouraged neighbours 0.014*** 0.014*** 0.013*** 0.021***
(0.001) (0.001) (0.001) (0.001)
F-statistic 2799.98 1379.80 2219.39 912.55
Observations 7,525 7,098 9,758 4,865
R-squared 0.201 0.219 0.175 0.137
Mean of child deprivation 0.430 0.384 0.421 0.237
Chow test: LR chi2 (17): (2) = (3): 51.22*** (3) = (4): 239.47***

Standard errors in parentheses

*
p < 0.05, ** p < 0.01, *** p < 0.001
NB: The degree of freedom is 17 in these models because the gender and location variables are alternately
dropped in Columns (2) & (3) and (3) & (4) respectively

borrower discouragement increases multidimensional child deprivation by 4.7 percentage

points. The link between borrower discouragement and multidimensional child deprivation
is not significant in Column 4. Therefore, we conclude that child deprivations emanating
from borrower discouragement is mainly observed among rural children. This observation,
however, is not far-fetched since rural homes have higher incidence of discouraged bor-
rowers coupled with acute deprivations among children. This finding also complements
the literature regarding policy alternatives considered to bridge rural–urban gaps in child
deprivations.

5.3 Robustness/Sensitivity checks

In this subsection, several sensitivity checks are performed to test the robustness of
our estimates. First, we assess the robustness of our standard 2SLS estimates using
the Lewbel (2012) 2SLS technique. In Table 4 (Column 1), we report results from the
Lewbel 2SLS method which displays a significant effect of borrower discouragement
on child deprivation. Here, we combine our number of discouraged neighbours instru-
ment with internally generated instruments. Based on the reported Hansen J statistic in
Column 1, the null hypotheses of valid instruments are not rejected. In Column 1, we
observe that borrower discouragement increases multidimensional child deprivations

13
Borrower discouragement and multidimensional child deprivation in Ghana

Table 4  Borrower Variables (1) (2)

discouragement and child
deprivation (Lewbel and CFA) Lewbel internal Control
and external Function
Approach

Discouraged borrower 0.041*** 0.040**

(0.016) (0.016)
Residual –0.031*
(0.018)
Child-specific variables Yes Yes
Household head variables Yes Yes
Household-level variables Yes Yes
Region fixed effect Yes Yes
Child year-of-birth cohort effects Yes Yes
First stage
No. of discouraged neighbours 0.014*** 0.014***
(0.001) (0.001)
F-statistic 153.24 166.08
Observations 14,623 14,623
R-squared 0.209 0.209

Robust standard errors in parentheses

by 4.1 percentage points. Comparing the estimates of Lewbel 2SLS method and those
from the baseline and the externally instrumented results in Tables 1 and 2 respec-
tively reveals consistent results especially with the latter (externally instrumented).
Scrutinising the three coefficients, the Lewbel 2SLS method is marginally higher than
the externally instrumented results but much higher than the results from the baseline.
This observation further confirms the downward bias evident in our baseline estimates
and conforms with extant studies that leveraged this technique (Martey et al. 2022;
Churchill and Marisetty 2020).
Table 4 also shows the results for the CFA using column 2. The estimates are the
same as the externally instrumented results in Table 2, confirming the consistency in
our estimates. Overall, the findings from the Lewbel 2SLS method and the CFA are
indicative of the consistency and reliability in our core findings in Tables 1 and 2.
It implies that the positive association between borrower discouragement and mul-
tidimensional child deprivation is robust to different methods used in addressing
endogeneity.
Thirdly, we test whether the coefficients of borrower discouragement are sensitive
to alternative cut-offs of multidimensional child deprivation. Given that child depriva-
tion score is a continuum as asserted by Gordon (2003), the closer the cut-off to uni-
tary (1), the more severe the multiple deprivations experienced by the child. Our initial
measurement of child multidimensional poverty used a conservative cut-off of 0.33
which indicates multidimensional deprivations but in the sensitivity analysis presented
in Table 5, we used alternative cut-offs of 0.5 and 0.6 which shows severity of multiple
deprivations experienced by children in Columns 1 and 2 respectively. In addition, we
also used the raw deprivation scores in Column 3. Results from the measures that use
alternative cut-offs presented in Columns 1, 2 and that which uses the deprivation raw

13
 R. E. Kofinti et al.

Table 5  Borrower discouragement and child deprivation (Alternative cut-offs and measure)
Variables (1) (2) (3)
Dependent variable: Child deprivation
Deprivation cut-off Deprivation cut-off Deprivation score
(0.5) (0.6)

Discouraged borrower 0.027* 0.030** 0.027***

(0.015) (0.014) (0.007)
Child-specific variables Yes Yes Yes
Household head variables Yes Yes Yes
Household-level variables Yes Yes Yes
Region fixed effect Yes Yes Yes
Child year-of-birth cohort effects Yes Yes Yes
First stage
No. of discouraged neighbours 0.014**** 0.014**** 0.014****
(0001) (0001) (0001)
F-statistic 2799.98 2799.98 2799.98
Observations 14,623 14,623 14,623
R-squared 0.107 0.091 0.257

Robust standard errors in parentheses

***
p < 0.01, ** p < 0.05, * p < 0.1

scores presented in Column 3 are all consistent with our findings indicating that our
results are robust to alternative cut-offs, and the continuum nature of child depriva-
tions using the raw scores. This implies that the effect of borrower discouragement in
worsening child deprivation is consistent across different cut-offs used in identifying
the multidimensionally deprived child.

5.4 Potential channel analysis

Several potential channels through which borrower discouragement can influence child
deprivations were discussed in Sect. 2. Drawing from the many potential channels, we

Table 6  Borrower discouragement and household non-farm business

(1)
Variables Non-farm business log (income per capita)

Discouraged borrower –0.051*** –0.219***

(0.007) (0.022)
All control variables Yes Yes
Region fixed effects Yes Yes
Child year-of-birth cohort effects No No
Observations 14,620 14,526
R-squared 0.118 0.195

Robust standard errors in parentheses

***
p < 0.01, ** p < 0.05, * p < 0.1

13
Borrower discouragement and multidimensional child deprivation in Ghana

employ a two-step procedure to explore the potential role of non-farm business and house-
hold per capita income. This approach has been previously used by Alesina and Zhuravs-
kaya (2011); and Koomson and Churchill (2021). In the first step, we confirm that non-
farm business and household per capita income is significantly associated with borrower
discouragement. The results reported in Table 6 shows that borrower discouragement is
negatively associated with the probability of non-farm business venture start-up by 5.1 per-
centage points and reduces household per capita income by 2.2 percentage points respec-
tively which is consistent with the literature (Peprah et al. 2015; Peprah and Koomson
2015).
In the second step, we separately include non-farm business and household per capita
income as covariates in the multidimensional child deprivation model and report the results
in Table 7. To provide stronger grounds for estimates comparison, we re-estimate the base-
line model without the potential non-farm and household per capita income to guarantee
that all estimates are generated from the same sample. The re-estimated coefficient from
the baseline model is presented in Panel B. For non-farm business to qualify as a poten-
tial mediator, the coefficient of discouraged borrower must reduce in size or become sta-
tistically insignificant if the non-farm business is added as a covariate in the discouraged
borrower-multidimensional child deprivation model. This is the case in Table 7 (Column 1)
where the inclusion of non-farm business in the model renders the coefficient of discour-
aged borrower insignificant compared to our baseline model (see Panel B, Column 1). In
the same vein, the introduction of household per capita income in the multidimensional

Table 7  Effect of household non-farm business and income per capita (channel analysis)
Variables (1) (2)
Mediator: Non-farm busi- Mediator: log
ness (income per
capita)

Panel A: Results for mechanism

Discouraged borrower 0.012 0.009
(0.007) (0.007)
Non-farm business –0.067***
(0.008)
log(income per capita) –0.027***
(0.003)
All control variables Yes Yes
Region fixed effects Yes Yes
Child year-of-birth cohort effects Yes Yes
Observations 14,620 14,526
R-squared 0.213 0.215
Panel B: Initial results for comparison
Discouraged borrower (re-estimated Table 1) 0.015** 0.015**
(0.007) (0.007)
Observations 14,620 14,526
R-squared 0.209 0.209

Robust standard errors in parentheses

***
p < 0.01, ** p < 0.05, * p < 0.1

13
 R. E. Kofinti et al.

child deprivation model nullifies the significance of discouraged borrower as shown in Col-
umn 2. With varying opinions in favour of either income- or expenditure-based variables as
better indicators of welfare (Koomson et al. 2020; Beegle et al. 2012), we also perform the
mediation analysis using per adult household equivalent expenditure and report the results
in Appendix A4. Our findings using the expenditure variable yielded the same conclusion
as that of the income variable. The overall, implications of these observations are that non-
farm business and household per capita income are important channels through which bor-
rower discouragement influences multidimensional child deprivation.

6 Conclusion

In spite of the growing interest in child deprivation to understand its causes and poten-
tial policies needed for its relief, the discouraged borrower syndrome has received little
attention despite its possible role in stifling family resources required to meet children’s
needs. The few existing studies on this topic have not explored the relationship between
borrower discouragement and the multidimensional deprivations experienced by children.
To address the existing gaps in the literature, data was sourced from the GLSS7 to examine
the effect of borrower discouragement on multidimensional child deprivation. Endogene-
ity between borrower discouragement and multidimensional child deprivation is addressed
using the number of neighbours who are discouraged.
We also explore the potential channels through which borrower discouragement trans-
mits to multidimensional child deprivation. Overall, we find that borrower discouragement
has a positive effect on multidimensional child deprivation and this finding is robust to
alternative quasi-experimental methods. It is also robust to other sensitivity checks such
as alternative cut-offs used in identifying the multidimensionally deprived children. Gen-
der and location-wise, the results show consistency in increasing deprivations more among
girls and urban-located children. Furthermore, we identify inability of non-farm business
start-ups and constrained household per capita income as potential channels through which
borrower discouragement affects multidimensional child deprivation.
A policy option is for regulators of the financial systems to put in place the needed
structures that will provide information on how to access financial services to reduce the
prevalence of borrower discouragement. A limitation of this study is that our data permit-
ted us to only explore non-farm businesses and household per capita income as potential
channels despite the many possible channels identified in our literature review. This, how-
ever, provides fertile ground for future research to explore the other channels identified in
the literature and possible mediators that are peculiar to other geographical settings.
Supplementary Information The online version contains supplementary material available at https://​doi.​
org/​10.​1007/​s10888-​023-​09578-6.

Funding The authors did not receive support from any organization for the submitted work.

Data availability The GLSS7 dataset can be accessed from the websites below.
GLSS7: https://​www2.​stats​ghana.​gov.​gh/​nada/​index.​php/​catal​og/​97/​study-​descr​iption.

Declarations
Competing interests Authors declare no competing interest for the current study.

13
Borrower discouragement and multidimensional child deprivation in Ghana

References
Afoakwah, C., Deng, X., Onur, I.: Women’s bargaining power and children’s schooling outcomes: Evidence
from Ghana. Fem. Econ. 26(3), 1–29 (2020a)
Afoakwah, C., Nghiem, S., Scuffham, P., Huynh, Q., Marwick, T., Byrnes, J.: Impacts of air pollution
on health: Evidence from longitudinal cohort data of patients with cardiovascular diseases. Eur. J.
Health Econ. 21(7), 1025–1038 (2020b)
Agyire-Tettey, F., Asuman, D., Ackah, C.G., Tsiboe-Darko, A.: Multidimensional child poverty in
Ghana: measurements, determinants, and inequalities. Child Indic. Res. 14(3), 957–979 (2021)
Akuffo, A.: HFC’S pioneering role and rational for conversion to full banking activities: sustainabil-
ity of specialized lenders. The World Bank/Housing Finance Conference 7(2), 16–17. Retrieved
from https://​www.​macro​think.​org/​journ​al/​index.​php/​ijrd/​artic​le/​view/​17540 (2006). Accessed 13
Nov 2021
Alesina, A., Zhuravskaya, E.: Segregation and the quality of government in a cross section of countries.
Am. Econ. Rev. 101(5), 1872–1911 (2011)
Alkire, S., Foster, J.: Counting and multidimensional poverty measurement. J. Public Econ. 95(7), 476–
487 (2007)
Ansong, D., Okumu, M., Otchere, F., Koomson, I., Sherraden, M.: Addressing the burden of education
financing in low and lower-middle-income countries: The role of savings accounts, cash transfers,
and other income sources. J. Fam. Econ. Issues 42(4), 745–756 (2021)
Baiyegunhi, L.J.S., Fraser, G.C.: Determinants of household poverty dynamics in rural regions of the
Eastern Cape Province, South Africa. In: Poster Presented at the Joint 3rd African Association of
Agricultural Economists (AAAE) and 48th Agricultural Economists Association of South Africa
(AEASA) Conference, Cape Town, South Africa. (2010)
BAPPENAS and UNICEF: SDG Baseline Report on Children in Indonesia: Indonesia Ministry of
National Development Planning (BAPPENAS) and the United Nations Children’s Fund (UNICEF).
Jakarta: BAPPENAS and UNICEF, 1(1), 1–124. Retrieved from https://​www.​unicef.​org/​indon​esia/​
media/​731/​file/​SDG%​20Bas​eline%​20Full%​20Rep​ort.​pdf.​pdf (2017). Accessed 7 Nov 2021.
Beegle, K., De Weerdt, J., Friedman, J., Gibson, J.: Methods of household consumption measurement
through surveys: Experimental results from Tanzania. J. Dev. Econ. 98(1), 3–18 (2012)
Betz, T., Cook, S.J., Hollenbach, F.M.: On the use and abuse of spatial instruments. Polit. Anal. 26(4),
474–479 (2018)
Bima, L., Nurbani, R., Diningrat, R., Marlina, C., Hermanus, E., Lubis, S.: Urban child poverty and dis-
parity: The unheard voices of children living in poverty. The SMERU Research Institute. Retrieved
from https://​smeru.​or.​id/​en/​file/​1602/​downl​oad?​token=​bVzF_​H0Z (2017). Accessed 8 Nov 2021
Caudill, S.B.: An advantage of the linear probability model over probit or logit. Oxford Bull. Econ. Stat.
50(4), 425–427 (1988)
Chakravarty, S., Xiang, M.: The international evidence on discouraged small businesses. J. Empir.
Financ. 20, 63–82 (2013)
Chow, G.C.: Tests of equality between sets of coefficients in two linear regressions. Econometrica 28(3),
591–605 (1960). https://​doi.​org/​10.​2307/​19101​33
Churchill, S.A., Marisetty, V.B.: Financial inclusion and poverty: A tale of forty-five thousand house-
holds. Appl. Econ. 52(16), 1777–1788 (2020)
de Milliano, M., Plavgo, I.: Analysing multidimensional child poverty in sub-Saharan Africa: Findings
using an international comparative approach. Child Indic. Res. 11(3), 805–833 (2018)
Djoumessi, Y.F., Kamdem, C.B., Afari-sefa, V., Bidogeza, J.-C.: Determinants of smallholder vegetable
farmers credit access and demand in Southwest region, Cameroon. Econ. Bull. 38(2), 1231–1240
(2018)
Dornan, P.: Children, poverty and the sustainable development goals. Child. Soc. 31(2), 157–165 (2017)
Duncan, G.J., Yeung, W.J., Brooks-Gunn, J., Smith, J.R.: How much does childhood poverty affect
thelife chances of children? Am. Sociol. Rev. 63(3), 406–423 (1998). Retrieved from https://​smeru.​
or.​id/​en/​file/​1602/​downl​oad?​token=​bVzF_​H0Z. Accessed 7 July 2021
Fonta, C.L., Yameogo, T.B., Tinto, H., Van Huysen, T., Natama, H.M., Compaore, A., Fonta, W.M.:
Decomposing multidimensional child poverty and its drivers in the Mouhoun region of Burkina
Faso, West Africa. BMC Public Health 20(1), 1–17 (2020)
Freel, M., Carter, S., Tagg, S., Mason, C.: The latent demand for bank debt: Characterizing “discouraged
borrowers.” Small Bus. Econ. 38(4), 399–418 (2012)
Frempong, R.B., Orkoh, E., Kofinti, R.E.: Household’s use of cooking gas and Children’s learning out-
comes in rural Ghana. Energy Econ. 103, 105617 (2021)

13
 R. E. Kofinti et al.

Fruehwirth, J.C., Iyer, S., Zhang, A.: Religion and depression in adolescence. J. Polit. Econ. 127(3),
1178–1209 (2019)
Gentry, W., Hubbard, R.: Entrepreneurship and Household Saving. Advances in Economic Analysis &
Policy 4(1), 1–57 (2004)
Ghana Statistical Service: Ghana living standards survey round 7 (GLSS 7). Main report. Ghana Statis-
tical Service, Accra. Retrieved from https://​open.​africa/​datas​et/​38102​d96-​1393-​4918-​a1a75​56eca​
8491ad/​r esou​r ce/​8 39a1​7 58-​1 46c-​4 0cd-​9 57d-​3 7d26​a a84f​b 6/​d ownl​o ad/​g lss7-​m ain-​r eport_​f inal.​
pdf (2019). Accessed 7 July 2021
Gordon, D.: The concept and measurement of poverty. Poverty and social exclusion in Britain. The mil-
lennium survey, pp. 29–69. Policy Press, Bristol (2006)
Gordon, D., Nandy, S., Pantazis, C., Pemberton, S., Townsend, P.: The distribution of child poverty in
the developing world. Centre for International Poverty Research, Bristol (2003)
Gordon D.: Child poverty in the developing world. Retrieved from Policy Press. Retrieved from https://​books.​
google.​com/​books?​hl=​en&​lr=​&​id=​gxJLH​WGiCQ​C&​oi=​fnd&​pg=​PR5&​dq=​Gordon+​et.+​al(2003)+​
Pover​ty&​ots=​iVwPj​PefeE​&​sig=​hSKIb​FkUmx​C2Qd0​8G1yR​VarzI​5s (2003). Accessed 10 July 2021
Honohan, P., & King, M.: Cause and effect of financial access: Cross-country evidence from the Finscope
surveys, pp. 45–84. Banking the World: Empirical Foundations of Financial Inclusion (2012).
Hossain, M., Asadullah, M.N., Kambhampati, U.: Empowerment and life satisfaction: Evidence from Bang-
ladesh. World Dev. 122, 170–183 (2019)
Kingdon, G. G., & Knight, J.: Well-being poverty versus income poverty and capabilities poverty in South
Africa. In: TIPS/DPRU forum paper on African development and poverty reduction. The Macro-Micro
Linkage. Retrieved from https://​www.​tips.​org.​za/​files/​Kingd​onKni​ght-​Wel-​being_​Pover​ty_​versus_​
income_​pover​ty.​PDF (2004). Accessed 10 July 2021
Kofinti, R.E., Annim, S.K.: Child deprivation and income poverty in Ghana. In: Measuring poverty and
wellbeing in developing countries, p. 178. (2016). https://​doi.​org/​10.​1093/​acprof:​oso/​97801​98744​801.​
003.​0012
Kon, Y., Storey, D.J.: A theory of discouraged borrowers. Small Bus. Econ. 21(1), 37–49 (2003)
Koomson, I., Danquah, M.: Financial inclusion and energy poverty: Empirical evidence from Ghana.
Energy Econ. 94, 105085 (2021)
Koomson, I., Villano, R.A., Hadley, D.: Effect of financial inclusion on poverty and vulnerability to poverty:
Evidence using a multidimensional measure of financial inclusion. Soc. Indic. Res. 149(2), 613–639
(2020)
Koomson, I., Bukari, C., Villano, R.A.: Mobile money adoption and response to idiosyncratic shocks:
Empirics from five selected countries in sub-Saharan Africa. Technol. Forecast. Soc. Chang. 167,
120728 (2021)
Koomson, I., Afoakwah, C.: Can financial inclusion improve children’s learning outcomes and late school
enrolment in a developing country? Appl. Econ. 55(3), 1–18 (2022). https://​doi.​org/​10.​1080/​00036​
846.​2022.​20866​83
Koomson, I., Churchill, S.A.: Ethnic diversity and food insecurity: evidence from Ghana. J. Dev. Stud.
57(11), 1912–1926 (2021). https://​doi.​org/​10.​1080/​00220​388.​2021.​19286​41
Koomson, I., Churchill, S.A., Munyanyi, M.E.: Gambling and Financial Stress. Soc. Indic. 163, 473–503 (2022).
Retrieved from https://​link.​sprin​ger.​com/​artic​le/​10.​1007/​s11205-​022-​02898-6. Accessed 10 July 2021
Koomson, I., Kofinti, R.E., Laryea, E.: Financial inclusion and multidimensional child poverty. Rev. Econ.
Househ. 1–24 (2023). https://​doi.​org/​10.​1007/​s40609-​022-​00256-5
Lenze, J., Klasen, S.: Does women’s labor force participation reduce domestic violence? Evidence from
Jordan. Fem. Econ. 23(1), 1–29 (2017)
Lewbel, A.: Using heteroscedasticity to identify and estimate mismeasured and endogenous regressor mod-
els. J. Bus. Econ. Stat. 30(1), 67–80 (2012)
Martey, E., Etwire, P.M., & Koomson, I.: Parental time poverty, child work and school attendance in Ghana.
Child Indic. Res. 15, 1489–1515 (2022). https://​doi.​org/​10.​1007/​s12187-​022-​09926-4
Minujin, A., Delamonica, E.: Multidimensional child poverty in Tanzania: analysis of situation, changes and
sensitivity of thresholds1. Global Child Poverty and Well-Being. 1, 263–286. (2012). https://​doi.​org/​
10.​51952/​97814​47
Minujin, A.: Child poverty in East Asia and the Pacific: Deprivations and disparities, A study of seven coun-
tries. Retrieved from http://​ideas.​repec.​org/p/​uce/​wpaper/​1109.​html (2011). Accessed 10 July 2021
Nunoo, J., Koomson, I., Orkoh, E.: Household deficiency in demand for water: do water source and travel
time matter? Glob. Soc. Welf. 5(3), 179–187 (2018). https://​doi.​org/​10.​1007/​s40609-​017-​0092-6
Opportunity International: Getting more children into better schools with social finance. Retrieved from
https://​edufi​nance.​org/ (2022). Accessed 12 July 2021

13
Borrower discouragement and multidimensional child deprivation in Ghana

Ortiz, I., Moreira Daniels, L., Engilbertsdóttir, S.: Child poverty and inequality: New perspectives. (2012).
https://​doi.​org/​10.​2139/​ssrn.​20397​73
Oshio, T., Sano, S., Kobayashi, M.: Child poverty as a determinant of life outcomes: Evidence from nation-
wide surveys in Japan. Soc. Indic. Res. 99(1), 81–99 (2010)
Pemberton, S., Gordon, D., & Nandy, S.: Child rights, child survival and child poverty: The debate. In:
Global child poverty and well-being: Measurement, concepts, policy and action, pp. 19–37. (2012).
https://​doi.​org/​10.​51952/​97814​47301​141.​ch002
Pemberton, S., Gordon, D., Nandy, S., Pantazis, C., Townsend, P.: Child rights and child poverty: Can the
international framework of children’s rights be used to improve child survival rates? PLoS Medicine
4(10), e307 (2007)
Peprah, J.A., Afoakwah, C., Koomson, I.: Savings, entrepreneurial trait and self-employment: Evidence
from selected Ghanaian Universities. J. Glob. Entrep. Res. 5(1), 1–17 (2015)
Peprah, J.A., Koomson, I.: Capital and credit sources and household non-farm income in Ghana. J. Arts Soc.
Sci. 3(1) (2015). Retrieved from https://​ssrn.​com/​abstr​act=​29644​39. Accessed 6 July 2021
Peprah, J.A.: Microcredit, Child Education, and Health Outcomes: A Case Study from Ghana. In: Financing
sustainable development in Africa, pp. 339–365. Springer (2018)
Persson, T., Tabellini, G.: Democratic capital: The nexus of political and economic change. Am. Econ. J.
Macroecon. 1(2), 88–126 (2009)
Prime-Stat, P.V.C.: Access to long-term capital: The case of Ghana home loans. Centre for affordable hous-
ing finance in Africa, 11, 1–26, (2018). Retrieved from https://​housi​ngfin​ancea​frica.​org/​app/​uploa​ds/​
Ghana-​Home-​Loans-​Case-​Study_​25-​June-​2018-1.​pdf. Accessed 6 July 2021
Ratcliffe, C., McKernan, S.-M.: Childhood poverty persistence: Facts and consequences. Urban Institute,
Washington DC (2010)
Rui, L., Xi, Z.: Econometric analysis of credit constraints of Chinese rural households and welfare loss.
Appl. Econ. 42(13), 1615–1625 (2010)
Sanfilippo, M., De Neubourg, C., Martorano, B.: The impact of social protection on children. UNICEF Office of
Research. Retrieved from https://​www.​unicef-​irc.​org/​publi​catio​ns/​pdf/​iwp_​2012_​06.​pdf (2012). Accessed
6 July 2021
Sen, A.: Poverty: An ordinal approach to measurement. Econometrica 44(2), 219–231 (1976). https://​doi.​
org/​10.​2307/​19127​18
Stock, J.H., Yogo, M.: Testing for weak instruments in linear IV regression. National Bureau of Economic
Research Cambridge, Mass (2002)
Swamy, V.: Financial inclusion, gender dimension, and economic impact on poor households. World Dev.
56, 1–15 (2014)
TchetgenTchetgen, E.J.: Identification and estimation of survivor average causal effects. Stat. Med. 33(21),
3601–3628 (2014)
Townsend, P.: Poverty in the United Kingdom: A survey of household resources and standards of living.
Univ of California Press. Retrieved from http://​pover​ty.​ac.​uk/​system/​files/​towns​end-​book-​pdfs/​PIUK/​
piuk-​preli​ms.​pdf (1979). Accessed 10 May 2021
Tran, M.M.C., Gan, C., Hu, B.: Credit constraints and impact on farm household welfare: Evidence from
Vietnam’s North Central Coast region. Lincoln University. Retrieved from https://​resea​rchar​chive.​linco​
ln.​ac.​nz/​bitst​ream/​handle/​10182/​6033/​FoC_​wp_​10.​pdf;​seque​nce=3 (2014). Accessed 10 May 2021
UNICEF & Global Coalition against Child Poverty: A world free from child poverty: A guide to the tasks to
achieve the vision. United Nations Children’s Fund (UNICEF) and the Global Coalition to End Child
Poverty. Retrieved from https://​www.​unicef.​org/​media/​65161/​file/​Child-​Pover​ty-​SDG-​Guide-​EN.​pdf
(2017). Accessed 10 May 2021
United Nations: Convention on the rights of the child, the General Assembly resolution 44/25. Retrieved from
https://​www.​un.​org/​en/​devel​opment/​desa/​popul​ation/​migra​tion/​gener​alass​embly/​docs/​globa​lcomp​act/A_​
RES_​44_​25.​pdf (1989). Accessed 10 May 2021.
Wooldridge, J.M.: Introductory econometrics: A modern approach. Cengage learning. Retrieved from http://​
lms.​aambc.​edu.​et:​8080/​xmlui/​bitst​ream/​handle/​12345​6789/​217/​Intro​ducto​ry%​20eco​nomet​rics.%​20A%​
20mod​ern%​20app​roach%​20(%​20PDF​Drive%​20).​pdf?​seque​nce=1 (2015). Accessed 10 May 2021

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a
publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript
version of this article is solely governed by the terms of such publishing agreement and applicable law.

13
Appendix
Table A1: Multidimensional child deprivation indicators
Dimension Indicator Deprivation cut-off definition Age
Cohorts
Cooking fuel Child is deprived if household of dwelling is cooking with solid Under 5
fuels and not cooking outside the house and 6 to 17
years
Water Deprived if household gets drinking water from unclean source Under 5
(Tanker supply/Vendor provided; unprotected well; unprotected and 6 to 17
spring; river/stream; dugout/pond/lake/dam/canal; other) OR it years
takes 30 minutes or more to collect water, round trip
Overcrowding Deprived if household has more than 3 people per sleeping room, on Under 5
average and 6 to 17
Living
years
conditions
Housing Deprived if household of dwelling uses inadequate flooring Under 5
(earth/mud, other) or walls (earth/mud, Palm leaves/Thatch and 6 to 17
(grass/Raffia), other) years
Electricity Deprived if household of dwelling does not have electric power Under 5
and 6 to 17
years
Toilet Deprived if household of dwelling has no toilet facilities, uses Under 5
bucket/pan, uses public toilet, other OR toilet is shared outside the and 6 to 17
house years
School Deprived if child (6-17years) is not attending school Age 6 to 17
Education
attendance years
Nutrition Deprived if child under 5 who is undernourished (underweight or Under 5
stunted) and 6 to 17
years
Health Deprived if child not covered by the national health insurance Under 5
insurance scheme and 6 to 17
years
Vaccination Deprived if child under 5 did not complete the six vaccinations (bcg, Under 5
Health
polio, penta, measles, yellow fever, and vitamin A)
Current Deprived if during the past 2 weeks the child suffered from either an Under 5
Wellness illness or injury and 6 to 17
years
Quality of Deprived if no household member 24 years or older has at least a Under 5
childcare Senior High (Secondary) School education and 6 to 17
years
Table A2: Summary of deprivation indicators across age groups
Under-five 6 to 17 All children
Dimensions Indicators (%) (%) (%)
Cooking Fuel 32.0 35.3 34.3
Water 24.1 24.7 24.5
Overcrowding 63.5 58.4 59.9
Living Conditions
Housing 41.0 40.0 40.3
Electricity 21.0 22.0 21.7
Toilet 90.5 88.5 89.1
Education School Attendance NA 9.8 9.8
Health Insurance 25.6 17.7 20.1
Undernourished 25.9 16.5 19.3
Health Vaccination 90.5 NA 90.5
Recent Illness 18.2 8.5 11.4
Quality Care 31.8 31.3 31.5
NA: Not Applicable
 Table A3: Summary statistics
Variable Description Full sample Discouraged Non-
sample discouraged
sample
Mean Std. Mean Std. Mean Std.
Dev. Dev. Dev.
Child deprivation (cut-off=0.33) Dummy variable equals 1 if multidimensional child deprivation score exceeds 0.33 0.408 0.491 0.419 0.494 0.400 0.490
Child deprivation (cut-off=0.5) Dummy variable equals 1 if multidimensional child deprivation score exceeds 0.5 0.153 0.360 0.159 0.366 0.149 0.356
Child deprivation (cut-off=0.6) Dummy variable equals 1 if multidimensional child deprivation score exceeds 0.6 0.106 0.308 0.112 0.316 0.101 0.302
Child deprivation score Continuous variable for multidimensional child deprivation score 0.301 0.195 0.307 0.198 0.297 0.193
Discouraged borrower Binary variable equals 1 if household is discouraged 0.393 0.488
Female child Dummy variable equals 1 if child is female 0.485 0.500 0.494 0.500 0.480 0.500
Age of household head Age of the household head 3.305 0.542 3.300 0.535 3.308 0.547
Child year-of-birth cohorts
Born between 2002-2005 Binary variable equals 1 if child was born between 2002-2005 0.273 0.445 0.274 0.446 0.272 0.445
Born between 2006-2009 Binary variable equals 1 if child was born between 2006-2009 0.224 0.417 0.225 0.417 0.223 0.417
Born between 2010-2013 Binary variable equals 1 if child was born between 2010-2013 0.166 0.372 0.163 0.370 0.167 0.373
Born between 2014-2017 Binary variable equals 1 if child was born between 2014-2017 0.165 0.371 0.160 0.367 0.168 0.374
Female-headed home Binary variable equals 1 if household head is female 0.544 0.498 0.549 0.498 0.540 0.498
Educated head Binary variable equals 1 if household head is educated 0.631 0.483 0.598 0.490 0.652 0.476
account ownership by head Binary variable equals 1 if household head owns an account 0.140 0.347 0.135 0.341 0.144 0.351
Rural household Binary variable equals 1 if household is located in a rural area 0.667 0.471 0.672 0.470 0.665 0.472
Married Binary variable equals 1 if household is married 0.398 0.490 0.405 0.491 0.394 0.489
Separated/Divorced/Widowed Binary variable equals 1 if household is Separated/Divorced/Widowed 0.095 0.294 0.092 0.289 0.098 0.297
Household size Number of persons in the household 6.822 3.463 7.208 3.718 6.572 3.264
log(income per capita) Continuous variable for household income per capita 7.152 1.580 6.984 1.578 7.261 1.571
No. of discouraged neighbours Continuous variable for number of discouraged neighbours 13.63 17.185 23.60 20.44 7.179 10.48
4 9 7 1
No.: number
Table A4: Mediation analysis using per adult household equivalent expenditure
(1) (2)
Variables Dependent variables: log(per adult Dependent variables: Child
household equivalent expenditure) deprivation
Mediator: per adult household
equivalent expenditure
Panel A: Results for mechanism
Discouraged borrower -0.061*** 0.010
(0.011) (0.007)
log(per adult household -0.088***
equivalent expenditure)
(0.005)

Observations 14,623 14,623

R-squared 0.398 0.224
Panel B: Initial results for
comparison
Non-farm entrepreneurship (re- 0.016**
estimated Table 1)
(0.007)

Observations 14,623
R-squared 0.209
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1