Remittance Inflows and Energy Transition of the Residential Sector in Developing Countries

Abstract

The energy transition is crucial for the United Nations’ Sustainable Development Goal 7 (affordable and clean energy). As remittances account for a significant share of household incomes in developing countries, they may be associated with the energy transition from low-efficiency residential fuels (e.g., coal and wood) to high-efficiency residential fuels (e.g., gas and electricity). This study examines the association between remittances and residential energy transition in developing countries by employing a pooled mean group autoregressive distributed lag (PMG-ARDL) model for 27 developing nations from 1995 to 2018. The results indicate that a 1% increase in remittances (ratio to GDP) is associated with a 0.24% increase in the share of high-efficiency energy sources in residential energy consumption in the long run.

1. Introduction

The energy ladder hypothesis suggests that households shift from the use of dirty and low-efficiency residential energy sources (e.g., coal, wood, and animal waste) toward clean and high-efficiency residential energy sources (e.g., kerosene, electricity, and gas) as their economic conditions improve [1,2]. However, energy transition is not smooth in developing nations for several reasons, such as low income and income uncertainty [3], lack of modern fuel availability [4,5], the higher upfront cost of modern energy appliances [5,6], and sociocultural preferences for conventional fuels [7,8]. Several studies have emphasized the role of stable and higher income in enabling households to shift from low- to high-efficiency residential energy sources [1,2]. Recently, many developing countries have recorded higher levels of remittance inflows [9,10,11], and research has revealed that remittances are associated with various socioeconomic transformations [9,10,12,13,14], including the energy transition process [15,16,17]. As the use of conventional fuels contributes to adverse economic, health, and environmental consequences in developing countries [18,19,20], this calls for an examination of how remittance inflows relate to energy transition in the residential sector. To this end, this paper aims to study how remittances are associated with the residential energy transition from low- to high-efficiency energy sources in the long-run in developing countries.

The energy ladder argument is a widely recognized theoretical framework for explaining the dynamics of the energy-switching behavior of households from low- to high-efficiency residential energy sources; this shift is primarily driven by changes in economic conditions [1,2]. Residential consumers have an inherent rank of preference for fuels based on their attributes, such as cleanliness, efficiency, safety, and ease of use, and generally prefer the most sophisticated fuels affordable, subject to their income level [21]. Thus, households often move up and down the energy ladder by adjusting their energy portfolios as their economic status changes [22,23,24]. In addition, households’ energy-switching behavior also depends on various socioeconomic factors, including urbanization, education, and access to modern energy [25,26,27].

In recent years, remittance inflows to developing countries have witnessed impressive growth, rising from USD 68 billion in 1990 to USD 553 billion in 2015 [10], currently accounting for 30–40% of household incomes in developing countries [28]. Remittance inflows have become a crucial driver of economic transition in developing countries by reducing poverty and inequality [10,14], improving health and education [9], promoting financial development and access to credit [12,29], supporting technology diffusion [13,30], and providing insurance against income volatility and uncertainty [31,32]. Although past studies generally suggest that remittances are associated with favorable socioeconomic transformation in developing countries, there exist some adverse effects of remittances, such as the Dutch Disease effect [33] and an increase in CO₂ emissions [34], as well. As remittances enable households in developing countries to improve their economic status through income stability, access to credit, and technological diffusion, an association may exist between remittances and the energy transition from low- to high-efficiency residential fuels, as implied by the energy ladder hypothesis.

This study adds to the existing literature by exploring the long-run association between remittance inflows and the share of high-efficiency energy sources in residential energy consumption in developing countries. We utilized country-level panel data on residential energy consumption for 27 developing nations from 1995 to 2018. The existing literature primarily focuses on household-level data covering a specific region or country (Manning and Taylor [15], for rural Mexico; Martey [16], for Ghana; Taylor et al. [17], for Guatemala). However, to our best knowledge, no empirical study has examined the linkage between remittance inflows and the transition of the residential sector toward high-efficiency residential energy sources at the national level, with comprehensive coverage of developing countries. Our study contributes to the literature by confirming the energy ladder hypothesis at the national level and the links of the residential energy transition process with socioeconomic factors, such as real GDP per capita, remittance inflows, urbanization, dependent population, and agricultural practice in developing countries. As remittance inflows have become a significant source of income for the residents of developing countries, the long-run linkage between remittance inflows and the transition toward high-efficiency residential energy sources helps us further understand the energy transition dynamics in the residential sector of developing countries. Understanding the role of remittance inflows in the energy transition dynamics of developing countries could help regulators formulate effective energy policies to facilitate clean and high-efficiency energy use in the residential sector.

This study employs a panel autoregressive distributed lag (ARDL) model to ascertain the nexus between remittance inflows and the share of high-efficiency energy sources in residential energy consumption in developing countries. Our results reveal that the remittance inflows are positively associated with the share of high-efficiency residential energy sources in residential energy consumption in the long run. In particular, the estimated coefficient suggests that a 1% increase in the remittance inflow (ratio to GDP) is associated with a 0.24% increase in the share of high-efficiency energy sources in residential energy consumption. There are three plausible explanations for this association. First, remittances provide stable income to families and protect them against income volatility and shortfalls [32,35], which could ultimately assist households in transitioning to and sustaining the use of high-efficiency residential fuels. Second, remittances facilitate financial development and access to credit [12,29], which can help households afford expensive modern energy appliances. Third, remittances through international migration foster technology diffusion [13,30] and educational development [36,37], which could contribute to the adoption of high-efficiency residential energy sources.

Given that many households in developing countries often suffer from pollution, low productivity, and health hazards associated with the use of dirty conventional fuels [18,19,20] and rely on remittance income from abroad [9,10,11,28], our results confirm that remittances are an important factor in understanding energy transition dynamics in the residential sector of developing countries. As international communities and governments emphasize the need for universal access to clean and modern energy sources in developing countries [20], this study could provide insight into the dynamics of residential energy transition in developing countries.

The remainder of this paper is organized as follows. Section 2 provides a literature review of access to clean and modern energy, energy transitions, and remittance inflows. Section 3 presents details of the methodology and data used in this study. Section 4 presents the results of the empirical analysis and their implications. Finally, Section 5 concludes with policy suggestions and the limitations of our analysis.

2. Literature Review

2.1. Access to Clean and Modern Energy

Millions of poor households use hazardous conventional energy sources to satisfy their daily energy requirements. As of 2019, one-third of the world’s population still lacks access to modern cooking technologies, and 759 million people do not have access to electricity [20]. Using inefficient conventional fuels results in several adverse economic, health, and environmental consequences [18,19,20]. For example, approximately 2.5 million people die prematurely from pollution-related illnesses caused by burning dirtier fuels for cooking and heating [38]. In addition, the collection and use of conventional fuels take considerable time that inhibits individuals from participating in other earning activities and reduces productivity [39,40]. In particular, women and children are the most vulnerable groups as they spend more time on fuel collection and staying home with indoor pollution [41,42,43].

Universal access to clean and modern energy is vital for the inclusive and sustainable development of the world. The United Nations has planned 17 interlinked global goals—sustainable development goals (SDGs)—to be achieved by 2030, of which affordable and clean energy is one. Affordable and clean energy goals are linked to several SDGs, such as health and well-being (SDG 3), affordable and clean energy (SDG 7), and sustainable cities and communities (SDG 11). The United Nations aims to ensure universal access to affordable, reliable, and modern energy services; substantially increase the share of renewable energy in the global energy mix; and double the global energy efficiency rate by 2030. According to a recent report by the United Nations [20], the global electricity access rate improved from 83% in 2010 to 90% in 2019. In addition, the share of the global population with clean cooking fuels and technologies increased from 57% in 2010 to 66% in 2019. Despite such progress, universal access to clean and modern energy remains challenging. One-third of the global population still lacks access to clean cooking fuels and technologies, and 759 million people lack access to electricity. Among them, Sub-Saharan Africa alone constitutes 85% of its population using inefficient and dangerous cooking systems and 75% without electricity. Currently, 660 million people remain without electricity, and one-third of the global population may still be without clean cooking technologies in 2030.

2.2. Energy Transition

Several studies have examined the energy ladder hypothesis to explain households’ transition from dirty and low-efficiency residential fuels toward clean and high-efficiency residential fuels in developing and emerging countries. Hosier and Dowd [1] and Leach [2] pioneered the energy ladder hypothesis suggesting that households tend to shift to high-efficiency and sophisticated energy sources as their income level increases (see Figure 1). This hypothesis is associated with three main observations regarding the linkages between development and energy transition. First, several studies have found a positive association between economic growth and modern energy uptake [1]. Second, transitions occur where access to modern fuels has been improved due to the development and urbanization processes [44]. Third, the rate of energy transition varies across income groups and between urban and rural populations [2].

In the developing world, households often do not completely switch from low- to high-efficiency residential energy sources as their incomes increase. Instead, they rearrange their energy portfolios by using low- and high-efficiency residential fuels. Heltberg [23] studied eight selected developing countries and found that households, particularly in rural areas, continue using conventional residential fuels even after they have started using modern residential fuels. Similarly, Kroon et al. [24] conducted a meta-analysis of existing empirical studies on fuel-switching behavior and concluded that the displacement of conventional fuels does not occur linearly at higher income levels. Instead, households continue to use an energy portfolio consisting of conventional and modern residential fuels to deal with income volatility, safeguard themselves from market uncertainties, and maintain cultural practices while benefitting from the use of modern residential fuels. In contrast, Hanna and Oliva [22], using field experiment data from India, showed that households might even purchase more conventional fuels with a rise in income due to the wealth effect, so that the substitution effect may not dominate the wealth effect unless conventional fuels are an inferior product.

In addition to economic well-being, multiple factors such as the availability of modern energy, education, urbanization, and social and cultural preferences play an essential role in households’ fuel transition [25,26,27]. However, household income level is the pivotal driver for their fuel transition. Many empirical studies recognize income level as a crucial factor in explaining energy transition in the residential sector [1,46,47,48].

2.3. Remittance Inflows and the Energy Transition

Remittance inflows are a source of stable income for households in developing countries [31,32,35]. Consequently, remittance inflows have a significant impact on recipient countries in several contexts, including poverty and inequality [10,14,49], human capital formation [9,50,51], and growth and structure of the economy [52,53,54]. In addition, several studies have confirmed that remittance inflows influence energy consumption in developing countries. For example, Rahman et al. [55] and Zaman et al. [56] discovered a positive long-run association between remittance inflow and energy consumption in selected thirteen high-remittance-receiving countries. Manning and Taylor [15] investigated the impact of remittance inflow of Mexican migrants into the United States at the household level in Mexico and found that remittances reduce households’ reliance on firewood collection and increase the use of stove and gas purchases. Their findings imply that remittance inflows are an essential element in the energy consumption transition in developing countries.

Although several studies show a positive linkage between remittances and socioeconomic well-being in the developing world, there are few studies on the association between remittances and energy transitions in the residential sector. Most studies have examined this association at the micro-level [7,15,16]. However, these studies did not investigate how remittance inflows are associated with high-efficiency residential energy uptake by households. To our best knowledge, there are no empirical studies on the remittance–energy transition nexus at the national level covering a broad range of developing countries. As developing countries are significantly behind in the adoption of modern residential energy technologies, these countries have received priority from the international community in meeting the universal access to clean and modern energy sources goal by 2030 [20]. Thus, this study contributes to the literature by examining the long-run relationship between remittance inflows and the adoption of high-efficiency residential energy sources in the residential sector of developing countries.

3. Methodology and Data

3.1. Data

This study uses panel data from 27 developing countries from 1995 to 2018 to examine the nexus between remittance inflows and the share of high-efficiency energy sources in residential energy consumption in developing countries. Our motivation to explore the association between remittances and high-efficiency residential energy uptake is driven mainly by two reasons. First, a large population in developing countries still lacks access to clean and modern energy sources, such as modern cooking technologies and electricity, leading to a range of economic, health, and environmental problems [18,19,20]. Energy transition in developing countries is critical to achieving the United Nations’ sustainable development goal of ensuring universal access to affordable and clean energy. Second, many developing countries receive large remittances from abroad [9,10], and past studies have shown that remittances are associated with high-efficiency residential energy uptake in developing countries [15,16,17].

We follow the World Bank’s country classification and choose low- and lower-middle-income countries as developing countries. Recent reports by the United Nations [20] and the International Energy Agency [38] have suggested that Sub-Saharan African and South Asian countries are significantly lagging in adopting clean and modern residential energy. Thus, most of the countries in our sample are from Sub-Saharan Africa and South Asia. Furthermore, to capture the long-run relationship between remittances and energy transitions, we include large and small remittance recipient countries in the sample. The selection of sample countries is subject to data availability.

The dataset comprises the energy consumption in the residential sector, personal remittance inflow, real GDP per capita, urban population, oil price, dependent (young and old) population, and agricultural land area variables. Energy consumption in the residential sector and oil price series were obtained from the World Energy Balances database of the International Energy Agency and the US Energy Information Administration (EIA), respectively. The data for other variables were taken from the World Bank’s World Development Indicators (WDI).

Table 1. High-efficiency residential energy consumption, personal remittances inflows, and GDP per capita statistics of the sample countries.

	High-Efficiency Residential Energy Consumption (Ratio to Residential Energy Consumption)			Personal Remittances, Received (Ratio to GDP)			GDP Per Capita (Constant 2010 USD)
	1995	2006	2018	1995	2006	2018	1995	2006	2018
East Asia and Pacific	0.12	0.22	0.30	0.04	0.07	0.08	954	1314	2197
Cambodia	0.03	0.09	0.11	0.00	0.03	0.06	342	669	1203
Philippines	0.22	0.35	0.49	0.07	0.12	0.10	1565	1959	3191
Latin America and the Caribbean	0.18	0.33	0.48	0.04	0.15	0.14	1675	1989	2536
Bolivia	0.39	0.64	0.94	0.00	0.05	0.03	1493	1751	2560
El Salvador	0.17	0.47	0.69	0.12	0.22	0.21	2529	2905	3507
Honduras	0.09	0.13	0.16	0.02	0.22	0.20	1575	1852	2219
Nicaragua	0.07	0.09	0.15	0.02	0.10	0.12	1104	1447	1857
Middle East and North Africa	0.76	0.76	0.84	0.04	0.04	0.05	2264	3177	3858
Algeria	1.00	1.00	1.00	0.03	0.00	0.01	3240	4298	4754
Egypt, Arab Rep.	0.81	0.85	0.90	0.05	0.05	0.10	1659	2255	2909
Morocco	0.70	0.67	0.87	0.05	0.08	0.06	1727	2513	3361
Tunisia	0.52	0.50	0.61	0.04	0.04	0.05	2429	3640	4408
South Asia	0.11	0.17	0.26	0.03	0.08	0.10	761	1103	1850
Bangladesh	0.14	0.25	0.39	0.03	0.08	0.06	460	650	1203
India	0.14	0.22	0.32	0.02	0.03	0.03	675	1107	2086
Nepal	0.03	0.03	0.05	0.01	0.16	0.28	400	511	818
Pakistan	0.20	0.24	0.28	0.03	0.04	0.07	804	970	1198
Sri Lanka	0.05	0.10	0.24	0.06	0.08	0.08	1469	2276	3946
Sub-Saharan Africa	0.05	0.08	0.09	0.01	0.03	0.03	801	975	1328
Benin	0.04	0.23	0.04	0.05	0.03	0.01	863	999	1211
Cameroon	0.06	0.04	0.06	0.00	0.01	0.01	1055	1251	1502
Cote d’Ivoire	0.07	0.07	0.13	0.01	0.01	0.01	1347	1208	1668
Ethiopia	0.01	0.01	0.01	0.00	0.01	0.01	184	252	571
Ghana	0.08	0.15	0.27	0.00	0.01	0.05	874	1117	1808
Kenya	0.05	0.04	0.05	0.01	0.02	0.03	848	886	1201
Mozambique	0.01	0.01	0.11	0.02	0.01	0.02	216	402	593
Nigeria	0.03	0.02	0.02	0.01	0.07	0.06	1342	1910	2383
Senegal	0.13	0.22	0.20	0.02	0.08	0.10	1028	1235	1547
Sudan	0.03	0.06	0.15	0.03	0.02	0.01	850	1261	1812
Tanzania	0.03	0.03	0.03	0.00	0.00	0.01	479	660	959
Togo	0.05	0.07	0.05	0.01	0.10	0.08	526	517	677
Overall Sample	0.19	0.24	0.31	0.03	0.06	0.07	1151	1500	2043

lists the countries in the sample with their respective statistics on high-efficiency residential energy consumption, personal remittances (ratio to GDP), and GDP per capita. Statistics show that Sub-Saharan Africa is the most backward region in adopting high-efficiency residential energy in the residential sector. In our sample, less than 10% of the residential energy consumption came from low-efficiency residential energy sources in Sub-Saharan African countries. In contrast, in the Middle East and North African countries, more than 80% of residential energy consumption comes from high-efficiency residential energy sources. Similarly, Figure 2 depicts the region-wise trends of high-efficiency residential energy consumption and personal remittance inflows in developing countries. On average, all regions witnessed an increase in the share of high-efficiency residential energy sources in the total residential energy consumption and personal remittance inflows.

This study investigates how remittance inflows are associated with the households’ fuel-switching behavior in developing countries. According to the energy ladder hypothesis, households transition from inefficient and dirty fuel sources (such as wood, agricultural residues, animal waste, and charcoal) to efficient and cleaner fuel sources (such as liquefied petroleum gas (LPG), solar, and electricity) as their income level rises [2]. Several studies categorize residential energy consumption into solid and non-solid fuels, in which they regard solid as dirty and low-efficiency and non-solid as clean and high-efficiency residential fuel sources, to examine households’ fuel-switching behavior [57,58,59]. Many studies discuss the roles of clean energy in the context of renewable energy sources [60,61,62]. Differently from these studies, our study focuses on the empirical validity of the energy ladder argument by using the classification of solid and non-solid fuels to divide residential fuels into dirty and clean energy sources. Following this categorization, we divide residential energy sources into two broad groups: the first group (“low-efficiency” residential energy sources) consists of solid fuels, such as coal, biofuels, and waste, while the second group (“high-efficiency” residential energy sources) consists of non-solid fuels, such as oil products, natural gas, solar, wind, and electricity. Our categorization of residential energy sources is based on end-users’ perspectives for cleanliness, efficiency, and comfort. We obtained the residential consumption of solid and non-solid fuels from the World Energy Balances database. The World Energy Balances database is a proprietary database from the International Energy Agency. From this database, we obtained residential energy consumption on coal and coal products, biofuels and waste, oil products, natural gas, solar/wind/other, and electricity variables. The technical details on these variables are available in the database documentation that can be found online at https://www.iea.org/subscribe-to-data-services/worldenergy-balances-and-statistics (accessed on 16 August 2022).

Table 2. Descriptive statistics of the variables employed.

	High-Efficiency Residential Energy Consumption (Ratio to Residential Energy Consumption)	Personal Remittances, Received (Ratio to GDP)	Log of GDP Per Capita	Urban Population (Ratio to Total Population)	Log of Real Oil Price	Log of Dependent Population (Young and Old)	Agricultural Land (Ratio to Total Land Area)
	RES	REM	Y	URB	OP	DEP	AGR
Countries	27	27	27	27	27	27	27
Obs	648	648	648	648	635	648	632
Mean	0.25	0.05	7.13	0.40	3.80	16.30	0.45
Max	1.00	0.31	8.48	0.73	6.34	19.92	0.76
Min	0.01	0.00	5.21	0.11	0.72	14.53	0.03
Std. Dev	0.28	0.05	0.69	0.16	0.92	1.23	0.18
Average values in years
1995	0.19	0.03	6.83	0.36	2.39	16.13	0.43
2000	0.23	0.03	6.94	0.38	3.11	16.21	0.44
2005	0.24	0.06	7.07	0.40	3.81	16.28	0.45
2010	0.26	0.06	7.21	0.42	4.51	16.34	0.46
2015	0.28	0.07	7.38	0.44	4.25	16.41	0.46
2018	0.31	0.07	7.46	0.45	4.65	16.46	0.46

provides statistical descriptions of the variables employed in the current study. The average share of high-efficiency energy sources in residential energy consumption is 25%, with a standard deviation of 28%. The average personal remittance inflows account for 5% of the GDP, ranging from 0.05% to 31% in our sample countries. The high-efficiency energy share in residential energy consumption and the personal remittance inflow to GDP ratio grew steadily over the sample period.

Table 3. Correlation matrix of the variables employed.

	RES	REM	Y	URB	OP	DEP	AGR
RES	1.00
REM	0.07	1.00
Y	0.67	0.23	1.00
URB	0.63	0.13	0.74	1.00
OP	0.10	0.25	0.25	0.19	1.00
DEP	−0.03	−0.24	−0.18	−0.38	0.02	1.00
AGR	−0.19	0.08	0.02	0.10	0.03	0.10	1.00

presents the correlation estimates of the variables. Consistent with the energy ladder hypothesis, the simple correlation analysis showed a positive correlation between the share of high-efficiency energy sources in residential energy consumption and real GDP per capita.

3.2. Model Specification

The PMG-ARDL model is popular and widely used in several fields, such as environmental and energy studies, to establish the long-run relationship between the variables of interest. We used a PMG-ARDL model to identify the long-run relationship between remittance inflows and energy transition in the residential sector of developing countries. To do so, we considered the following model specification,

$$RE{S}_{it}={\beta }_0+{\beta }_{1}RE{M}_{it}+{\displaystyle \sum }_k{\beta }_k{X}_{k,it}+u_{it}$$

(1)

where $RE{S}_{it}$is the share of high-efficiency energy sources in residential energy consumption, $RE{M}_{it}$ is the personal remittance inflow (ratio to GDP), $X_{k,it}$ are other control variables, and $u_{it}$ is the error term. Other control variables include the log of real GDP per capita and urban population (ratio to total population). Although both remittances and real GDP per capita are an indicator of income level for developing countries, there are some inherent differences between them. For example, real GDP per capita captures a country’s overall income or development level. Unlike real GDP per capita, remittances are counter-cyclical to the economic situation, increasing in times of crisis or economic disturbance in recipient countries [63]. In addition, remittance is a stable source of personal income and provides insurance against income volatility and shortfalls in times of economic crisis [32,35,63]. Provided these distinctions, the long-run association between remittance and residential energy uptake in developing countries may follow a different path from that between real GDP per capita and residential energy uptake. Therefore, we include both remittances and real GDP per capita as independent variables in our analysis. The selection of other control variables is based on existing literature [2,24,64], where income level and urbanization play a crucial role in determining fuel-switching behavior in the residential sector.

A panel ARDL model is used to estimate the long-run association between remittance inflows and energy mix in the residential sector. The panel ARDL model is preferred over other dynamic panel models for ascertaining the long-run dynamics among variables. Dynamic panel models, such as the generalized method of moments and fixed effects estimation, may suffer from inconsistent estimations [65]. This model allows the estimation of the long-run linkage among variables regardless of whether the variables are I(0), I(1), or a combination of the two [66,67] and produces unbiased estimates even when some regressors are endogenous [68,69]. Additionally, the model is suitable for a small sample size [70].

Equation (1) takes the following error-correction form for the panel ARDL model specification,

$$\mathsf{\Delta }RE{S}_{it}={\phi }_{i}EC{T}_{it}+{{\displaystyle \sum }}_{j=1}^{p-1}{\alpha }_{ij}^{\ast }\mathsf{\Delta }RE{S}_{it-j}+{{\displaystyle \sum }}_{j=0}^{q-1}\mathsf{\Delta }{Z}_{it-j}{\beta }_{ij}^{\ast }+{\epsilon }_{it}$$

(2)

$$EC{T}_{it}=RE{S}_{it-1}-Z_{it}{\mathsf{\theta }}_i$$

(3)

where Δ is a difference operator, $RE{S}_{it}$ is the share of high-efficiency energy sources in residential energy consumption, $Z_{it}$ is a set of explanatory variables (personal remittance inflows (ratio to GDP), real GDP per capita expressed in natural logarithms, urban population (ratio to total population)), $EC{T}_{it}$ is an error-correction coefficient, and ${\epsilon }_{it}$ is residuals. The term ${\phi }_i=-(1-{\displaystyle \sum }_{j=1}^p{\alpha }_{ij})$ represents the convergence speed of the model to the long-run equilibrium. The coefficient ${\phi }_i$ must bear a statistically significant negative sign (i.e., ${\phi }_i<0$) for the system to converge toward the long-run equilibrium. The long-run coefficients are given by ${\mathsf{\theta }}_i=-\frac{{{\displaystyle \sum }}_{j=0}^q{\mathsf{\beta }}_{ij}}{{\phi }_i}$, whereas the short-run coefficients are given by ${\alpha }_{ij}^{*}=-{\displaystyle \sum }_{d=j+1}^p{\alpha }_{i,d}$ and ${\mathsf{\beta }}_{ij}^{*}=-{\displaystyle \sum }_{d=j+1}^p{\mathsf{\beta }}_{i,d}$.

The model estimates can be derived using either the mean group (MG) or the panel mean group (PMG) estimator [66,71]. In the MG estimation, all parameters are allowed to vary across countries. In contrast, there is a homogeneity restriction on the long-run estimates, but error variances and short-run coefficients are allowed to differ across countries in the PMG estimation. The Hausman test was performed to determine the suitability of the MG or PMG estimations for our model.

4. Results

4.1. Panel Unit Root Tests

The first step in the panel ARDL model estimation begins with stationarity tests of the variables. Panel ARDL models require that each variable be either I(0) or I(1) [66,67]. For this, we can use first-generation panel unit root tests, such as Breitung [72], Hadri [73], Im et al. [74], and Levin et al. [75], and second-generation panel unit root tests, such as Pesaran [76]. First-generation tests are based on the assumption that cross-sectional units are error independent. This assumption will likely produce biases and inconsistencies during unit root tests if the variables are cross-sectionally dependent [77,78]. Second-generation panel unit root tests, on the other hand, aim to overcome the drawbacks of the cross-sectional independence assumption in first-generation tests. We used EViews (version 12) and Stata (version 16) to conduct empirical analysis.

We ascertained the cross-sectional dependence (CD) presence in the variables by performing four tests: the Lagrange multiplier (LM) test by Breusch and Pagan [79], scaled LM and CD tests by Pesaran [80], and bias-corrected scaled LM test by Baltagi et al. [81]. The test statistics in

Table 4. Cross-section dependence tests of the variables employed.

Variables		Breusch and Pagan [79] LM	Pesaran [80] Scaled LM	Baltagi et al. [81] Bias-Corrected Scaled LM	Pesaran [80] CD
Share of high-efficiency energy sources in the residential energy consumption	RES	2809.35 ***	92.78 ***	92.20 ***	30.68 ***
Personal remittances, received (ratio to GDP)	REM	2765.51 ***	91.13 ***	90.54 ***	23.94 ***
Log of real GDP per capita	Y	6852.54 ***	245.38 ***	244.80 ***	80.04 ***
Urban population (ratio to total population)	URB	6673.73 ***	238.64 ***	238.05 ***	68.72 ***
Log of real oil price	OP	7659.51 ***	275.84 ***	275.26 ***	87.27 ***
Log of dependent population (young and old)	DEP	5408.45 ***	190.8 8***	190.29 ***	55.95 ***
Agricultural land (ratio to total land area)	AGR	3898.91 ***	140.18 ***	139.62 ***	22.39 ***

Note: null hypothesis: no cross-section dependence (correlation). Significance levels: 1% (***).

indicate that all variables are cross-sectionally dependent at the 1% significance level. These test statistics suggest that the second-generation panel unit root tests were appropriate for our analysis.

Once we established that each variable is cross-sectionally dependent, we used the cross-sectionally augmented IPS (CIPS) test by Pesaran [76], which permits cross-sectional dependence and heterogeneity in the panel data. The CIPS test extends the standard augmented Dickey–Fuller (ADF) test, which incorporates the lagged level cross-section averages and the individual series first differences. The test statistics are shown in

Table 5. Unit root tests of the variables employed.

Variables		Pesaran [76]
		CIPS t-Stat (Level)	CIPS t-Stat (Difference)
Share of high-efficiency energy sources in the residential energy consumption	RES	−2.92 ***	−3.99 ***
Personal remittances, received (ratio to GDP)	REM	−2.70 **	−3.94 ***
Log (real GDP per capita)	Y	−1.73	−3.00 ***
Urban population (ratio to total population)	URB	−3.22 ***	−3.98 ***
Log (real oil price)	OP	−2.46	−2.98 **
Log (dependent population—young and old)	DEP	−1.54	−4.94 ***
Agricultural land (ratio to total land area)	AGR	−2.70 **	−3.44 ***
Null hypothesis: the series is non-stationary.

Note: the tests are performed with a constant and trend deterministics. Significance levels: 1% (***) and 5% (**).

, indicating that the variables are either I(0) or I(1) at the 5% significance level.

4.2. Panel Cointegration Test

After the unit root tests, we proceeded with the cointegration test. The presence of cointegration among variables is essential for justifying the use of an ARDL model to examine the long-run nexus among them. First, we conducted residual-based cointegration tests by Pedroni [82,83]. This test produces several test statistics for testing the null hypothesis that there is no cointegration among the variables, as presented in

Table 6. Cointegration tests—Pedroni [82,83].

	Statistics
Within dimension (panel statistic—weighted)
Panel v	−4.69
Panel rho	2.92
Panel PP	−3.99 ***
Panel ADF	−5.42 ***
Between dimension (group statistics)
Group rho	4.35
Group PP	−8.63 ***
Group ADF	−7.05 ***

Note: null hypothesis: no cointegration. Significance levels: 1% (***).

. Most test statistics are significant at the 1% level, suggesting a cointegrating relationship among the variables. Second, we performed the structural dynamics-based Westerlund [84] cointegration test, addressing the cross-sectional dependency issue. The Westerlund cointegration test reports group mean statistics, G_𝜏 and G_α, and panel statistics, P_𝜏 and P_α. The purpose of the group mean statistics is to determine if at least one cross-section is cointegrated, while the purpose of the panel statistics is to determine if the variables are cointegrated at the overall panel level. The test statistics are presented in

Table 7. Cointegration tests—Westerlund [84].

	Value	Z-Value	p-Value
Group mean statistics
G𝜏	−3.3 ***	5.89 ***	0.00
Gα	−2.76	6.06	1.00
Panel statistics
P𝜏	−13.3 ***	3.14 ***	0.00
Pα	−2.72	3.76	1.00

Note: null hypothesis: no cointegration. Significance levels: 1% (***).

. G_𝜏 and P_𝜏 test statistics are significant at the 1% level, indicating that a cointegrating relationship exists among variables.

4.3. ARDL Estimation

We first conducted the Hausman test to determine the suitability of the MG or PMG estimation for our model. The statistically insignificant Hausman test statistic at the 10% significance level (see

Table 8. PMG-ARDL estimates.

Dependent Variable: Share of High-Efficiency Energy Sources in the Residential Energy Consumption
(A) Long-run estimates
Personal remittances, received (ratio to GDP)	REM	0.244 ***
		(0.082)
Log (real GDP per capita)	Y	0.091 ***
		(0.009)
Urban population (ratio to total population)	URB	0.573 ***
		(0.128)
(B) Short-run estimates
Error correction term	ECT	−0.227 ***
		(0.049)
Personal remittances, received (ratio to GDP)	REM	0.047
		(0.111)
Log (real GDP per capita)	Y	−0.038
		(0.043)
Urban population (ratio to total population)	URB	−1.966
		(3.531)
Intercept	C	−0.13 ***
		(0.032)
Hausman test statistics (p-value shown in parentheses)		1.29 (0.73)
No. of countries		27
No. of obs		621

Significance levels: 1% (***).

) indicates that we should use PMG estimation. The PMG-ARDL results are listed in Table 8. Table 8 shows that the error correction term is negative and statistically significant at the 1% level, indicating that the system converges to the long-run equilibrium. As we analyzed the long-run coefficients, we observed positive long-run relationships between high-efficiency residential energy consumption and each of the variables (personal remittances, real GDP per capita, and urban population). The positive long-run relationship between high-efficiency residential energy consumption and personal remittances is consistent with the findings of previous studies [2,15,25,85] and the trends observed in Table 1 and Figure 2, where the East Asia and Pacific, Latin America and the Caribbean, Middle East and North Africa, South Asia, and Sub-Saharan Africa regions exhibit an increasing trend of high-efficiency residential energy consumption and remittance inflows for developing countries in the sample period of 1995 to 2018. The long-run coefficient reveals that a 1% increase in personal remittances (ratio to GDP) is associated with a 0.24% increase in the share of high-efficiency energy sources in residential energy consumption. Concerning the estimated short-run coefficients of the ARDL model, the coefficients fail to support the existence of any short-run association between high-efficiency residential energy consumption and remittance inflow, that is, the current change in the share of high-efficiency energy sources in residential energy consumption ($\mathsf{\Delta }RE{S}_{it}$) is not associated with the previous year’s change in remittance inflow ($\mathsf{\Delta }RE{M}_{it-1}$). However, the estimated coefficient of the error correction term is significantly negative at −0.227, indicating that the share of high-efficiency energy sources in the residential energy consumption adjusts to its long-run equilibrium given short-run disturbances, with 22.7% of the adjustment occurring in the first year.

4.4. Discussion

Many people have emigrated from developing countries to find jobs, pursue education, and escape adverse situations in their homeland, such as war, political unrest, and natural disasters [11,86]. These people often send their earnings back home to support their families, inducing a large influx of remittances into developing countries. Currently, remittance inflows account for a large share of residential income in developing countries [9,10,11,28]. According to the United Nations [87], remittances constitute more than three times the amount of official development assistance and foreign direct investment combined in developing countries. About half of these remittances flow directly into rural areas supporting low-income families, where three-quarters are used for essentials such as food, medicine, and education, and the rest are either saved or invested in income-generating activities. In addition, remittances have helped developing countries develop education, entrepreneurship, investment, credit access, and social protection and ultimately reduce poverty and inequality [9,10,12,13,14,29,30,31,32]. However, the association between households’ adoption of high-efficiency residential fuels and remittance inflows has not yet been extensively studied at the national level.

The positive long-run relationship between remittance inflows and the share of high-efficiency energy sources in residential energy consumption suggests that remittances can be an essential driver of high-efficiency residential energy uptake in developing countries. One possible reason for this positive association might be that remittance inflows provide a stable source of income for households. In developing nations, households tend to lack sound social insurance programs, liquid forms of savings, and access to credit, which inhibits them from escaping poverty and having better living standards by adopting high-efficiency modern technologies [35]. In this context, remittance inflows have become an important income source for households and provide insurance against income volatility and shortfalls [32,35], allowing households to stabilize income over an extended period and sustain the energy transition process.

Moreover, two additional factors are plausible for the positive association between remittances and energy transition. First, residential energy appliances for high-efficiency fuels require considerable upfront investments and lumpy payments to purchase a unit of fuel [2,5,6]. As households in developing countries face a scarcity of liquidity and access to credit [88,89], they face budget constraints for adopting high-efficiency residential energy technologies. Remittance inflows stimulate financial development and provide access to credit [12,29], facilitating households to embrace high-efficiency residential energy technologies. Second, international migration promotes technology diffusion in developing countries through diaspora networks, knowledge transfers, and financial supports [13,30]. In particular, remittances support technology diffusion by improving recipient households’ access to credit, promoting business investment, and encouraging entrepreneurship [90,91]. Moreover, remittances have a positive impact on educational development [36,37], ultimately contributing to switching to cleaner fuels [8].

Concerning other control variables, we find that a 1% increase in real GDP per capita is associated with a 0.09% increase, and a 1% increase in the urban population (ratio to total population) is associated with a 0.57% increase in the share of high-efficiency energy sources in residential energy consumption in the long run. These results confirm the findings of the seminal work of Leach [2] on energy transitions [55,64,92]. The reason behind such a positive association between high-efficiency residential energy consumption and income level could be that most families have an ideal preference for high-efficiency over low-efficiency residential fuels [2,21]. This preference for fuels forms an energy ladder running from biomass fuels to kerosene, LPG, and electricity [2], and a higher income level assists them in ascending the energy ladder. Similarly, the positive association between high-efficiency residential energy consumption and urbanization can be justified by the argument that urban residents tend to use high-efficiency fuels, given the ease of access to fuels and appliances, sociocultural preference for high-efficiency fuels, and housing infrastructure in urban areas.

Our findings highlight several key contributions to the literature and their implications for policy. First, this study revealed that remittances are associated with energy transitions in the residential sector of developing countries. Developing countries have attempted to achieve the United Nations’ sustainable development goal of “ensuring access to affordable, reliable, sustainable, and modern energy for all”. Our results show a positive long-run relationship between remittances and high-efficiency residential energy uptake in developing countries. This does not imply that policies targeting the promotion of remittances should be recommended to promote the energy transition from low- to high-efficiency residential energy sources, because remittances induce favorable and unfavorable effects on macroeconomic conditions. Instead, the role of remittances should be considered in developing countries, as they are associated with many benefits, such as ensuring smooth energy transition in the residential sector. Second, remittance inflows in developing countries depend on various factors, such as exchange rates, diplomatic ties, macroeconomic conditions, and natural disaster events in host and home countries [93,94]. Such dependency makes high-remittance recipient countries vulnerable to abrupt changes in remittance flows. Previous studies have shown that remittances provide a safety net against income volatility and help families stabilize their incomes [31,32], and sudden disruption of remittance flow into high-remittance-receiving countries could force households to switch back to low-efficiency residential energy sources, endangering current achievements. Therefore, governments of high-remittance-receiving countries should focus on safeguarding households from the negative impact of a sudden disruption in remittance inflow.

4.5. Robustness Check

In this subsection, we validate the main results by conducting a sensitivity test. Previous studies consider other factors, such as fuel price, demography, agriculture dependency, family size, and education, to understand households’ fuel transition dynamics [25,26,27,40,95]. Among these factors, we consider three additional control variables for the sensitivity test: the log of the real oil price, the dependent population, and agricultural land. We calculate domestic real oil prices using the US real oil price and domestic consumer price indices of the respective countries.

The logic behind the choice of additional control variables for the sensitivity test is as follows. First, because energy markets are not well developed, obtaining high-efficiency residential fuel prices is difficult in developing countries. Thus, we used real oil prices as a proxy for high-efficiency residential fuel prices. We used real oil prices as a proxy for high-efficiency residential fuel prices because oil products, such as kerosene, diesel, and gasoline, constitute a significant portion of households’ high-efficiency residential energy demand. In addition, oil prices affect the prices of other high-efficiency residential fuels, such as gas and electricity. Second, the demographic structure could also influence energy usage patterns. Especially, households in developing countries are left with less disposable incomes for energy consumption when the dependent population becomes higher in the total population. The lower level of disposable income for energy consumption forces households to opt for low-efficiency fuels, which could create friction in energy transition progression in the residential sector. For this reason, this study used a population of young (age < 15 years) and old (age > 64 years) people. Third, many people in developing countries are dependent on agriculture. For these people, agriculture is not only a source of income but also a source of various forms of low-efficiency residential fuels, such as agricultural waste, animal dung, and firewood. Consequently, residents are more likely to use low-efficiency residential fuels when engaging in agricultural practices. In this study, we proxied the level of agricultural practices by the share of agricultural land to the total land area to represent the country’s dependency on agricultural practices.

The results of the sensitivity test are presented in

Table 9. PMG-ARDL estimates for robustness check.

Dependent Variable: Share of High-Efficiency Energy Sources in the Residential Energy Consumption.
(A) Long-run estimates
Personal remittances, received (ratio to GDP)	REM	0.323 ***
		(0.079)
Log (real GDP per capita)	Y	0.073 ***
		(0.016)
Urban population (ratio to total population)	URB	1.845 ***
		(0.165)
Log (real oil price)	OP	−0.005 ***
		(0.002)
Log (dependent population—young and old)	DEP	−0.464 ***
		(0.037)
Agricultural land (ratio to total land area)	AGR	−0.168 *
		(0.101)
(B) Short-run estimates
Error correction term	ECT	−0.242 ***
		(0.059)
Personal remittances, received (ratio to GDP)	REM	0.0
		(0.13)
Log (real GDP per capita)	Y	0.009
		(0.045)
Urban population (ratio to total population)	URB	4.418
		(4.4)
Log (real oil price)	OP	0.0
		(0.002)
Log (dependent population—young and old)	DEP	0.443
		(0.546)
Agricultural land (ratio to total land area)	AGR	733.698
		(732.313)
Intercept	C	0.15
		(0.13)
Hausman test statistics (p-value shown in parentheses)		1.75 (0.94)
No. of countries		26
No. of obs		585

Significance levels: 1% (***) and 10% (*).

. The results broadly show positive long-run coefficient estimates of personal remittances, real GDP per capita, and urban population, which confirms the validity of our main result. In addition, the sensitivity test results provide additional information. For example, the coefficient of the real oil price indicates that a 1% increase in the real oil price is associated with a 0.005% reduction in the share of high-efficiency energy sources in residential energy consumption. A negative association between real oil price and high-efficiency residential energy uptake is expected as oil products, such as kerosene, diesel, and gasoline, form part of high-efficiency residential fuels for households. However, the weak linkage could be justified by the argument that high-efficiency residential energy sources are composed of fewer oil products in developing countries. In addition, the results indicate that a 1% increase in the dependent population (the population of young and older people) is associated with a 0.46% decline in the share of high-efficiency energy sources in residential energy consumption. Such a negative association may suggest that households in developing countries would have less disposable income to purchase high-efficiency residential fuels when the dependent population increases. Moreover, the coefficient of the ratio of agricultural land (to total land area) indicates that a 1% increase in the ratio of agricultural land is associated with a 0.17% decline in the share of high-efficiency residential energy sources. The negative association could be because more low-efficiency residential fuels, such as agricultural waste, animal dung, and firewood, will be easily available to households as by-products of increased agricultural practices, leading to a reduction in the share of high-efficiency residential energy sources.

5. Conclusions

This study investigated the long-run association between the share of high-efficiency energy sources in residential energy consumption and remittance inflows in a sample of 27 developing countries by employing a PMG-ARDL model. Our findings confirm that remittances are associated with the households’ energy transition from dirty and low-efficiency (such as wood, charcoal, and animal waste) to clean and high-efficiency (such as kerosene, gas, and electricity) residential energy sources in developing countries. Furthermore, the results indicate that real GDP per capita, urban population, real oil price, dependent (young and old) population, and agricultural land variables help explain households’ energy switching dynamics. The estimates of the PMG-ARDL model reveal that a 1% increase in personal remittance inflow (ratio to GDP) is associated with a 0.24% increase in the share of high-efficiency energy sources in residential energy consumption.

Although our results reveal a long-run association between the share of high-efficiency energy sources in residential energy consumption and remittance inflows in developing countries, this study has several limitations that call for a more comprehensive investigation in future research. First, this study investigated the energy transition dynamics in the residential sector, accounting for solid and non-solid fuels, where solid fuels are regarded as low-efficiency and non-solid fuels as high-efficiency residential energy sources [57,58,59]. Our analysis in this study, however, does not account for the roles of various fuel types within the solid and non-solid fuel categories. Even for solid (non-solid) fuels, there is a hierarchy of fuel preferences in the residential sector, depending on the attributes of fuels, such as cleanliness, efficiency, safety, and user-friendliness [21]. Energy transition within solid fuels may be important for developing countries, especially for the least developed countries, given their low income and technological constraints. Thus, an examination of the energy ladder argument within solid fuels is also required to precisely understand how remittances help households switch between solid fuels.

Second, developing countries receive several forms of financial transfer through remittances, foreign aid, portfolio flows, and foreign direct investment. This study considers only the role of remittance inflows in relation to the adoption of high-efficiency residential energy sources in the residential sector. Other forms of financial transfer could also play a crucial role in the dynamics of energy transition in the residential sector in developing countries, which calls for further investigation. Third, developing countries have different characteristics in terms of various aspects, such as economic structure and energy resource endowments, which could potentially influence the relationship between remittances and high-efficiency residential energy use. Our study only examined the general long-run association between remittances and energy transition in the residential sector of developing countries. Nevertheless, it is important to examine this issue at the country level and make a country-by-country comparison, which would allow for a deeper understanding of how remittances are linked to the energy ladder hypothesis in developing countries. Fourth, we only identified the association (relationship or correlation) between remittances and high-efficiency residential energy uptake in the residential sector of developing countries and did not examine the causal relationship given our limited dataset. Our results are based on broad statistical observation and cannot evaluate a direct causal link between remittances and residential energy transition. Identifying the causal relationship would be helpful for better understanding how remittances help households ascend the energy ladder in developing countries.

Fifth, remittances are an income source for households in developing countries; thus, the variables of remittance and non-remittance income should be incorporated into the model simultaneously. However, the non-remittance income variable is not available in our dataset at the macro level. In addition, owing to data limitations and to make our model parsimonious and simple, this study did not incorporate other socioeconomic indicators, such as inequality, education, subsidies, foreign investment, official development assistance, and carbon offsets, which can be crucial for determining the energy mix in the residential sector. To identify the comprehensive role of remittances, we need to incorporate more disaggregated data such as non-remittance income and other socioeconomic indicators. Sixth, the energy ladder hypothesis is mainly related to households’ energy transition behavior; therefore, we confined our study to the residential sector of developing countries. However, sectors of activity, in particular agriculture, may play an important role in determining the share of high-efficiency residential energy, which requires us to examine how agricultural activities relate to the energy ladder argument. Although we admit that several issues remain to be addressed, this study could contribute to a better understanding of how remittance inflows are associated with the energy ladder argument related to energy transition dynamics in the residential sector in developing countries.