How the age of the head of households matter in the incidence of financial inclusion on energy poverty in Ghana

doi:10.21203/rs.3.rs-2080376/v1

We extend Koomson and Danquah (2021, EP) in order to understand if and how the age of the household head matters in the incidence of financial inclusion on energy poverty in Ghana. We revisit the underlying study within a framework of interactive regressions in order to establish household head ages that should be avoided in order for financial inclusion to continuously reduce energy poverty. In the estimation exercise that is based on pooled data and two stage least squares, we establish a negative net effect from the role of household head age in moderating the incidence of financial inclusion on energy poverty. An extended analysis provides household head ages that should be avoided in order for financial inclusion to reduce energy poverty. These critical household head ages are contingent on the area of household (i.e. rural versus urban) as well as the gender of household age( male versus female).

JEL Classification: D03; D12; D14; I32; Q41

Energy poverty

Financial inclusion

Consumption poverty

Education

Household income

The present study extends Koomson and Danquah (2021, EP)¹ in order to articulate the relevance of age in the nexus between financial inclusion and energy poverty. The study is premised on two main motivational elements, notably: (i) the importance of replication studies in social sciences and (ii) the relevance of extending Koomson and Danquah (2021) with an emphasis on how age affects the established nexus in order to provide more scholarly insights and room for policy implications. These motivational elements are expanded in the same chronology as highlighted.

First, consistent with contemporary replication studies (Pridemore et al., 2018; Asongu et al., 2020, 2021), for policy makers to be better informed, comparative replication exercises are fundamental. To put this insight into more perspective: “the replicability of research results is also a central tenet to the scientific research process” (Cook, 2014, p. 233) and “Replications are an important part of the research process because they allow for greater confidence in the findings” (McEwan et al., 2018, p. 235) and “the replicability of research results is also a central tenet to the scientific research process” (Cook, 2014, p. 233). It is partly on theses premise that the present study extends Koomson and Danquah (2021).

Second, while Koomson and Danquah (2021) have concluded that financial inclusion reduces energy poverty in Ghana, the nexus could be contingent on many other factors. For instance, with regard to age, the favorable incidence of financial inclusion on energy poverty could be contingent on the age of the household head, such that it is only below or above a certain age of the household head that financial inclusion reduces energy poverty. On the basis of this argument, the blanket nexus between financial inclusion and energy poverty established by Koomson and Danquah (2021), while relevant as a basis of the present study, can be more policy worthwhile and effective within the framework of interactive regressions. In this perspective, policy makers can be informed on specific household head age groups that should be targeted in order for the established finding in the underlying study to be more policy worthwhile.

In the light of the above, actionable age thresholds or critical masses in the financial inclusion and energy poverty nexus are fundamental in tailoring the positive externalities of financial inclusion, especially in the light of achieving the sustainable development goals (SDGs) of the United Nations, not least, because financial inclusion has been established to be relevant in the achievement of attendant poverty-related SDGs (UNCDF, 2022). For instance, it has been concluded by contemporary poverty and SDGs literature that unless concerns about inequality such as financial exclusion are addressed, most Sub-Saharan African countries are not likely to achieve the extreme poverty target of the United Nations’ SDGs by 2030 (Bicaba et al., 2017; Asongu & Odhiambo, 2018).

The rest of the study is organized as follows. The literature review and testable hypothesis are covered in Section 2 while Section 3 is concerned with the data and methodology. The empirical results are disclosed in Section 4. Section 5 concludes with implications and future research directions.

¹ The terms ‘underlying literature’ ‘underlying study’ and Koomson and Danquah (2021), are used interchangeably throughout this study.

This study can be summarized in the following: household head age as an argument for the nexus between financial inclusion and energy poverty. Age matters because being too old could decrease financial inclusion possibilities (employment, activity, updates on financial instruments). In this line of logic, Baah-Boateng (2015) argues that unemployment in Ghana increases with education and declines with age. As a result, higher unemployment rates among the youths than the old. Besides, Canagarajah et al. (2001) posit that the distribution of income within the Ghanaian economy varies based on several factors among which are gender and age. All these differences in the distribution of finance within different age groups and inability by some age groups to easily procure a paid employment leads to different attitudes towards the adoption of financial inclusion instruments (opening a bank account, using ATM, using mobile baking services etc.). At the same time, Koomson and Danquah (2021) argue that increase in financial inclusion in Ghana reduces energy poverty especially for those who are in the category of the employees and in households with more income.

From the above development, the first hypothesis can be stated thus:

Hypothesis 1: household age moderates financial inclusion for an overall favorable incidence in reducing energy poverty.

Focusing uniquely on age, Zins and Weill (2016) for the African continent as a case argue that the probability of an individual or household to adopt financial inclusion tools increases with age, and that this effect is more pronounced with the individual increases its education and income as he/she grows older. Moreover, Allen et al. (2016) argue that the probability of owning an account in an established financial institution is higher with age and that the older people turn to borrow more from these institutions. On their part, Fungácová and Weill (2015) put it that women are less likely to be financially included and that older people are more concerned about lack of money, distance and religious reasons and as a result, prefer the use of financial services for security.

The second hypothesis can therefore be stated thus:

Hypothesis 2: some age limits should not be exceeded in order for Hypothesis 1 to remain valid.

Away from the established nexus, several other determinants of financial inclusion have been examined in the literature, including age, marital status. In this regard, Crentsil et al. (2019) examine the explaining factors of multidimensional energy poverty in Ghana and posit that energy poverty is higher among households headed by females and in rural households. They further argue that increase in education cannot really explain energy poverty in Ghana, whereas, increase in the number of household members increases energy poverty. Adusah-Poku and Takeuchi (2019a) found similar results and further argued that there is need for fighting income poverty at first place as one of the strategies of energy poverty in Ghana. Recently, Martey (2022) has argued that one of the ways of energy poverty is through crop diversification by Ghanaian household, with the effect prominent for urban and high-income earning households. Besides, household income is identified as the modulating mechanism for this effect to be feasible. At the same time, Lin and Okyere (2022) examine the impact of acting prosocial on multidimensional energy poverty in Ghana and suggest that people who acted prosocial by volunteering were more likely to be energy poor than their counterparts who did not volunteer. Besides, the number of times volunteers got involved in prosocial activities augmented their chances of being energy-poor by about 13.6%.

3.1 Data

The data that is used for the present research is consistent with Koomson and Danquah (2021) and sourced from the 6th (GLSS6) and 7th (GLSS7) rounds of the Ghana Living Standards Survey (GSS, 2014; 2019) that were gathered from 10 regions in the country by the Ghana Statistical Service (GSS). It is important to articulate that the GLSS7 and GLSS6 were collected respectively in 2016/2017 and 2012/2013. Consistent with the narrative, the surveys are based on a two-stage probability sampling approach and entails dimensions of fuel use and energy, housing conditions and demography, water and sanitation, health, employment, insurance services, financial access, migration, governance, non-farm activities and agriculture, inter alia. The motivation for selecting GLSS6 and GLSS7 is primarily owing to data availability constraints at the time of study. Moreover, the attendant survey rounds consistently entail variables that are employed in empirical analyses of this type, especially as it pertains to proxies that are relevant in deriving indexes, explanatory variables for potential mechanisms as well as attendant instruments.

It is worthwhile to note that 18,000 and 15,000 households are respectively covered by the GLSS6 and GLSS7 with corresponding rates of responses of 93.4% for GLSS7 and 93.2% for GLSS6. In the light of these insights, the most updated sample size of the GLSS7 and GLSS6 were respectively 14, 009 and 16,772. Upon combining the sections/file entailing the variables of interest, the corresponding sample size are slightly reduced to a total pool of 30, 606, consisting of 16, 760 for GLSS6 and 13,846 for GLSS7. Moreover, in the light of missing information, the corresponding regressions in the analyses consists of 6,545 forGLSS7 and 16,169 forGLSS6, leading to a pooled data entailing 22, 714 households. The substantial drop in observation in the post-estimation is associated with the financial inclusion proxy in the GLSS7 in which 6,910 observations are not apparent owing to non-responses that are linked to the indicators that constitute the index. The descriptive statistics of the variables employed is shown in Appendix 1.

3.1 Energy Poverty

In accordance with the energy poverty literature (Koomson & Danquah, 2021), both subjective and objective measures as used. First, on the front of the objective approach the energy expenditure-income approach is put in perspective with respect to energy poverty (% of household income) that is spent on fuel and energy. In accordance with Boardman (2013) and Churchill and Smyth (2020), the higher the proportion of the energy measure, the higher the existing energy poverty level. Moreover, a secondary objective framework is to utilize 10% of energy poverty as a threshold within the remit of the expenditure-income in order to articulate households that are poor in energy as households that invest above 10% of their household income on fuel and energy (Bouzarovski & Petrova, 2015; Koomson & Danquah, 2021; Boardman, 2013). Second, on the subjective front, poverty in energy can be appreciated in terms of material deprivation, especially when seasons are cold. Consistent with Churchill and Smyth (2020), for the most part, this indicator usually takes the value of 1 in a scenario where the household cannot heat the house given lack of funds and 0 if otherwise. Note should also be taken on the perspective, the underlying measures are for the most part used in studies involving developed countries owing to comprehensive data availability on issues surrounding heating as well as expenditures related to household fuel and energy.

A measurement that embodies both the subjective and objective measures of poverty in energy is the multidimensional energy poverty index (MEPI). This attendant measure is often employed in developing nations owing to its conceptualization and how such relates to economic conditions and rate of adoption of clean energy (Nussbaumer et al., 2013; Churchill & Smyth, 2020). In accordance with the extant literature focusing on developing nations (Nussbaumer et al., 2013; Adusah-Poku & Takeuchi, 2019 b; Crentsilet al., 2019), the MEPI measure is used owing to constraints in availability of data, especially in relation to energy poverty measures in GLSS.

As reported in the extant literature (Nussbaumer et al., 2013; Crentsil et al., 2019; Adusah-Poku & Takeuchi, 2019b; Koomson & Danquah, 2021), the MEPI entails five dimensions, consisting of six indicators. As apparent in Appendix 2, the five dimensions entail lighting, cooking, communication, education/entertainment and household appliances that are connected.

In accordance with Alkire and Foster (2011), the MEPI is founded on the proxy of multidimensional poverty from the Oxford Poverty and Human Development Initiative which is based on the contributions of Amartya Sen to the literature on capabilities and deprivations. As documented by Koomson and Danquah (2021), the attendant five dimensions can be weighted equally with a corresponding weight of 0.2 assigned to respective dimensions. Notwithstanding this perspective, the lighting and cooking dimensions are provided more weights compared to the other three dimensions owing to their relative importance in energy poverty in the light of Nussbaumer et al. (2013) and Adusah-Poku and Takeuchi (2019b). When cooking and lighting are compared, more weight is attributed to cooking owing to the fact that it is a critical need in energy for households in developing countries. On this premise, the two indicators in the cooking dimension are allocated a 0.205 weight whereas 0.200 is attributed to the lighting dimension. Of the three dimensions that are left, each of them is attributed a weight of 0.13. Appendix 2 captures the attendant indicators that also reflect the relative deprivations and by extension, are employed to compute the scores on energy deprivation. For each of the households, the deprivation score is appreciated as the total of deprivations that ranges from 0 to 1 represented as follows:

$${d_i}={w_1}{I_1}+{w_2}{I_2}+ \cdots +{w_n}{I_n}$$

1

where ${d_i}$ represents the score on household energy deprivation, ${I_i}=1$in a situation where the household is deprived in indicator i and ${I_i}=0$ if otherwise. ${w_i}$denotes the weight associated with indicator i with $\sum\nolimits_{{i=1}}^{d} {{w_i}=1}$. In the light of Nussbaumer et al. (2013), a cut off threshold of 0.33is employed and such implies that a household having a score of energy deprivation of at least 0.33 is a household that is energy poor.

3.2 Financial inclusion (FI)

Following Koomson and Danquah (2021), the study employs a FI multidimensional proxy which is consistent with the attendant contemporary energy poverty literature (Zhang & Posso, 2019; Churchill & Marisetty, 2020; Churchill et al., 2020). Hence, within the current empirical setting, the four dimensions of FI used by Koomson and Danquah (2021) are also employed, namely: credit/loan access, ownership of bank account, receipt of financial remittance from a financial institution by means of mobile money and ownership of insurance. These measurements of FI which are consistent with Koomson and Danquah (2021) are provided in Appendix 3. Concerning the attribution of weights, 0.25 is attributed to each of the dimensions employed to compute the household deprivation score provided in Eq. (1). Still following the underlying literature, a cut-off point or threshold of 0.50 is employed and a corresponding value of 1 is assigned for households which are associated with a deprivation score of less than 0.50 and the value of 0 is assigned if the household financial deprivation score exceeds 0.50.

3.2. Methodology

Following Koomson and Danquah (2021), this study employs the linear probability model (LPM) such that financial inclusion is tailored to influence energy poverty contingent on the age of the household head. It is important to articulate that consistent with the attendant narrative, a pooled ordinary least squares (OLS) approach is employed as opposed to the random and fixed effects models because collected data (i.e. GLSS7 and GLSS6) are repeated cross sections that do not encompass a panel data structure. In essence, consistent with previous research on the problem statement (Churchill & Marisetty, 2020; Koomson et al., 2020), the concern of endogeneity owing to simultaneity or reverse causality is addressed in the estimation process with an instrumental variable approach as apparent in Equations (2) and (3). In the attendant equations, while financial inclusion is postulated to reduce energy poverty, financial inclusion is also a function of other determinants as clarified in the subsequent paragraphs. The concern of reverse causality builds on the perspectives that: (i) financial inclusion mitigates energy poverty and (ii) the condition of energy poverty can also motivate a households to create a bank account and save within the attendant formal financial institution in order to have better initial financial conditions that are essential for financial access in view of ultimately mitigating energy poverty. The first stage and second stage of the instrumental variable estimation process are provided in Eq. (1) and Eq. (2), respectively.

Reduced form equation (stage 1)

$${FI}_{it}=\delta +\gamma {Dist}_{it}+\eta {X}_{it}+{\vartheta }_{r}+{\mu }_{t}+{\epsilon }_{it}$$

2

Structural equation (stage 2)

$${ EPov}_{it}=\alpha +\beta {\widehat{FI}}_{it}+\lambda {X}_{it}+{\vartheta }_{r}+{\mu }_{t}+{\nu }_{it}$$

3

where ${EPov}_{it}$ denotes the status of energy poverty of a household$i$ at time $t$, with time corresponding to the period of each GLSS round; ${FI}_{it}$ reflects an$i$ household’s financial inclusion statusat time $t$; while$X$ denotes a vector of covariates that are established in the extant energy poverty literature, inter alia, gender, age, education, marital status, household size, location, employment status of head of household.$\delta and\alpha$ respectively, represent constant values; ${\vartheta }_{r}$ and ${\mu }_{t}$ denote fixed effects pertaining to the region and round of GLSS, respectively whereas $\epsilon$ and $\nu$ are the random error terms.

It is worthwhile to emphasize that consistent with Koomson and Danquah (2021), in Eq. (2) above, ‘distance to the nearest bank’ (i.e. $Dist$) is used as an instrument for financial inclusion. The attendant instrument has been used in the extant literature that is oriented towards the nexus between financial inclusion and poverty (Churchill et al., 2020; Koomson et al., 2020; Churchill & Marisetty, 2020; Koomson et al., 2020). It is argued in the attendant literature that financial inclusion is connected to distance to the nearest bank, not least because if households are nearer to a bank, there is obviously greater access to financial services and less associated costs (e.g. transport cost) (Demirgüç-Kunt & Klapper, 2012; Koomson et al., 2020; Churchill et al., 2020). It is further argued that, the underlying instrument or “distance to the nearest bank” only influence energy poverty through financial inclusion because extant studies have employed the considered instrument and confirmed its validity using microfinance operation modalities as well as other types of financial institutions located in the rural areas (Reiter & Peprah, 2015; Churchill et al., 2020; Churchill & Marisetty, 2020; Koomson et al., 2020; Koomson & Danquah, 2021).

The findings are presented in this section in Tables 1–5. While Table 1 focuses on the full sample, the remaining tables present the findings in terms of regions (rural versus urban areas) and gender (male versus female). In essence, Table 2 and Table 3 are concerned with the rural area and urban area respectively, while Table 4 and Table 5 focus respectively on the male and female gender. Each of the tables is presented in three specifications, notably a pooled regression in the first column and the GLSS6 and GLSS7 respectively in Column 2 and Column 3.

This underlying approach to presentation of results is consistent with Koomson and Danquah (2021). Given that the main findings are already reported in the underlying study, in order to avoid duplication of information, the focus of the empirical results section is on the problem statement being examined, notably: the importance of the age of the household head in moderating the incidence of financial inclusion on energy poverty. Before delving into these specificities, it is worthwhile to articulate that consistently, the findings in the pooled data are in accordance with those from the respective survey rounds in the last-two columns. Accordingly, in the light of the problem statement, the attendant consistency in findings is in terms of: (i) financial inclusion unconditionally decreasing energy poverty and (ii) the age of the household head contributing to mitigating the favorable incidence of financial inclusion on energy poverty. In other words, as the age of the household head increases, the negative incidence of the financial inclusion on energy poverty decreases. However, in order for this finding to be properly articulated the net effects of financial inclusion as well as corresponding threshold have to be computed in the paragraphs that follow. The computation of net effect and corresponding thresholds are consistent with the assessment of Hypothesis 1 and Hypothesis 2, respectively.

In the light of the above, the estimated coefficients in this study are not interpreted as in a linear additive model as it the case with Koomson and Danquah (2021). Hence, in order to avoid pitfalls of interactive regressions documented in Brambor et al. (2006), this study computes net effects and thresholds in accordance with contemporary interactive regressions literature (Nchofoung et al., 2021, 2022; Nchofoung & Asongu, 2022a, 2022b). To put these in perspective, in the first column of Table 1, the net effect from the relevance of age of the household head in moderating the incidence of financial inclusion on energy poverty is -0.078 = ([0.002 × 46.900] + [-0.172]). In the corresponding computation, 46.900 is the average age of the household head, -0.172 is the unconditional incidence of financial inclusion on energy poverty while 0.002 is the conditional or interactive incidence of financial inclusion with the average age of the household head on energy poverty.

After establishing the net incidence of household age in moderating the effect of financial inclusion on energy poverty, it is also worthwhile to capture the dynamics of household age in terms of understanding what household age should not be exceeded in order for the favorable incidence of financial inclusion on energy poverty to be maintained. Still consistent with the example focusing on the first column of Table 1, the household age that should not be exceeded in order for financial inclusion to continuously mitigate energy poverty is 86 years (or 0.172/0.022). It follows that in order to financial inclusion to mitigate energy poverty, households’ heads should be younger than 86 years.

In the light of the above, the net effects from the role of the household head in the incidence financial inclusion on energy poverty are consistently negative across tables and specifications. The corresponding household head thresholds that are should not be exceeded for the underlying net negative incidence to be maintained are between: (i) 78.5 and 86 years for the full sample; (ii) 59.5 and 62 years for the rural sample; (iii) 78.5 and 85.5 years for the urban sample; (iv) 71 and 87 years for the male sample and (v) 54.66 and 82.5 years of female sample. In terms of policy implications, we take a minimalist perspective by recommending the lowest age in each table. Hence, the critical household head ages that should be avoided in order for the negative or favorable incidence of financial inclusion on energy poverty to be maintained are: 78.5 years for full sample, 59.5 years for the rural sample, 78.5 years to the urban sample, 71 years for the male sample and 54.66 years for female sample.

Table 1

Full Sample
Dependent variable: Energy Poverty
	(1)	(2)	(3)
Variables	Pooled	GLSS6	GLSS7
age	0.001***	0.001***	0.001*
	(0.000)	(0.000)	(0.000)
	0.029	0.034	0.021
female	-0.014**	-0.016**	-0.011
	(0.006)	(0.007)	(0.011)
	-0.016	-0.018	-0.013
married	-0.028***	-0.022***	-0.045***
	(0.006)	(0.006)	(0.011)
	-0.036	-0.028	-0.057
edu	-0.159***	-0.148***	-0.186***
	(0.005)	(0.006)	(0.009)
	-0.207	-0.193	-0.240
hhsize	0.012***	0.011***	0.013***
	(0.001)	(0.001)	(0.001)
	0.085	0.081	0.097
rural	0.193***	0.194***	0.189***
	(0.005)	(0.006)	(0.010)
	0.250	0.251	0.242
1.empstat	-0.037***	-0.025	-0.046***
	(0.012)	(0.018)	(0.017)
	-0.028	-0.017	-0.039
2.empstat	-0.131***	-0.121***	-0.131***
	(0.012)	(0.017)	(0.018)
	-0.139	-0.128	-0.142
3.empstat	0.020**	0.037**	0.003
	(0.010)	(0.015)	(0.013)
	0.025	0.045	0.003
rounds	-0.008
	(0.005)
	-0.009
FI_mpi	-0.172***	-0.177***	-0.157***
	(0.016)	(0.019)	(0.030)
	-0.215	-0.222	-0.194
finclusion_age	0.002***	0.002***	0.002***
	(0.000)	(0.000)	(0.001)
	0.144	0.143	0.144
Constant	0.766***	0.742***	0.791***
	(0.014)	(0.019)	(0.023)
Net effects	-0.078	-0.083	-0.063
Age threshold	86	88.5	78.5
Observations	22,706	16,161	6,545
R-squared	0.274	0.279	0.265
Robust standard errors in parentheses. * p < 0.01, p < 0.05, * p < 0.1 Age = Age of household head, engyptydummy = energy poverty, FI_mpi = financial inclusion, female = female-headed household, married, edu = educated, hhsize = household size, rural = rural area, 1.empstat = retired/inactive, 2.empstat = employee, 3.empstat = self-employment, finclusion_edu = financial inclusion*education. The average age of a household is 46.9.

Table 2

Rural sample
Dependent variable: Energy Poverty
	(1)	(2)	(3)
Variables	Pooled	GLSS6	GLSS7
age	0.000**	0.000***	0.000
	(0.000)	(0.000)	(0.000)
	0.020	0.028	0.006
female	-0.015***	-0.019***	-0.008
	(0.005)	(0.006)	(0.010)
	-0.031	-0.039	-0.015
married	-0.006	-0.006	-0.011
	(0.005)	(0.005)	(0.010)
	-0.015	-0.013	-0.023
edu	-0.066***	-0.057***	-0.089***
	(0.004)	(0.005)	(0.009)
	-0.146	-0.133	-0.177
hhsize	0.005***	0.005***	0.007***
	(0.000)	(0.001)	(0.001)
	0.076	0.067	0.091
1.empstat	-0.016*	-0.015	-0.013
	(0.009)	(0.015)	(0.011)
	-0.020	-0.018	-0.018
2.empstat	-0.147***	-0.113***	-0.187***
	(0.013)	(0.017)	(0.020)
	-0.216	-0.168	-0.273
3.empstat	0.012*	0.025**	0.006
	(0.007)	(0.011)	(0.009)
	0.024	0.048	0.012
rounds	-0.008**
	(0.004)
	-0.018
FI_mpi	-0.119***	-0.121***	-0.124***
	(0.017)	(0.020)	(0.031)
	-0.250	-0.265	-0.239
finclusion_age	0.002***	0.002***	0.002***
	(0.000)	(0.000)	(0.001)
	0.187	0.191	0.188
Constant	0.966***	0.949***	0.974***
	(0.010)	(0.014)	(0.017)
Net effects	-0.025	-0.027	-0.030
Rural age threshold	59.5	60.5	62.0
Observations	12,966	9,143	3,823
R-squared	0.139	0.120	0.183
Robust standard errors in parentheses. * p < 0.01, p < 0.05, * p < 0.1.Age = Age of household head, engyptydummy = energy poverty, FI_mpi = financial inclusion, female = female-headed household, married, edu = educated, hhsize = household size, rural = rural area, 1.empstat = retired/inactive, 2.empstat = employee, 3.empstat = self-employment, finclusion_edu = financial inclusion*education. The average age of a household is 46.9.

Table 3

Urban sample
Dependent variable: Energy Poverty
	(1)	(2)	(3)
Variables	Pooled	GLSS6	GLSS7
age	0.002***	0.002***	0.001*
	(0.000)	(0.000)	(0.001)
	0.057	0.065	0.043
female	-0.025**	-0.029**	-0.014
	(0.011)	(0.013)	(0.021)
	-0.025	-0.029	-0.015
married	-0.064***	-0.056***	-0.084***
	(0.011)	(0.013)	(0.021)
	-0.067	-0.058	-0.088
edu	-0.286***	-0.278***	-0.305***
	(0.009)	(0.011)	(0.018)
	-0.285	-0.274	-0.309
hhsize	0.027***	0.028***	0.027***
	(0.002)	(0.002)	(0.004)
	0.135	0.136	0.132
1.empstat	-0.079***	-0.063*	-0.105***
	(0.025)	(0.035)	(0.039)
	-0.050	-0.039	-0.069
2.empstat	-0.130***	-0.134***	-0.107***
	(0.022)	(0.032)	(0.032)
	-0.129	-0.133	-0.106
3.empstat	0.002	0.015	-0.025
	(0.021)	(0.031)	(0.029)
	0.002	0.016	-0.026
rounds	-0.015
	(0.010)
	-0.014
FI_mpi	-0.171***	-0.170***	-0.157***
	(0.028)	(0.032)	(0.056)
	-0.178	-0.178	-0.162
finclusion_age	0.002***	0.001*	0.002*
	(0.001)	(0.001)	(0.001)
	0.080	0.065	0.101
Constant	0.793***	0.771***	0.815***
	(0.029)	(0.038)	(0.048)
Net effects	-0.077	-0.123	-0.063
Urban age threshold	85.5	170	78.5
Observations	9,740	7,018	2,722
R-squared	0.185	0.192	0.172
Robust standard errors in parentheses.* p < 0.01, p < 0.05, * p < 0.1. Age = Age of household head, engyptydummy = energy poverty, FI_mpi = financial inclusion, female = female-headed household, married, edu = educated, hhsize = household size, rural = rural area, 1.empstat = retired/inactive, 2.empstat = employee, 3.empstat = self-employment, finclusion_edu = financial inclusion*education. The average age of a household is 46.9.

Table 4

Male sample
Dependent variable: Energy Poverty
	(1)	(2)	(3)
Variables	Pooled	GLSS6	GLSS7
age	0.001***	0.001***	0.001*
	(0.000)	(0.000)	(0.000)
	0.025	0.026	0.027
married	-0.021***	-0.017**	-0.036***
	(0.007)	(0.008)	(0.014)
	-0.024	-0.019	-0.040
edu	-0.135***	-0.123***	-0.163***
	(0.005)	(0.006)	(0.011)
	-0.176	-0.161	-0.210
hhsize	0.011***	0.011***	0.013***
	(0.001)	(0.001)	(0.002)
	0.088	0.082	0.102
rural	0.216***	0.216***	0.213***
	(0.007)	(0.008)	(0.013)
	0.275	0.277	0.266
1.empstat	-0.038**	-0.040*	-0.032
	(0.016)	(0.023)	(0.021)
	-0.025	-0.024	-0.025
2.empstat	-0.114***	-0.109***	-0.108***
	(0.014)	(0.020)	(0.022)
	-0.129	-0.122	-0.122
3.empstat	0.023**	0.038**	0.006
	(0.012)	(0.017)	(0.016)
	0.029	0.047	0.008
rounds	-0.013**
	(0.006)
	-0.016
FI_mpi	-0.166***	-0.174***	-0.142***
	(0.019)	(0.023)	(0.037)
	-0.205	-0.217	-0.170
finclusion_age	0.002***	0.002***	0.002***
	(0.000)	(0.000)	(0.001)
	0.124	0.129	0.111
Constant	0.737***	0.720***	0.742***
	(0.017)	(0.022)	(0.028)
Net effects	-0.072	-0.080	-0.048
Male age threshold	83.0	87.0	71.0
Observations	15,905	11,499	4,406
R-squared	0.282	0.290	0.264
Robust standard errors in parentheses. * p < 0.01, p < 0.05, * p < 0.1. Age = Age of household head, engyptydummy = energy poverty, FI_mpi = financial inclusion, female = female-headed household, married, edu = educated, hhsize = household size, rural = rural area, 1.empstat = retired/inactive, 2.empstat = employee, 3.empstat = self-employment, finclusion_edu = financial inclusion*education. The average age of a household is 46.9.

Table 5

Female Sample
Dependent variable: Energy Poverty
	(1)	(2)	(3)
Variables	Pooled	GLSS6	GLSS7
age	0.001***	0.001***	0.000
	(0.000)	(0.000)	(0.000)
	0.033	0.047	0.002
married	-0.031***	-0.020	-0.053***
	(0.010)	(0.012)	(0.018)
	-0.033	-0.022	-0.059
edu	-0.212***	-0.203***	-0.231***
	(0.011)	(0.013)	(0.019)
	-0.268	-0.255	-0.294
hhsize	0.013***	0.014***	0.011***
	(0.002)	(0.002)	(0.003)
	0.074	0.077	0.065
rural	0.151***	0.149***	0.151***
	(0.008)	(0.010)	(0.014)
	0.197	0.192	0.202
1.empstat	-0.043**	-0.017	-0.069**
	(0.020)	(0.031)	(0.027)
	-0.038	-0.015	-0.066
2.empstat	-0.182***	-0.166***	-0.191***
	(0.023)	(0.034)	(0.033)
	-0.156	-0.138	-0.176
3.empstat	0.010	0.031	-0.009
	(0.017)	(0.028)	(0.021)
	0.013	0.036	-0.011
rounds	0.005
	(0.009)
	0.006
FI_mpi	-0.165***	-0.165***	-0.164***
	(0.029)	(0.035)	(0.051)
	-0.209	-0.208	-0.213
finclusion_age	0.002***	0.002***	0.003***
	(0.001)	(0.001)	(0.001)
	0.157	0.146	0.182
Constant	0.791***	0.750***	0.861***
	(0.025)	(0.035)	(0.035)
Net effects	-0.071	-0.071	-0.023
Female age threshold	82.5	82.5	54.66
Observations	6,801	4,662	2,139
R-squared	0.269	0.267	0.278
Robust standard errors in parentheses. * p < 0.01, p < 0.05, * p < 0.1. Age = Age of household head, engyptydummy = energy poverty, FI_mpi = financial inclusion, female = female-headed household, married, edu = educated, hhsize = household size, rural = rural area, 1.empstat = retired/inactive, 2.empstat = employee, 3.empstat = self-employment, finclusion_edu = financial inclusion*education. The average age of a household is 46.9.

Concerning nexus between the findings and the extant literature on the subject, the importance of financial inclusion in reducing energy poverty is consistent with prior studies on the subject (Levaï et al., 2011; Boutabba et al., 2020; Koomson & Danquah, 2021) and by extension, the attendant literature on the importance of financial inclusion in inclusive and economic development outcomes (Sarma & Pais, 2011; Kuri & Laha, 2011 ; Sharma, 2016; Danquah et al., 2017; Li, 2018 ; Koomson & Ibrahim, 2018 ; Park & Mercado, 2018; Stein & Yannelis, 2019; Matekenya et al., 2020 ; Omar & Inaba, 2020 ). However, in considering potential channels in accordance with another strand of literature on mechanisms for development outcomes (Alesina & Zhuravskaya, 2011; Churchill & Smyth, 2020), we have shown that some household head age limit should not be exceeded for financial inclusion to favor energy poverty reduction.

The present research has extended Koomson and Danquah (2021, EP) in order to understand if and how the age of the household head matters in the incidence of financial inclusion on energy poverty in Ghana. To make the assessment, we have revisited the underlying study within a framework of interactive regressions in order to establish household head ages that should be avoided in order for financial inclusion to continuously reduce energy poverty. In the estimation exercise that is based on pooled data and two stage least squares, we establish a negative net effect from the role of household head age in moderating the incidence of financial inclusion on energy poverty. An extended analysis provides household head ages that should be avoided in order for financial inclusion to reduce energy poverty. These critical household head ages are contingent on the area of household (i.e. rural versus urban) as well as the gender of household age (i.e. male versus female). As a main policy implication, rural, urban, male and female specific household head ages should be avoided for financial inclusion to reduce energy poverty. These household age limits are respectively, 78.5 years for the full sample, 59.5 years for the rural sample, 78.5 years to the urban sample, 71 years for the male sample and 54.66 years for female sample.

In the light of the above, by providing actionable critical levels of household head ages that policy makers can act upon in order to maintain the favorable relevance of financial inclusion on energy poverty, we have contributed to both scholarly and policy literature. On the scholarly front, it is relevant to improve studies providing direct linkages between microeconomic and macroeconomic variables within the framework of interactive regressions because microeconomic and macroeconomic outcomes do not depend on isolated influences from corresponding determinants. On the policy front, contingent on the area in which the household is located (rural versus urban) as well as the gender of the household head (male versus female) policy makers have been provided with critical household head ages that should not be exceeded for financial inclusion to continuously have a favorable outcome in terms of mitigating energy poverty.

This study can be extended by considering other policy variables from which understanding of the nexus between financial inclusion and energy poverty can be improved. Accordingly, we have shown that blanket or direct linkages can be improved with policy variables within the remit of interactive regressions in order to establish specific policy thresholds that are relevant in improving the policy understanding of how energy poverty can be reduced through mechanisms such as financial inclusion. In considering the recommended future research direction, other countries and inclusive development outcomes should be engaged.

Availability of supporting data: The data for this paper are available upon request.

Funding: This research is not funded by any organization.

Competing interest: The authors have no financial and non-financial competing interests.

Acknowledgement: None.

Abdul-Mumuni, A., &Koomson, I. (2019). “Household remittance inflows and child education in Ghana: Exploring the gender and locational dimensions”. Journal of Economic Research(JER), 24(2), 197-222.
Adusah-Poku, F., & Takeuchi, K. (2019 b). “Household energy expenditure in Ghana: A double-hurdle model approach”. World Development, 117(May), pp. 266–277.
Adusah-Poku, F., & Takeuchi, K. (2019 a). Energy poverty in Ghana: any progress so far?.Renewable and Sustainable Energy Reviews, 112, 853-864.
Alesina, A., &Zhuravskaya, E. (2011). “Segregation and the Quality of Government in a Cross Section of Countries”. American Economic Review, 101(5), pp. 1872–1911.
Alkire, S., & Foster, J. (2011). “Counting and multidimensional poverty measurement”. Journal of Public Economics, 95(7–8), pp. 476–487.
Allen, F., Demirguc-Kunt, A., Klapper, L., &Peria, M. S. M. (2016). The foundations of financial inclusion: Understanding ownership and use of formal accounts. Journal of financial Intermediation, 27, 1-30.
Asongu, S. A., &Odhiambo, N. M., (2018). “ICT, Financial Access and Gender Inclusion in the Formal Economic Sector: Evidence from Africa”, African Finance Journal,20(2), pp. 46-66.
Asongu, S. A., Biekpe, N., &Cassimon, D., (2020). “ Understanding the greater diffusion of mobile money innovations in Africa”, Telecommunications Policy, 44(8), September 2020, 102000.
Asongu, S. A., Biekpe, N., &Cassimon, D., (2021). “On the diffusion of mobile phone innovations for financial inclusion”, Technology in Society, 65 (May), 101542.
Baah-Boateng, W. (2015). Unemployment in Ghana: a cross sectional analysis from demand and supply perspectives. African Journal of Economic and Management Studies.
Bicaba, Z., Brixiova, Z., &Ncube, M., (2017). “Can Extreme Poverty in Sub-Saharan Africa be Eliminated by 2030?,” Journal of African Development, 19(2), pp. 93-110.
Boardman, B. (2013). Fixing fuel poverty: Challenges and solutions. Routledge.
Boutabba, M. A., Diaw, D., Laré, A., &Lessoua, A. (2020). “The impact of microfinance on energy access: A case study from peripheral districts of Lomé, Togo”. Applied Economics, 52(45), pp. 4927-4951.
Bouzarovski, S., &Petrova, S. (2015). “A global perspective on domestic energy deprivation: Overcoming the energy poverty–fuel poverty binary”. Energy Research & Social Science, 10(November), pp. 31–40.
Brambor, T., Clark, W. M., & Golder, M. (2006). “Understanding Interaction Models:Improving Empirical Analyses”, Political Analysis, 14 (1), pp. 63-82.
Canagarajah, S., Newman, C., &Bhattamishra, R. (2001). Non-farm income, gender, and inequality: evidence from rural Ghana and Uganda. Food policy, 26(4), 405-420.
Churchill, S. A., &Marisetty, V. B. (2020). “Financial inclusion and poverty: A tale of forty-five thousand households”. Applied Economics, 52(16), pp. 1777-1788.
Churchill, S. A., & Smyth, R. (2020). “Ethnic diversity, energy poverty and the mediating role of trust: Evidence from household panel data for Australia”. Energy Economics, 86(February), 104663.
Churchill, S. A., Nuhu, A. S., & Smyth, R. (2020). Financial Inclusion and Poverty: Micro-level Evidence from Nigeria. In Moving from the Millennium to the Sustainable Development Goals (pp. 11–36). Springer.
Cook, B. G., (2014). “A call for examining replication and bias in special education research”. Remedial and Special Education, 35(4), pp. 233-246.
Crentsil, A. O., Asuman, D., & Fenny, A. P. (2019). “Assessing the determinants and drivers of multidimensional energy poverty in Ghana”. Energy Policy, 133 (October), 110884.
Crentsil, A. O., Asuman, D., & Fenny, A. P. (2019). Assessing the determinants and drivers of multidimensional energy poverty in Ghana. Energy Policy, 133, 110884.
Danquah, M., Quartey, P., &Iddrisu, A. M. (2017). “Access to financial services via rural and community banks and poverty reduction in rural households in Ghana”. Journal of African Development, 19(2), pp. 67–76.
Demirgüç-Kunt, A., &Klapper, L. (2012). Financial inclusion in Africa: An overview. The World Bank. http://documents.albankaldawli.org/curated/ar/534321468332946450/pdf/WPS608 8.pdf
Fungácová, Z., & Weill, L. (2014). Understanding financial inclusion in China. BOFIT Discussion Papers, 2014(10), 1.
Ghana Statistical Service (GSS), Ghana Health Service (GHS), and ICF International. (2015). Ghana demographic and health survey 2014. GSS, GHS and ICF International, Rockville, Maryland, USA.
GSS (2014). Ghana living standards survey: Main report. Ghana Statistical Service, Accra.http://www.statsghana.gov.gh/docfiles/glss6/GLSS6_Main%20Report.pdf
GSS (2019). Ghana living standards survey round 7 (GLSS 7): Main report. Ghana Statistical Service, Accra. (GLSS7) https://www.statsghana.gov.gh/gssmain/fileUpload/pressrelease/%20MAIN%20REPORTFINAL.pdf
Koomson, I., &Danquah, M. (2021). “Financial inclusion and energy poverty: Empirical evidence from Ghana”. Energy Economics, 94(February), 105085.
Koomson, I., & Ibrahim, M. (2018). Financial Inclusion and Growth of Non-farm Enterprises in Ghana. In Financing Sustainable Development in Africa (pp. 369–396). Springer.https://doi.org/10.1007/978-3-319-78843-2_14
Koomson, I., Villano, R. A., & Hadley, D. (2020). “Effect of Financial Inclusion on Poverty and Vulnerability to Poverty: Evidence Using a Multidimensional Measure of Financial Inclusion”. Social Indicators Research, 25(4), pp. 375–387. https://doi.org/10.1007/s11205-019-02263-0
Kuri, P. K., &Laha, A. (2011). “Financial inclusion and human development in India: An inter-state analysis”. Indian Journal of Human Development, 5(1), pp. 61–77.
Levaï, D., Rippey, P., Rhyne, E., &Allderdice, A. (2011). “Microfinance and energy poverty: Findings from the energy links project”. Center for Financial Inclusion at Accion International Publication, 13.
Li, L. (2018). “Financial inclusion and poverty: The role of relative income”. China Economic Review, 52, pp. 165–191.
Lin, B., &Okyere, M. A. (2022). Are people energy poor because of their prosocial behavior? Evidence from Ghana. Energy, 239, 122455.
Martey, E. (2022). Empirical analysis of crop diversification and energy poverty in Ghana. Energy Policy, 165, 112952.
Matekenya, W., Moyo, C., &Jeke, L. (2020). “Financial inclusion and human development: Evidence from Sub-Saharan Africa”. Development Southern Africa, 38(5), pp. 683-700.
McEwan, B., Carpenter, C. J., &Westerman, D. (2018). “On replication in communication science”,Communication Studies, 69(3), pp. 235-241.
Nchofoung, T. N., &Asongu, S. A., (2022a). “ICT for sustainable development: Global comparative evidence of globalisation thresholds”. Telecommunications Policy, 46(5), 102296
Nchofoung, T. N., &Asongu, S. A., (2022b). “Effects of infrastructures on environmental quality contingent on trade openness and governance dynamics in Africa”, Renewable Energy, 189(April), 152-163.
Nchofoung, T. N., Asongu, S. A., Kengdo, A. A. N., (2022). “Linear and non-linear effects of infrastructures on inclusive human development in Africa”, African Development Review, 34(1), pp. 81-96.
Nchofoung, T.N., Achuo, E.D. &Asongu, S.A., (2021). “Resource rents and inclusive human development in developing countries”. Resources Policy, 74(4), 102382.
Nussbaumer, P., Nerini, F. F., Onyeji, I., & Howells, M. (2013). Global insights based on the multidimensional energy poverty index (MEPI). Sustainability, 5(5), 2060–2076.
Omar, M. A., &Inaba, K. (2020). “Does financial inclusion reduce poverty and income inequality in developing countries? A panel data analysis”. Journal of Economic Structures, 9, 1–25.
Park, C.-Y., & Mercado Jr, R. V. (2018). “Financial inclusion, poverty, and income inequality in developing Asia”. Singapore Economic Review, 63(01), pp. 185–206.
Pridemore, W. A., Makel, M. C., &Plucker, J. A. (2018). “Replication in criminology and the social sciences”,Annual Review of Criminology, 1, pp. 19-38.
Reiter, B., &Peprah, J. A. (2015). “Assessing African Microfinance: An Exploratory Case Study of Ghana’s Central Region”. Journal of International Development, 27(7), pp. 1337–1342.
Sarma, M., &Pais, J. (2011). “Financial inclusion and development”. Journal of International Development, 23(5), pp. 613–628.
Sharma, D. (2016). “Nexus between financial inclusion and economic growth: Evidence from the emerging Indian economy”. Journal of Financial Economic Policy, 8(1), pp. 13–36.
Stein, L. C., &Yannelis, C. (2019). “Financial inclusion, human capital, and wealth accumulation: Evidence from the freedman’s savings bank”. The Review of Financial Studies, 33(11), pp. 5333–5377.
UNCD (2022). “Financial Inclusion and SDGs”, United Nations Capital Development Fund. https://www.uncdf.org/financial-inclusion-and-the-sdgs (Accessed: 22/03/2022).
Zhang, Q., &Posso, A. (2019). “Thinking inside the Box: A Closer Look at Financial Inclusion and Household Income”. The Journal of Development Studies, 55(7), pp. 1616-1631.

No competing interests reported.

Appendices.docx

How the age of the head of households matter in the incidence of financial inclusion on energy poverty in Ghana

Status:

Version 1

Abstract

1. Introduction

2. Literature Review And Testable Hypotheses

3. Data And Methodology

3.1 Data

3.1 Energy Poverty

3.2 Financial inclusion (FI)

3.2. Methodology

4. Empirical Results

5. Concluding Implications And Future Research Directions

6. Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1