Household Energy Poverty in European Union Countries: A Comparative Analysis Based on Objective and Subjective Indicators

Abstract

The study aims to assess household energy poverty in European Union (EU) countries, comparing them based on the Objective Energy Poverty Index and the Subjective Energy Poverty Index. The Objective Energy Poverty Index is derived from indicators such as energy expenditure share, risk-of-poverty rate, and electricity prices. The Subjective Energy Poverty Index includes indicators such as the inability to keep the home adequately warm, arrears on utility bills, and bad housing conditions. Both indices aggregate the indicators mentioned above using equal and non-equal weighting approaches. The analysis uses country-level data from 2019 to 2023 sourced from Eurostat. The findings indicate considerable variation in household energy poverty across the EU, with more pronounced inequalities in subjective indicators than objective ones. Additionally, the study reveals a weak correlation between the Objective Energy Poverty Index and the Subjective Energy Poverty Index, leading to differing country rankings based on these indices. However, the choice of weights in constructing the energy poverty indices does not significantly impact a country’s energy poverty ranking. The paper also identifies countries where household energy poverty decreased in 2023 compared to 2019 and those where it increased. Regarding the Subjective Energy Poverty Index, Croatia and Hungary showed the most notable improvement in their rankings among European countries, while France, Germany, and Spain deteriorated their positions. According to the Objective Energy Poverty Index, Bulgaria, Croatia, Portugal, and Spain demonstrated the most significant improvement, whereas Greece experienced a considerable decline.

1. Introduction

Household energy poverty is a crucial social challenge that hinders the social inclusion of European Union (EU) citizens. Recent changes in energy supply conditions, coupled with shifts in the broader economic landscape, have raised new challenges regarding the affordability of energy and the difficulties households face in meeting their basic energy needs [1]. This issue is increasingly recognized as one of the most pressing problems likely to affect societies in the near future [2].

The world is undergoing an energy transformation that aims to balance three elements: energy security, energy equity (availability and affordability), and environmental sustainability [3]. However, this transformation has recently been cumulatively influenced by a series of shocks [4]. Significant barriers to alleviating energy poverty include political instability and violence, climate change, post-COVID-19 recovery, widespread inequality, and the cost of living issues. It is important to emphasize that alleviating energy poverty is identified alongside energy security and climate change as one of the most important paradigms for the future energy sector, which has received less attention in research and inclusion in political agendas [5].

Energy poverty has no uniform and universally accepted definition. According to the EU directive [6], ‘energy poverty’ means a household’s lack of access to essential energy services, where such services provide basic levels and decent standards of living and health, including adequate heating, hot water, cooling, lighting, and energy to power appliances, in the relevant national context, existing national social policy, and other relevant national policies, caused by a combination of factors, including at least non-affordability, insufficient disposable income, high energy expenditure, and poor energy efficiency of homes. More broadly, energy poverty can be understood in terms of the availability and affordability of modern energy sources that meet the basic needs of a household [3,7]. The causes of energy poverty can be found in a combination of the following three contributing factors: low-income level (economic situation), inadequate quality of buildings/low energy efficiency of dwellings (poor technical condition), and high energy prices [8].

Energy poverty is one of the forms of material deprivation, separate from income poverty [9]. Additionally, due to the inadequate quality of buildings and the low energy efficiency of dwellings, energy poverty is closely linked to housing poverty. Energy poverty affects many dimensions of human life, impacting health, social inclusion, environmental quality, and life satisfaction [3]. It also determines productivity [9], limits job opportunities [10], and acts as a barrier to education [11]. The issue is relevant in both developing and developed countries. In developing economies, studies emphasize the lack of access to clean energy [12]. In contrast, in developed economies, the challenge lies in providing economically affordable energy services [2,3].

There are many threads in the literature on energy poverty research. Research primarily focuses on the definitional dilemmas related to energy poverty. Another key issue in energy poverty studies is the development of measurement methods [4]. These methods pose significant challenges for researchers due to the complex, multi-dimensional nature of energy poverty and its variability across different spatial and temporal contexts [2].

The growing importance and relevance of energy poverty for both households and entire societies [11] prompted us to focus on measuring household energy poverty. Our study contributes to the literature in several respects. Firstly, we focus on analyzing household energy poverty at the country level, an approach currently underrepresented in the research [13]. Secondly, in response to the conclusions of various researchers [8,14,15,16] that only multidimensional measures can fully capture energy poverty, we apply both subjective and objective approaches in our measurement. This dual approach helps to fill a significant research gap in the field. Our research aligns with the trend in international comparative studies by focusing on European Union (EU) countries, which experience varying degrees of energy poverty [17]. The COVID-19 crisis, which led to increased unemployment, reduced household income, and heightened energy demand, has exacerbated the energy vulnerability of many households [18]. Moreover, this likely contributed to the deepening of energy poverty in EU countries [8], though the extent varies across countries [2,19]. Additionally, the escalation of energy prices due to the Russia–Ukraine war in 2022 further exacerbated energy poverty, causing a sharp increase in inflation [2,13,20]. Given these developments, understanding the level of energy poverty in EU countries and capturing the recent changes, particularly since 2019, has become more critical than ever. The accurate measurement of energy poverty is essential for informing policies that effectively address this issue. In this study, we assess multidimensional household energy poverty using composite indicators, also known as synthetic measures [8]. Following Kryk and Guzowska [8], we use a dynamic approach that provides a consistent reference point throughout the analysis period. In addition to using their proposed method of aggregating indicators with equal weights, we also apply a method that assigns different weights to the indicators.

The aim of our research was to assess and compare household energy poverty in EU countries over the period 2019–2023 using different measurement methods. We focus on answering the following research questions (RQ):

(RQ1): How do countries differ in energy poverty indicators?

(RQ2): Which countries exhibited the most significant changes in energy poverty between 2019 and 2023?

(RQ3): Do the country rankings based on the objective indicators of energy poverty align with those based on subjective indicators?

(RQ4): Does the choice of weights in constructing composite indicators matter in determining a country’s energy poverty position?

The paper is organized as follows: Section 1 introduces the topic. Section 2 presents a literature review on the measurement of energy poverty. Then, Section 3 describes the data and outlines the research methodology used. Section 4 and Section 5 report the research findings and provide a discussion of the results. Finally, the paper concludes with a summary of the key insights and implications.

2. Literature Review—Measuring Energy Poverty

Research into energy poverty has gained increasing importance in recent years, attracting significant interest from many researchers who continuously expand their investigations to understand this issue better [21]. The evolving research trends and issues surrounding energy poverty have also been acknowledged by national and international policies aimed at addressing the issue [22]. This recognition is evident in successive EU initiatives and directives, the objectives of which have evolved over time [8,23,24,25,26].

Researchers emphasize that energy poverty is a complex and multidimensional concept, which complicates its definition [27,28]. A review of the definitions of energy poverty in the literature reveals that it encompasses all the aspects of poverty related to energy [10,26].

Energy poverty is most commonly defined as the inability to maintain an adequate temperature within a household [10,16]. However, individual authors additionally emphasize different dimensions of the problem of energy poverty, including the ability to meet basic household needs [5]; perform essential household activities, e.g., lighting, cooking, heating, and cooling [27,29,30]; and issues related to affordability [31,32] and access to modern energy sources [27]. It should be noted that the terms ‘energy poverty’ and ‘fuel poverty’ are often used interchangeably in the literature [24,28,33]. However, some researchers highlight distinctions between them, defining fuel poverty as primarily concerning issues related to low energy availability, whereas energy poverty is viewed as a broader term encompassing all the problems associated with insufficient access to energy [10,23,27]. In our study, we use the term ‘energy poverty’ as a synonym for ‘fuel poverty’.

Researchers are investigating a range of specific issues related to energy poverty. The first stream of research, which seems to be fundamental, focuses on defining energy poverty and developing indicators for its measurement [10,26]. Additionally, other research threads include the determinants of energy poverty [11,12,19,34], exposure to energy poverty [28,35,36], support policies and strategies for coping with energy poverty [24], the importance of energy poverty for the natural environment, and the use of new energy sources [37]. Studies on energy poverty and its various dimensions are conducted in individual countries as well as groups of countries, a task made challenging by the availability of statistical data. The definitional dilemmas surrounding energy poverty lead to the use of various methods for its measurement. Each method used has its advantages and disadvantages (

Table 1. Features of different research approaches/methods for measuring energy poverty.

Specification	Characteristic	Disadvantages	Advantages
Outcome-based approach (direct measurement)	the assessment of the implementation of basic energy services, such as heating, compared to established benchmarks	technical impracticality (difficulty in defining sufficient criteria for energy services, e.g., geographical differences occur; individual preferences) ethical issues (entering homes and monitoring households)	great potential in the future—with the transition to ‘smart homes’, as well as the smart meter rollout across the EU
Expenditure-based approach (objective)	assessing the ratio of household income to energy expenditure, taking into account both absolute and relative thresholds	complexity potential inaccuracies in calculating energy needs and income (underestimation) difficulties in application in different European contexts exclusion of the most vulnerable households (those unable to bear the costs) narrow approach to energy needs (focusing mainly on heating or energy affordability)	commonly used objectivity and measurability
Consensual-based approach (subjective)	self-assessment of individuals regarding their ability to meet specific energy needs	exclusion errors (households do not report that they are poor) susceptibility to bias, dependence on the socio-cultural environment; subjectivity of indicators (different individuals might have different understandings of what each indicator) contingent nature on individual interpretations	simple, make it easier to understand energy poverty analysis data collection is less complicated than expenditure data, allow us to capture the broader elements of energy poverty, such as social exclusion and material deprivation provide a concise and consistent picture of the level of energy poverty in a given region or country over time transparent and accessible to decision makers

Source: own study based on [2,13,16,18,26,38].

), which are influenced by factors such as data availability and the specific objectives of measuring energy poverty [2].

The presented approaches (Table 1) to measuring energy poverty focus on the technological, physical, or financial thresholds of access to energy and are complementary to each other [4]. Various methodologies are used to assess energy poverty, including individual indicators that focus on specific aspects, such as energy availability and affordability. These methodologies can vary depending on the country or group of countries being analyzed, as well as the indicators used to measure energy poverty. It is important to note that these indicators are subject to ongoing scientific debate [14] and may evolve over time. Additionally, some indicators serve as a foundation for further research and inspire new approaches. For example, the ‘Low Income High Cost’ (LIHC) indicator was used in the United Kingdom to identify fuel-poor households. A household was considered fuel-poor if (i) its ‘required fuel costs’ were above the national median and (ii) paying these costs would leave the household with a residual income below the official poverty line [30]. Recently, the LIHC indicator in the United Kingdom has been replaced by the ‘Low Income Low Energy Efficiency’ (LILEE) fuel poverty metric, which is defined as follows: a household is considered fuel-poor if (i) it lives in a property with a fuel poverty energy efficiency rating of band D or below, and (ii) after spending the required amount to heat their home, the household is left with a residual income below the official poverty line [26].

Research often begins with a single indicator, but due to the difficulties in capturing the complexity of energy poverty [13,14], multiple subjective and objective indicators are commonly used. The literature provides numerous examples of the multidimensional measures of energy poverty, with various indicators highlighting different aspects of this issue. However, no consensus exists on which indicators are the best or how to integrate them [8]. This lack of agreement may reflect a limited recognition of energy poverty and its indicators in many countries [39]. Based on the review of the relevant literature,

Table 2. Selected research approaches to measuring energy poverty in international analyses.

Indicators *	Authors/Spatial Scope
	Kryk and Guzowska EU24 [8]	Bollino and Botti EU27 + 4 other European countries [17]	Thomson and Snell EU25 [23]	Healy and Clinch EU14 [40]	Halkos and Gkampoura EU28 [37]	Dubois and Meier EU28 [1]	Karpinska and Śmiech 11CEE [36]	Kashour and Jaber EU27 [2]
Electricity and/or gas prices for household consumers	+							+
Final energy consumption in households per capita	+				+
The share of household spending on electricity, gas, and other fuels in relation to total household expenditure						+
The budget share allocated to a minimum amount of electricity for households at risk of poverty						+
Purchasing power-adjusted GDP per capita					+
Household annual mean income								+
Equaivalized disposable income less than 60% of median after-housing-costs income							+
Final energy consumption in households by fuel type [%]					+
Population unable to keep home adequately warm due to poverty status [%]	+
Household energy efficiency score								+
(In)ability to keep home adequately warm		+	+	+	+	+		+
Arrears on utility bills		+	+	+	+	+		+
Leaking roof, damp walls, or rotten windows	+	+	+	+	+	+		+
The lack of adequate heating facilities and absence of central heating				+
The share of the population who reported that their dwelling was neither comfortably warm nor cool						+
The efficiency of the heating/cooling system in maintaining the appropriate temperature in the apartment		+

* Note: the names of the indicators may vary slightly in wording. Additionally, some indicators were combined (e.g., damp walls and rotten window frames were studied separately by the authors), so the number of indicators may differ from those originally described in the approach. Source: own study.

presents selected indicators that have been used in comparative analyses between countries to date.

Energy poverty can be analyzed on macro-, and microeconomic scales depending on the analysis’s purpose and the data aggregation level [1,41]. Regardless of the scales analyzed, most studies considered indicators related to the situation as perceived by households, such as the inability to keep their home adequately warm, living in a home with leakage, dampness, and rot, and having arrears on utility bills [2,8,13]. The international comparisons of EU countries using these indicators are enabled by data from the EU Survey on Income and Living Conditions (EU-SILC), collected by Eurostat. For microdata analysis, household-level EU-SILC data can be utilized, while for macrodata analysis, aggregated country-level data published by Eurostat can be used. It is worth mentioning that some energy poverty indicators, such as arrears on utility bills and inability to keep home adequately warm, are used in the analysis of material deprivation [42,43], while some, such as living in a dwelling with a leaking roof, damp walls, floors, or foundation, or rot in window frames of floor—in the analysis of housing deprivation [44,45].

Analyzing the presented indicators used by researchers to measure energy poverty (Table 2), it is evident that various sets of indicators are used, encompassing energy prices, energy expenditures, thermal comfort, and energy efficiency of housing. It is crucial to consider and examine all these factors when assessing energy poverty. Additionally, it is important to note that a limitation in selecting indicators for international analyses may be the lack of available data [17] and challenges related to transparency, comparability, and effectiveness across different contexts [22,26].

3. Materials and Methods

To assess the multidimensional energy poverty of households in EU-27 countries, we used Eurostat data for 2019 and 2023 [46]. We selected 2019 as a baseline year before the COVID-19 pandemic in the EU and 2023 for the most recent data available across all the analyzed countries. This time frame allows us to capture the changes in EU countries due to changes in energy supply and demand related to, among others, the COVID-19 pandemic and the armed conflict between Ukraine and Russia.

The following countries are considered: Austria (AT), Belgium (BE), Bulgaria (BG), Croatia (HR), Cyprus (CY), Czech Republic (CZ), Denmark (DK), Estonia (EE), Finland (FI), France (FR), Germany (DE), Greece (GR), Hungary (HU), Ireland (IE), Italy (IT), Latvia (LV), Lithuania (LT), Luxembourg (LU), Malta (MT), The Netherlands (NL), Poland (PL), Portugal (PT), Romania (RO), Slovakia (SK), Slovenia (SI), Spain (ES), and Sweden (SE).

In our research, when selecting our indicators, we consider the indicators used by the EU Member States to estimate energy poverty in their energy and climate plans [6], e.g., (a) the inability to keep the home adequately warm; (b) the arrears on utility bills; (c) the total population living in a dwelling with a leaking roof; damp walls, floors, or foundation; or rot in window frames or floor; (d) at-risk-of-poverty rate (cutoff point: 60% of median equivalized income after social transfers). Moreover, our choice of indicators follows those previously used to measure energy poverty by other authors (see Table 2). In particular, our indicators align with the conceptualization of energy poverty presented by Kashur and Jaber [2], who adopt a cause-and-effect approach to household energy poverty [2]. The selected indicators encompass, on the one hand, high energy costs and low income/risk of poverty, reflecting the Member State’s capacity to address energy poverty. On the other hand, we include commonly accepted proxies for measuring energy poverty, such as insufficient heating in dwellings, energy arrears, and housing quality problems. Furthermore, the availability of statistical data was considered when selecting the indicators. For example, gas prices were excluded due to incomplete data for all the EU countries during the analyzed period.

In our study, as in other research [13,37,47,48], we distinguish between the objective and subjective indicators in energy poverty analysis. The objective indicators typically refer to household income and their energy expenditure [47,48]. However, unlike in the studies of Llorca et al. [47] and Sokołowski et al. [48], we analyze data at the country level rather than the household level, and we include the risk of poverty rates, electricity prices, and energy expenditure shares in our set of objective indicators. As for subjective indicators, they rely on the households’ self-assessments of factors such as housing conditions, the ability to pay utility bills, or achieving thermal comfort [38,40,48]. Following other studies [13,48], we include in our set of subjective indicators the inability to keep the home adequately warm; the presence of leaks, dampness, or rot; and the inability to pay utility bills. Therefore, we considered six indicators (

Table 3. Indicators used to measure energy poverty in EU countries.

No.	Indicator	Short Name of Indicator
	Objective approach
1	Expenditure on electricity, gas, and other fuels as a proportion of the total household expenditure [%]	Energy expenditure share
2	The share of persons with an equivalized disposable income, before social transfers, below the risk-of-poverty threshold, which is set at 60% of the national median equivalized disposable income (after social transfers) [%]	Poverty rate
3	Electricity prices for household consumers [Kilowatt-hour in Purchasing Power Standard (PPS), all taxes and levies included]	Electricity price
	Subjective approach
1	The share of the population who are unable to keep homes adequately warm [%]	Warm home
2	Share of the population having arrears on utility bills [%]	Utility bills
3	Share of the population living in a dwelling with a leaking roof; damp walls, floors, or foundation; or rot in window frames of floor [%]	Bad housing conditions

Source: own study.

), three for objective and subjective measurement of energy poverty. The selected indicators are widely accepted and facilitate the comparison of energy poverty across Member States [39]. Using a set of objective and subjective indicators independently has advantages: it captures the diverse definitions of energy poverty and provides a more comprehensive view than any single indicator alone [48]. The distinction between objective and subjective indicators at the country level is a novel aspect of our analysis, as most studies that differentiate these indicators focus on household-level data.

In the Results Section, we analyze each indicator individually and the aggregated values of the energy poverty indices. Initially, basic descriptive statistics for all the input indicators—such as the mean, standard deviation, minimum, and maximum—are computed and presented in the preliminary data analysis. Box and whisker plots are applied to detect outliers. In addition, the Gini index is used to measure the inequality of all the input indicators. The Gini index offers key advantages, providing a single summary measure that effectively captures the overall level of disparity in a distribution. This allows for straightforward comparisons between the different distributions of indicators. The index ranges from 0 to 1, where 0 indicates perfect equality, and 1 represents perfect inequality, making it an intuitive tool for comparing inequality levels across different datasets. Moreover, the Gini index is closely linked to the Lorenz curve, a valuable tool for visually assessing inequality. The Gini index is calculated as twice the area between the Lorenz curve and the line of perfect equality (the diagonal). A key advantage of the Lorenz curve is that it offers deeper insights into the inequality of a given indicator. The slope of the curve reveals the degree of inequality across different ranges of countries, making it a powerful tool for both visualizing and analyzing disparities within a distribution.

The multifaceted household energy deprivation is evaluated using a composite indicator approach. There are several approaches to building composite indicators [49,50]. The fundamental steps typically applied are as follows (see Floridi et al., [49]):

• The normalization of data.
• The weighting and aggregation of indicators.

In the initial step, the input values of all the indicators are normalized to make them comparable.

As in the study by Kryk and Guzowska [8], the values of the input indicators are rescaled to range from 0 to 1 using the min–max procedure, also called the zero unitarization method. The individual value of the indicator $x_j$ for the i-th country is transformed into the score $z_{jit}$:

$$z_{jit}={\displaystyle \frac{x_{jit}-\underset{\mathit{it}}{\mathit{min}}\left(x_{jit}\right)}{\underset{\mathit{it}}{\mathit{max}}\left(x_{jit}\right)-\underset{\mathit{it}}{\mathit{min}}\left(x_{jit}\right)}}$$

(1)

where $\underset{\mathit{it}}{\mathit{min}}\left(x_{jit}\right)$ and $\underset{\mathit{it}}{\mathit{max}}\left(x_{jit}\right)$ are, respectively, the minimum and the maximum values of $x_{jit}$ across the countries; i = 1, 2, …, 27 and time t = 1, 2, …, T for j-th indicator j = 1, 2, …, K; K is the number of indicators; and T is the number of years.

In the second step, the indicators need to be weighted and aggregated to derive composite indices. Thus, we define the energy poverty score for the i-th country as the weighted sum of scores $z_{jit}$:

$$s_{it}={\sum }_{j=1}^K{w}_j{z}_{jit}$$

(2)

where w_j is a weight representing the relative significance of j-th indicator, wherein

$${0\le w}_k\le 1\mathrm{and} {\sum }_{j=1}^K{w}_j$$

(3)

= 1

Most composite indices rely on equal weighting, meaning all the variables are assigned the same importance. In our study, we utilize both equal and unequal weights. For the latter, we apply a statistical-based weighting system proposed by Betti and Verma’s [51] method. This method is widely used in the analysis of housing poverty [52], material and social deprivation [43,53], food poverty and insecurity [54,55], and quality of life [56,57], among others. However, to our knowledge, it has not yet been applied to energy poverty analysis. It comprises two factors: the dispersion of the poverty indicator and its correlation with other poverty indicators. According to Betti and Verma [51], the weights can be defined as follows:

$${\mathsf{\omega }}_j={\mathsf{\omega }}_j^a\cdot {\mathsf{\omega }}_j^b$$

(4)

where the first component is the item’s coefficient of variation, and the second component is a measure that assigns less weight to items more highly correlated with others, thereby reducing the impact of redundancy. An accurate formulation of component ${\mathsf{\omega }}_j^b$ is provided by Betti and Verma; Betti et al. [43,51]:

$${\mathsf{\omega }}_j^b=\left({\displaystyle \frac{1}{1+{\sum }_{k\prime =1}^K{r}_{kk\prime }|r_{k{k}^{\prime }}<r^{\mathrm{*}}}}\right)\cdot \left({\displaystyle \frac{1}{{\sum }_{k\prime =1}^K{r}_{kk\prime }|r_{kk\prime }\ge r^{\mathrm{*}}}}\right)$$

(5)

where $r_{kk\prime }$ is the correlation coefficient between two different scores z_kit and z_k’it and $r^{*}$ is a predetermined cut-off correlation level.

To sum to one, the values (3) were normalized according to the following formula:

$$w_j={\displaystyle \frac{{\mathsf{\omega }}_j}{{\sum }_{k=1}^K{\mathsf{\omega }}_j}}$$

(6)

We calculate weights using mdepriv—the Stata procedure created by Pi Alperin and Van Kerm [58]. Higher values of all the analyzed indicators—input, normalized, and synthetic—reflect greater energy poverty in a given country. We refer to the energy poverty score calculated using Equation (2) as the Objective Energy Poverty Index (OEPI) for objective indicators and the Subjective Energy Poverty Index (SEPI) for subjective indicators. We separately examine the Objective Energy Poverty Index (OEPI) and the Subjective Energy Poverty Index (SEPI), investigating changes in these indices from 2019 to 2023. Additionally, we explore their interrelationship and correlations with classical macroeconomic variables used in the analyses of energy poverty, such as GDP per capita [16], social spending per capita [59,60], and the unemployment rate [61]. We also construct country rankings based on OEPI and SEPI and analyze changes in these rankings over time. The results were analyzed using classical statistical methods, including the Spearman correlation coefficient, to compare country rankings [62].

4. Results

Our results and discussion are organized into several subsections. First, we present a preliminary analysis of the input indicators of household energy poverty in the EU, focusing on the basic characteristics of their distributions. Next, we examine the changes in the Objective Energy Poverty Index (OEPI) and the Subjective Energy Poverty Index (SEPI) between 2019 and 2023. Furthermore, we explore the links between OEPI and SEPI and their correlations with GDP per capita, social spending per capita, and the unemployment rate. We also compare country rankings based on OEPI and SEPI and analyze the changes in these rankings over time. Finally, we discuss the strengths and limitations of our study.

4.1. Preliminary Data Analysis

In the first stage of our study, we analyzed the basic characteristics of the input indicators (see

Table 4. Descriptive statistics for the input indicators of household energy poverty in the EU.

Indicators	2019	2023
Energy expenditure share
Mean (Std. Dev.) Min.–Max.	4.293 (1.3666) 2.000–8.5000	5.126 (1.549) 2.200–7.900
Gini index	0.170	0.173
Poverty rate
Mean (Std. Dev.) Min.–Max.	16.337 (3.859) 10.100–23.800	16.256 (3.557) 9.800–22.500
Gini index	0.134	0.125
Electricity price
Mean (Std. Dev.) Min.–Max.	0.204 (0.038) 0.140–0.270	0.273 (0.072) 0.142–0.394
Gini index	0.106	0.151
Warm home
Mean (Std. Dev.) Min.–Max.	8.196 (7.707) 1.800–30.100	9.467 (5.986) 2.100–20.800
Gini index	0.466	0.335
Utility bills
Mean (Std. Dev.) Min.–Max.	8.244 (7.124) 1.500–32.500	7.533 (6.070) 1.100–32.900
Gini index	0.410	0.352
Bad housing conditions
Mean (Std. Dev.) Min.–Max.	13.607 (5.893) 4.100–31.100	13.538 (6.962) 4.800–31.600
Gini index	0.233	0.283

and Figure 1 and Figure 2).

Based on the results presented in the tables and figures, it can be concluded that in the whole EU, there is no consistent pattern of variation among energy poverty indicators after COVID-19 compared to the values before the pandemic, both in terms of average values and dispersion. The most remarkable change is in electricity prices—the interquartile ranges almost do not overlap. Energy expenditures and the share of the population unable to keep home adequately warm increased at first sight after COVID-19, but on the other hand, the outliers disappeared for both indicators. The objective indicators of household energy poverty have overall much fewer outliers, both before and after COVID-19 (only Slovakia in the case of the electricity share), while for subjective indicators, there are many outliers, especially in the case of the inability to keep home adequately warm. In each case, all the outliers consistently appear above the upper whisker in the box plots for our indicators.

Table 5. Countries with the lowest (L) and highest (H) levels of household energy poverty.

Level	2019	2023
Energy expenditure share
L	Malta and Luxembourg	Malta
H	Slovakia	Slovakia and Poland
Poverty rate
L	Czech Republic	Czech Republic
H	Romania	Latvia, Estonia
Electricity price
L	Finland and Luxembourg	Malta
H	Germany	Cyprus
Warm home
L	Austria, Finland, Sweden, Slovenia, Luxembourg, Germany	Luxembourg and Finland
H	Bulgaria, Lithuania, Cyprus, Greece, and Portugal	Portugal, Spain, and Bulgaria
Utility bills
L	The Netherlands, Czech Republic, and Germany	The Netherlands
H	Greece and Bulgaria	Greece, Bulgaria, and Romania
Bad housing conditions
L	Finland	Sweden and Finland
H	Cyprus and Portugal	Cyprus

shows the EU countries that performed best and worst on energy poverty in 2019 and 2023 based on the selected indicators. As mentioned earlier, higher values for all the indicators mean greater energy poverty, indicating a worse situation for the given country. If one or more countries deviated from the minimum/maximum by less than 0.1 standard deviations, we treated them as ex aequo best/worst. Additionally, all the listed outliers are marked in bold font.

Based on Figure 2 and Figure 3 and Table 4, it is not possible to clearly identify ‘losers’ in terms of objective indicators. However, in terms of the subjective situation, a particularly adverse situation is evident in Bulgaria, Cyprus, Greece, and Portugal. Notably, Bulgaria has shown the most significant improvement over the investigated period, with a reduction of over 9 percentage points in the inability to keep homes adequately warm and nearly a 10 percentage point decrease in utility bills.

We also calculated the inequality among the countries using the Gini index for unweighted data, where each country constitutes a single observation. The Lorenz curves for all the indicators, comparing 2019 and 2023, are presented in Figure 3.

There are decreases in inequality for ‘warm house’ and ‘utility bills’ (with a very slight decrease in the inequality of the ‘poverty rate’ as well). In contrast, inequalities in ‘electricity prices’ and ‘bad housing conditions’ increased. For ‘energy expenditure share’, inequality remained almost the same.

Visible differences in the Lorenz curves can be observed for the ‘electricity price’ indicator and all three subjective indicators, which are also reflected in the changes in the values of the Gini index, as presented in Table 4. The most notable difference is seen for the ‘warm home’ indicator. For 50% of the countries least affected by this aspect of energy poverty, the slope of the Lorenz curve for 2023 is steeper (closer to 1) compared to 2019, indicating a more equal distribution in this lower half. In contrast, for the upper 15% of countries, the 2019 curve has a significantly steeper slope than the 2023 curve, with outliers in 2019 contributing to this behavior. An interesting trend can also be observed in the Lorenz curve for the ‘utility bills’ indicator, wherein the lower part of the distribution, the slope for 2023 is mostly higher than for 2019, and although in the upper part, it was not consistently lower in 2023 (outliers appear both in 2019 and 2023), the lower inequalities in the lower part of distribution overwhelm. As a consequence, the Gini index is smaller. For the ‘bad housing conditions’ indicator, despite the disappearance of outliers, inequality increased in 2023 in both the lower and upper parts of the distribution. The same pattern holds for the final set of Lorenz curves, where clear differences between 2023 and 2019 suggest that inequalities have grown at both ends of the distribution.

4.2. Results for Objective Energy Poverty Indices (OEPIs) and Subjective Energy Poverty Indices (SEPIs)

In the next stage of our study, we assigned weights to the normalized objective indicators. The values of the weights for the Objective Energy Poverty Indices (OEPIs) used in our study are presented in

Table 6. Values of weights for Objective Energy Poverty Indices (OEPIs).

Indicator	Equal Weights	Unequal Weights
Energy expenditure share	1/3	0.2923
Poverty rate	1/3	0.4120
Electricity price	1/3	0.2957

.

In the unequal weights approach, the ‘energy expenditure share’ and the ‘electricity price’ have nearly the same weight, while the weight for the ‘poverty rate’ is approximately 40% higher than the weights of the other two indicators.

Using the equal and unequal weights presented in Table 6, we calculated the Objective Energy Poverty Indices in accordance with formula 2. The results are presented in Figure 4.

For nearly all the countries, the situation in 2019 was better than in 2023, as the curve for 2019 is closer to the origin of coordinates than for 2023. The significant exceptions are Belgium, Malta, and Portugal, for which the OEPI is lower in 2023 than in 2019. The case of Romania is particularly interesting, as other studies have indicated that the primary challenge is energy affordability [1].

For this country, the objective indicator of energy poverty has a lower value in 2023 than in 2019, but only if the indicator is calculated with unequal weights. While energy prices in Romania increased in this period (0.34 vs. 0.26 in 2023 and 2019, respectively), the energy expenditure share decreased from 3.7% to 3.4%, and the poverty rate decreased from 23.8% to 21.1%. Especially, this last change is striking and makes the difference between increasing the OEPI with equal weights and a slight decrease in this index with non-equal weights, as the latter puts greater weight (cf. Table 6) on the poverty rate.

The difference between the direction of changes in the 2019–2023 period while calculating the objective indicator of poverty with equal and non-equal weights stresses the importance of the decision of proper weights.

As with the objective indicators, we assigned weights to the normalized subjective indicators. The values of the weights for the Subjective Energy Poverty Indices (SEPIs) used in our study are presented in

Table 7. Values of weights for Subjective Energy Poverty Indices (SEPIs).

Indicator	Equal Weights	Unequal Weights
Warm home	1/3	0.2914
Utility bills	1/3	0.3936
Bad housing conditions	1/3	0.3150

.

The results presented in Table 7 show that the differences in unequal weights are smaller than those for the objective indicators. The Subjective Energy Poverty Index (SEPI) values were determined based on these weights.

As for the Subjective Energy Poverty Indices (SEPIs), the picture is less clear-cut. While most countries saw an increase in the SEPI values in 2023 compared to 2019, the indices decreased in as many as 12 countries, whether calculated with equal or unequal weights (see Figure 5). The unnoticeable differences between the indicators with equal and unequal weights might be expected, as the weights differ much less than in the case of OEPI. The huge positive difference between the 2023 and 2019 values is observed in Denmark, Spain, and France. In all these three countries, it is mostly due to one of the elements of the synthetic indicator: the fraction of the population unable to keep their home adequately warm. In 2023, this fraction nearly doubled in France and increased even more dramatically in Denmark and Spain, reaching 250% and 280% of the 2019 value, respectively.

It is worth noting that the results derived from the indices with equal weights do not differ much from those with unequal weights, as the weights themselves (see Table 6 and Table 7) are not highly differentiated. Consequently, the Pearson correlation coefficient between the SEPI with equal weights and SEPI with unequal weights for combined data (both years) was 0.997. A similar result was observed for the OEPI, with a slightly lower Pearson correlation coefficient at 0.983.

To investigate the nature of the objective and subjective indicators, we examine their relationships with other key economic indicators, namely GDP per capita, social expenditures per capita, and the unemployment rate. The correlation coefficients between these variables and the Objective Energy Poverty Indicator (OEPI) and Subjective Energy Poverty Indicator (SEPI) for 2019 and 2023 are presented in

Table 8. Correlations between key economic indicators and OEPI and SEPI for the year 2019.

Indicator	OEPI, Equal Weights	OEPI, Unequal Weights	SEPI, Equal Weights	SEPI, Unequal Weights
GDP per capita [PPS]	−0.582	−0.552	−0.446	−0.456
Social spending per capita [PPS]	−0.552	−0.545	−0.496	−0.506
Unemployment rate [%]	0.140	0.200	0.391	0.410

and

Table 9. Correlations between key economic indicators and OEPI and SEPI for the year 2023.

Indicator	OEPI, Equal Weights	OEPI, Unequal Weights	SEPI, Equal Weights	SEPI, Unequal Weights
GDP per capita [PPS]	−0.408	−0.420	−0.269	−0.274
Social spending per capita [PPS]	No data	No data	No data	No data
Unemployment rate [%]	0.244	0.312	0.553	0.564

.

The objective indicators are more closely related to a nation’s overall wealth, which is confirmed by the data showing that GDP is more strongly correlated with OEPIs than with SEPIs. Interestingly, while the strength of this relationship decreased from 2019 to 2023 for both types of indices, the reduction is much more pronounced for subjective indices. Conversely, the unemployment rate is more strongly related to SEPIs. This relationship—unlike the one with GDP—significantly increased in 2023 compared to 2019.

It is important to note that the overall correlation of economic indicators with OEPIs and with SEPIs reflects the interplay of the correlations between their constituent elements and their weights. For example, while GDP has a strong correlation with the share of energy expenditures (r = −0.323 for combined data), it correlates even more strongly with aspects of subjective indicators such as the ability to keep homes warm and utility bills (r = −0.404 and −0.394, respectively). The goal here is not to delve into the correlations of indicators with their constituent elements, which can be carried out easily, but rather to practice using synthetic indicators, as these are commonly employed in evaluating the effects of new policies and regulations.

4.3. Comparison of Relative Changes in Energy Poverty between 2019 and 2023 Using Both Objective and Subjective Approaches

Relative changes (using 2019 as the base year) for the OEPI and SEPI are presented in Figure 6 and Figure 7 for the indices with both equal and unequal weights, respectively. The positive values on the plots mean worsening, while the negative values—improvement.

Based on the plots above, we can categorize all the EU countries into four groups according to their position in the coordinate system. The countries in the I quadrant (with increasing values of both OEPI and SEPI) have experienced worsening conditions in both dimensions. There are either 14 or 13 countries in this quadrant, depending on whether equal or unequal weights are used to calculate the energy poverty indices. The countries in the III quadrant (with decreasing values of both OEPI and SEPI) have improved in both dimensions of energy poverty. This quadrant includes two countries: Malta and Bulgaria. The countries in the II quadrant exhibit higher values of the OEPI while showing lower values of the SEPI. This is a very interesting case, as one may wonder, do, e.g., the prices of electricity really matter if people can still afford to warm their homes and have no arrears on utility bills? In this quadrant, there are ten countries. Belonging to the last IV quadrant means a negative change in the objective indicator (which means lower energy poverty) and an increase in the subjective indicator of energy poverty. That would be quite a strange situation, rather difficult to explain, and—indeed—only Portugal and Romania (the latter only in case of unequal wages) lie in this quarter but on the very verge of it.

4.4. Country Rankings Based on OEPI and SEPI in 2019 and 2023

Given that the changes in the Objective Energy Poverty Index (OEPI) and the Subjective Energy Poverty Index (SEPI) appear to be almost independent—where one may increase while the other decreases—two issues need to be addressed. First, it is necessary to clarify how the performance of the different countries is related across these two dimensions (subjective and objective); specifically, whether the countries with high values for the subjective indicator also tend to have high values for the objective indicator and vice versa. Second, it is important to recognize how changes in these subjective and objective indicators in each country compare with those in other countries.

Therefore, we have found the rankings of the countries for both subjective and objective indices of energy poverty and compared those rankings in 2019 and 2023 years to check if they differ depending on the indicators taken into account. To answer the second question, we have checked the differences in rankings in 2023 compared to 2019.

Table 10. Country rankings based on the Objective Energy Poverty Index (OEPI) and the Subjective Energy Poverty Index (SEPI).

Country	OEPI				SEPI
	With Equal Weights		With Unequal Weights		With Equal Weights		With Unequal Weights
	2019	2023	2019	2023	2019	2023	2019	2023
Austria	22	20	24	20	24	21	25	21
Belgium	11	5	11	6	14	17	14	17
Bulgaria	3	13	3	9	1	5	1	5
Croatia	9	16	9	14	10	19	10	18
Cyprus	18	12	17	13	2	2	3	2
Czechia	15	11	16	16	26	24	26	24
Denmark	17	17	18	17	18	15	19	15
Estonia	8	2	5	2	15	22	15	22
Finland	27	24	27	25	25	25	24	25
France	21	23	21	21	16	6	16	6
Germany	12	9	14	11	23	11	23	11
Greece	16	8	15	8	3	1	2	1
Hungary	24	26	26	26	6	14	6	14
Ireland	20	19	20	22	13	13	13	13
Italy	7	3	6	3	11	10	11	10
Latvia	2	1	2	1	7	9	7	9
Lithuania	10	4	8	4	5	7	4	7
Luxembourg	25	25	22	24	21	16	21	16
Malta	26	27	23	27	17	23	18	23
The Netherlands	23	18	25	18	22	18	22	19
Poland	5	6	7	7	20	26	20	26
Portugal	14	21	12	19	4	3	5	4
Romania	1	7	1	5	9	8	9	8
Slovakia	6	10	10	12	19	20	17	20
Slovenia	19	22	19	23	8	12	8	12
Spain	4	14	4	10	12	4	12	3
Sweden	13	15	13	15	27	27	27	27

presents the rankings of all the countries based on the energy poverty indices calculated using both equal and unequal weights.

The rankings for OEPI and SEPI, as well as for 2019 and 2023, show little agreement. Specifically, the correlation between the OEPI and SEPI ranks is relatively weak: for 2019, the Spearman rank correlation is 0.316 with equal weights and 0.402 with unequal weights, while for 2023, it is 0.145 with equal weights and 0.262 with unequal weights. This indicates a significant decrease in correlation from 2019 to 2023. It appears that the objective and subjective aspects of energy poverty were already somewhat decoupled, but this decoupling has become even more pronounced following the COVID-19 pandemic.

This ‘decoupling’ may be visualized with, e.g., indicators calculated with equal weights, as pictured in Figure 8 below. If the objective and subjective indices of energy poverty were perfectly correlated, the countries would align along the diagonal. Below this line, some countries are better off with respect to subjective aspects than to objective aspects, while above it—vice versa. The regression line for the 2019 data has a slope of 0.427 and a coefficient of determination (R²) of 0.083. In 2023, an even greater decoupling is observed, with the slope decreasing to 0.127 and the R² dropping to 0.015.

This decoupling of the indicators aligns with the correlation values between 2019 and 2023, which are significantly less than one. For subjective indices (SEPIs) with equal weights, it is equal to 0.788 and with unequal weights 0.786. For objective indices (OEPIs), the correlation coefficient is 0.761 with equal weights and 0.782 with unequal weights.

The countries that improved their position among the European countries with respect to OEPI the most are Bulgaria, Croatia, Portugal, and Spain. Conversely, Greece shifted toward the worst countries. According to SEPI, Croatia and Hungary have improved their relative position most among the European countries concerning SEPI. In contrast, France, Germany, and Spain worsened their situation by increasing the Subjective Energy Poverty Index. We have listed those countries where the change in ranks is at least 10, but overall, there is no clear relationship between the changes in ranks with respect to the objective and subjective indicators. When we calculated the correlation between the changes in ranks in the objective and subjective indicators, we obtained as low as −0.126 for unweighted indicators and −0.100 for weighted indicators. The negative sign is worth noticing.

5. Discussion

Many of our results on energy poverty in EU countries align with the findings from other studies. Our research confirms the clear geographical variation in energy poverty across Europe (EU27), as Thomson and Snell [23] indicated based on data from 2007. Our findings reveal that energy poverty remains widespread in Bulgaria and Romania (as new EU members) and Cyprus in 2023. Similar conclusions were reached by Bollino and Botti [17], who found that Eastern and Southern European countries experience the highest levels of energy poverty, while Scandinavian countries are the least affected. Consistent with the results of Karpinska and Śmiech [36], who studied Central and Eastern European (CEE) countries, our research indicates that energy poverty impacts CEE countries differently—Bulgaria has the highest energy poverty rate, while the Czech Republic has the lowest.

We have found that certain EU countries, such as Bulgaria, are particularly vulnerable to energy poverty. It should be noted that previous studies on energy poverty in the EU28 countries from 2004 to 2019 also identified Bulgaria as having the highest poverty levels among European countries [16], despite a dramatic decrease in energy poverty over the years. Halkos and Gkampoura [16] further highlighted that Balkan countries, particularly Bulgaria, were significantly impacted by the economic crisis regarding energy poverty. Additionally, Bulgarian households continue to struggle with an excessive energy burden [9], and the Bulgarian authorities have made relatively minimal efforts to alleviate this issue [26].

Our analysis using a subjective approach to energy poverty has yielded intriguing insights. The increased percentage of the population reporting household energy deficiencies, including inadequate heating, from 2019 to 2023 may reflect a broader deterioration in social sentiments. This trend could be linked to the heightened public discussion about climate change, rising energy prices, and the need for changes in energy policy. Experience from Finland suggests that more extensive media coverage of climate action and future scenarios raises public awareness of climate threats [63]. Possibly, heightened public awareness and media focus on the energy transition may have contributed to more negative perceptions of certain aspects of energy poverty.

Specifically, the challenging situation regarding subjective energy poverty is observed in Bulgaria, Cyprus, and Greece, which aligns with the conclusions of Dubois and Meier [1]. Their analysis of energy service deprivation in the EU highlighted that energy poverty issues vary across countries. For instance, in Portugal and Cyprus, the primary challenge appears to be energy efficiency, whereas Bulgaria and Greece face difficulties related to both energy efficiency and affordability [1].

When explaining the mechanism of energy poverty in the studied countries, it is well established that there is a strong connection between it and household income (e.g., Romero et al., [64], for Spain; Hong and Park, [65], for South Korea; van Hove et al., [66], for various European countries). In the comparative analysis of energy poverty at the country level, it is worth referring to the overall level of development of each country and verifying the relationship between GDP and energy poverty. However, this relationship remains less clear, particularly in highly industrialized countries. Our analysis reveals a strong negative correlation between GDP per capita and energy poverty for both 2019 and 2023 across the OEPI and SEPI indicators. This finding supports the results of Zhou et al. [67], who examined household energy poverty in 82 countries within the Belt and Road Initiative and found a negative correlation between GDP and energy poverty. One might ask whether this effect merely reflects low incomes at the individual household level or if other factors are at play. The answer, however, is not straightforward. Our findings indicate that even after adjusting for income inequality (as measured by the Gini index), the relationship between inequality-adjusted GDP per capita and energy poverty becomes even stronger. For instance, in 2019, the correlation between GDP per capita and energy poverty was −0.526 for the OEPI with equal weights compared to −0.562 for inequality-controlled GDP. Similarly, after accounting for the inequality level, the correlation between GDP per capita and the SEPI with equal weights increased in absolute value from −0.426 to −0.486.

When analyzing the diversity of energy poverty across different countries, it is worth referring to classical economic factors related to the labor market. Unemployed persons typically have lower incomes, which can vary depending on a country’s unemployment benefits policy. Additionally, they may spend more time at home, leading to a higher consumption of electricity and other energy sources [68]. The relationship between unemployment and the risk of energy poverty at the household level has been demonstrated, for example, in Spain [61]. Our analysis also reveals a strong correlation at the national level. The unemployment rate is one of the indicators that constitute the structural energy poverty vulnerability (SEPV) index proposed by Recalde et al. [69].

In order to understand the conditions and explain the differences in the energy poverty rate in EU countries, it would be worthwhile to look at the policies implemented to counteract the exposure to energy poverty. Many countries have implemented some policies to reduce energy poverty through energy subsidies and other forms of financial assistance, which can also be used to mitigate energy poverty at the household level (see, e.g., [24,70]. Examples include Bono Social in Spain [59], Bonus Energia in Italy [71], and Energy Price Freeze and Compensation in Poland [72]. The effectiveness of energy subsidies has been investigated in countries such as Spain [59,60], with findings suggesting that their impact on reducing energy poverty at the individual level is relatively low. Our analysis shows that social spending per capita, measured in PPS, strongly negatively correlates with energy poverty indicators, with a stronger correlation observed for subjective measures than for objective ones. However, given the complex nature of energy poverty, this issue requires further detailed research.

To summarize our findings on energy poverty rates in EU countries, it is important to highlight that our results corroborate other studies showing that energy poverty persists not only in developing countries but also in developed ones [16]. When examining the causes of this situation, it is important to consider their varying nature, including the effects of support policies aimed at alleviating energy poverty in individual countries [24]. Researchers highlight that Eastern and Southern European countries are particularly vulnerable due to factors such as high levels of general income poverty, inefficient housing construction, insufficient infrastructure development, and various governance challenges [25]. Additionally, Stojilovska et al. [21] argue that the spatial division of energy poverty in Europe extends beyond a physical (infrastructural) divide; it is also a political division deeply rooted in economic and political contexts.

The results we obtained using different methods highlight the importance of selecting appropriate indicators in energy poverty studies. We have followed the conventional use of subjective and objective indicators; however, the selection of specific ones is open to debate [14]. Additionally, exploring and developing new measures that best capture energy poverty within a given research context is advisable. Researchers have pointed out [22] that insufficient statistical data and indicators can lead to overemphasizing the more easily recognizable symptoms of energy poverty, potentially overlooking other equally important aspects.

Halkos and Gkampoura [37], using panel data from 28 European countries for the period 2004–2019, show that GDP per capita and the use of fossil fuels are inversely associated with energy poverty conditions. Additionally, renewable energy sources and biofuels are inversely associated with the inability to keep homes adequately warm and the presence of leaks, damp, or rot in dwellings. However, they can also be considered a contributing factor to utility bill arrears. The push for decarbonization has accelerated the expansion of renewable energy sources, transforming energy production and consumption trends [73].

Using a composite measure to calculate household poverty indices, OEPI and SEPI, we applied two approaches: one with equal weights and another with unequal weights, the latter being determined based on the variability of indicators and their correlations. It was found that the weights did not differ significantly in the approach with unequal weights, likely due to the low correlation between the indicators used. Future research should consider conducting simulations with varying weights to assess their impact on the values of the energy poverty indices and, consequently, on country rankings.

This paper makes several significant contributions to the literature. Our research enhances the understanding of energy poverty in EU countries and enriches the methods used to measure it. Many researchers, such as Kryk and Guzowska [8], apply equal weights to all energy poverty indicators when creating an aggregated index. In our study, we adopt this approach but also propose incorporating unequal weights, allowing the data to ‘speak for themselves’. This aspect is a strength of our research and represents an innovative approach to using energy poverty measures. Taking into account unequal weights, we use the Betti and Verma [51] weighting method, which has not been used so far in the analysis of energy poverty at the EU country level. We use a method that accounts for the diversity of energy poverty indicators and addresses information redundancy by reducing the influence of strongly correlated indicators [43]. The advantage of our study lies in its use of both subjective and objective approaches, which have been applied by other authors [11,13,47,48], though typically in micro-level analyses and often limited to one or a few countries. In contrast, our study offers a comparative analysis of objective and subjective indicators at the country level, and it provides an international comparison across all the EU countries, addressing a significant research gap. Additionally, our study strengthens and updates the existing conclusions on the dimensions of energy poverty in EU countries between 2019 and 2023. This is especially important and valuable given the dynamic changes in the energy market caused by factors such as the COVID-19 pandemic [8] and recent global energy market disruptions [20].

A limitation of our study is the lack of pertinent statistical data [13,22], as well as the limited availability of studies measuring energy poverty. This constraint prevented us from using all the relevant energy poverty indicators, particularly space cooling-related ones, which is becoming an increasingly important aspect of energy poverty research [26]. Future research could focus on developing new measures of energy poverty based on available international data. The observed and persistent differences in energy poverty among EU countries warrant further investigation into their causes and the potential for convergence over time.

Our research on energy poverty has important implications for policymakers and practitioners addressing this issue. First, it offers new insights into methods for measuring energy poverty, enabling a more accurate assessment of its prevalence within the country. Second, we provide some recommendations for energy and social policies aimed at addressing the various manifestations of energy poverty. Our study highlights that modernizing homes to improve energy efficiency is a key strategy in combating energy poverty. This includes improving housing conditions, such as repairing leaking roofs, damp walls, floors, foundations, or rotting window frames. Additionally, it is crucial to address the need for financial support to cover energy costs for groups at risk of social exclusion. Exploring various solutions implemented in energy policies across different countries, such as voucher programs, is also advisable.

6. Conclusions

Energy poverty is a significant issue affecting both developing and developed countries. Due to its complexity and multidimensional nature, there are challenges in selecting the appropriate measures and methods to fully understand the issue.

To assess and compare household energy poverty in EU countries from 2019 to 2023, we considered two groups of indicators: the first group included objective indicators, and the second group included subjective indicators. We developed two indices to measure energy poverty for each group: the Objective Energy Poverty Index (OEPI) and the Subjective Energy Poverty Index (SEPI), respectively. To construct these indicators, we applied a composite measure approach using both equal weights and unequal weights for each indicator determined based on the variability of the indicators and their correlation with the other indicators within each group.

We used Eurostat, the country-level data from 2019 to 2023 in our analysis. Considering OEPI, we found that nearly all the countries experienced increased objective energy poverty in 2023 compared to 2019. According to the SEPI analysis, 12 countries experienced higher subjective energy poverty in 2023 compared to 2019, while 15 countries saw a decrease. However, 13 or 14 countries, depending on whether equal or unequal weights are used to calculate the energy poverty indices, experienced worsening conditions in the objective and subjective dimensions. This suggests that it is important to examine energy poverty separately in its objective and subjective dimensions, as they provide different insights.

The analysis of the subjective and objective approaches revealed that the indicators used to measure energy poverty can be influenced by both market phenomena (e.g., energy prices) and state policies. It is important to note that during the analyzed period, countries implemented various coping policies in response to the COVID-19 pandemic and the outbreak of the war in Ukraine, which may have impacted the energy market. These insights emphasize the need for the systematic monitoring of household energy poverty.