A Deep Dive into Institutional and Economic Influences on Poverty in Europe

Abstract

This study analyzed the evolution of the poverty rate between 2004 and 2023 in 29 European countries, using two categories of variables: institutional variables (Corruption Control Index and Rule of Law Index) and economic variables (unemployment rate, shadow economy, government expenditures on social protection and the Gini index). The methodology adopted included dynamic panel econometric models, applying a technique which involves the elimination of individual effects by a primary differencing of the variables and the use of the generalized method of moments (GMM) to evaluate the estimators. This methodology eliminates endogeneity caused by including the dependent variable with lag among the explanatory variables in the model. The results showed a strong negative correlation between the poverty rate and institutional variables, suggesting that improvements in governance and access to education and health resources are essential for poverty reduction. The shadow economy has also been identified as a poverty buffer, providing support in the absence of formal employment opportunities. The short-term impact of government expenditures on social protection was not significant, indicating the need for further analysis to better understand these dynamics. This research can make a significant contribution to the design of more effective public policies aimed at reducing shocks, reducing inequality and promoting sustainable economic growth.

1. Introduction

The risk of poverty is a very serious problem and a priority for all governments and societies, with deep and diverse consequences. It affects not only the individual well-being of citizens but also the overall socio-economic structure of all countries. The impact of poverty is beyond immediate economic issues, influencing access to education, health, adequate housing and employment opportunities, which can exacerbate inequalities and destabilize social cohesion.

In addition to individual impacts, poverty generates significant pressures on national economies, limiting the capacity for economic growth and slowing progress towards sustainable development goals. High poverty rates can also lead to an increase in welfare dependency, putting additional pressure on public budgets and potentially diminishing the resources available for other critical investments such as infrastructure or education.

Reducing the risk of poverty is essential not only to improve people’s living conditions but also to ensure long-term economic stability. This requires effective and targeted policies that address the root drivers of poverty and promote a more equitable distribution of resources. Combating poverty is thus not only an ethical obligation, but also a pragmatic strategy for sustainable development and long-term prosperity for society.

However, poverty remains a complex phenomenon, affected by economic fluctuations and structural inequalities in society. Economic downturns can rapidly erode household financial stability, sharply increasing the number of people at risk of poverty. In addition, inequalities in the access to education and health perpetuate poverty across generations, limiting social and economic mobility (Săseanu et al. 2024).

The objective of this study is to identify and analyze the impact of specific factors: institutional variables, such as the Corruption Control Index and the Rule of Law Index, and economic variables, including the unemployment rate, the Black Economy, government expenditure on social protection and the Gini index, on the poverty rate in 29 European countries.

The results of this study highlight that an integrated approach combining economic and institutional reforms is needed to address poverty effectively. Effective policies should promote improved coordination between economic and social initiatives to ensure an equitable distribution of resources and support sustainable development. This implies not only short-term interventions, but also long-term strategies that address the core causes of poverty and build sustainable resilience to economic shocks.

The risk of poverty in the European context therefore requires continued attention adapted to the specificities of each country and region. Through the application of advanced research methods and a detailed analysis of economic and institutional factors, this study provides a solid basis for the formulation and implementation of policies that minimize the risk of poverty and promote a more just and inclusive society.

The exploration of the effects of effective governance and economic, political and legislative stability on poverty brings into question how institutional and economic factors may intersect or may aggravate financial risks during crises.

This study uses indicators such as the Rule of Law Index and the Corruption Control Index to assess the impact of governance on economic stability. These indicators are essential in risk analysis, as weak governance and corruption can increase vulnerability to economic shocks and destabilize financial markets. The results of the study therefore suggest that improvements in governance contribute not only to poverty reduction but also to a more stable environment for financial markets.

Moreover, the research emphasizes the role of the shadow economy as a social shock absorber during economic crises. This is particularly relevant for analyzing capital market risks, as shadow economic activities can mitigate the negative impact of financial crises on vulnerable populations, but also introduce elements of unpredictability into the economy.

This study also emphasizes the need for an integrated strategy that combines economic and institutional factors to combat poverty and thus minimize the risks associated with economic and financial instability. By promoting improved coordination between economic and social initiatives, more effective public policies can be created, oriented towards the prevention and management of financial crises and the promotion of sustainable and inclusive economic growth (Batrancea et al. 2020).

This study brings an innovative perspective to the existing literature on poverty by introducing a variety of economic and institutional factors into the analysis. This approach reveals how the interplay between the unemployment rate, the shadow economy, government spending on social protection, the control of corruption and the rule of law influence the risk of poverty. A novel aspect is the analysis of the long-term impact of social protection spending on poverty. By exploring how and when government interventions become effective, the research clarifies why some measures may have delayed or cumulative results. Introducing the underground economy, which plays an important role in modulating the effectiveness of redistributive policies, into the analysis has allowed the assessment of the hidden impact of the shadow economy on poverty and inequality, opening new possibilities for exploring the complex interactions between the formal and shadow economy.

The methodology, which uses an advanced econometric model applied to a comprehensive panel dataset for 29 European countries over a 20-year period, increases the accuracy and relevance of the results. This approach not only tests hypotheses in a robust way, but also provides a good support for empirically based public policy recommendations.

Furthermore, the research emphasizes the importance of adapting policies to specific regional contexts. By analyzing how institutional stability and governance efficiency influence the implementation of economic policies, the study highlights the need for customized strategies that respond to the diverse needs of different regions, thereby improving the effectiveness of efforts to combat poverty and social exclusion. This study thus not only adds a new dimension to the understanding of poverty, but also opens new directions for future research, providing a solid basis for the formulation of more effective public policies.

Although, the research focuses mainly on the analysis of the actual poverty rate, we believe that this approach is closely linked to the risk of poverty, as highlighting and understanding the determinants of the current poverty level provides essential information for anticipating and preventing its increase in the future. The use of dynamic panel models allowed us not only to capture the present state, but also to identify short- and long-term causal relationships, which provides an appropriate framework for analyzing the risk of poverty in a broad sense.

We therefore believe that our paper contributes to the literature on poverty risk by empirically substantiating some of the institutional and economic factors that may influence poverty risk and aligns with the overall theme of the journal. We have clarified and emphasized this link more explicitly in the introduction, adding additional insights.

2. Literature Review

This section highlights significant studies that use various econometric techniques to analyze the impact of different factors, both institutional and economic, thus providing a solid background for our analysis.

Recent scientific research shows that economic crises can accentuate income inequality and poverty risk, disproportionately affecting vulnerable groups in society (Dai et al. 2023). A good example is Greece, where the response to economic crises has had severe consequences for disadvantaged citizens, amplifying poverty and inequality during that period (Petrakos et al. 2023). These observations are in line with the literature suggesting that austerity measures, commonly adopted during recessions, can worsen income distribution and intensify the risk of social exclusion (Bodea et al. 2021; De Beer 2012; Munteanu et al. 2020). It is also observed that reducing inequality can play an important role in tackling global poverty and reducing the shadow economy (Haruna and Alhassan 2022). One interesting study shows that a 1% annual reduction in the Gini index could decrease the global rate of extreme poverty by about 89 million people by 2030 (Lakner et al. 2022). These results underline the importance of adopting policies that aim not only at economic growth but also at an equitable distribution of the benefits of that growth.

In addition, several studies have revealed a critique of traditional measures to reduce poverty and inequality, as they often fail to capture the real dynamics of needs in different economic and social contexts. Au (2023) argues for a cost-of-living approach to poverty measurement, which can more accurately reflect actual household deprivation than relative methods based on a percentage of median income. The analysis of the impact of economic crises on poverty and inequality shows that while these crises directly influence poverty, income redistribution policies and fiscal measures can moderate these effects. Research conducted by Demsou (2023) indicates that a deprivation-based decomposition of the Gini index from multidimensional poverty can provide a deeper understanding of the impact of government policies on inequality. This methodology has allowed a more detailed assessment of the impact of different components of poverty, such as health and access to safe potable water, and emphasizes the importance of directing social policies to address these specific disparities.

Recent scientific research underlines the significant impact of government expenditures on poverty, illustrating how they can influence economic growth and reduce social disparities in the short and long term. One significant study in this regard is by Anderson et al. (2018), which highlights a diversity of results in the literature on the relationship between government expenditure and income levels. Their analysis shows that factors such as the type of expenditure and the geographical region determine the magnitude and direction of the impact of this expenditure. Cardoso et al. (2023) found that expenditures on social protection had a significant positive impact on GDP in 42 countries, suggesting that investments in social protection not only support economic growth but also contribute to the reduction of inequality, in contrast to the smaller effects of total government spending.

In Romania, Mehmood and Sadiq (2010) identified a cointegrating relationship between government spending and poverty, emphasizing the existence of long-run and short-run effects of fiscal policies on poverty. This reflects the importance of a prudent allocation of resources to effectively combat poverty. Yusoff et al. (2023) used a nonlinear autoregressive distributed lag (ARDL) model to examine how government expenditure on development influences poverty, finding that a reduction in government expenditure on development can lead to an increase in poverty in the long run, emphasizing the importance of efficiency in the use of public funds.

At the European level, the European Pillar of Social Rights (EPSR) has been identified as a focal point for the effective coordination and implementation of social policies at the EU level, emphasizing the essential role of government expenditure in promoting sustainable economic growth and human capital development, especially in education and health (Hacker 2023; Munteanu et al. 2024b). This perspective emphasizes the strategic importance of investing in essential services, which are important for the long-term reduction of poverty and inequality.

Human development and poverty reduction are significantly influenced not only by economic growth but also by its distribution at the local level. Several studies show that an increase in GDP does not necessarily guarantee an improvement in quality of life or a reduction in economic disparities. Naveed and Gordon (2024), who analyzed the Human Development Index, highlighting significant regional disparities within the same national economy, emphasized the need for decentralization and a more equitable distribution of resources.

In the European Union, differences between member states and internal regions accentuate these inequalities, influencing the risk of poverty. Similarly, Asmara et al. (2024) found that in Indonesia, between 2015 and 2020, economic growth did not significantly influence the Human Development Index, but rather the poverty rate, suggesting that economic policies need to address more aspects than just GDP growth.

The quality of institutions and their impact on poverty and financial inclusion are also intensely debated. Aracil et al. (2022) emphasize that countries with stronger institutions benefit more from access to financial services, with a stronger positive impact on poverty reduction in countries with a strengthened institutional framework. Other studies (Sanga and Aziakpono 2023; Munteanu et al. 2024a) show that political stability, regulatory quality, and the control of corruption are essential for financial sector development and access to credit, especially in middle- and high-income economies. In addition, Behnezhad et al. (2021) demonstrate that economies with effective governance and strong institutions more effectively translate economic growth into poverty reduction, identifying an institutional threshold below which economic growth does not yield significant poverty benefits.

Thus, poverty reduction policies must include not only economic measures, but also institutional reforms to increase government transparency and efficiency (Gallego-Álvarez et al. 2021). In this context, the European Pillar of Social Rights (EPSR) plays a central role in the effective coordination and implementation of social policies at the EU level, highlighting the importance of establishing effective economic and government structures in reducing economic inequality and promoting sustainable economic growth.

The rule of law is essential for economic development and poverty reduction, creating a stable and predictable framework that facilitates the functioning of markets and protects individual rights. Dessie (2014) shows that the absence of the rule of law excludes citizens from economic and political processes, perpetuating poverty and inequality. Implementing policies that respect the rule of law can improve access to justice, reduce corruption and support economic inclusion. However, Versteeg and Ginsburg (2017) emphasize the difficulties of measuring the rule of law, pointing out that existing indicators often reflect public perceptions of corruption more than actual law enforcement.

The literature on poverty in the European Union highlights the interconnections between economic and institutional factors. Barbero and Rodríguez-Crespo (2022) analyze how digital technologies and good governance can reduce social exclusion, although regional disparities persist that require specific territorial policies. Liotti (2024) counters hypotheses that suggest economic freedom would reduce poverty, showing that reduced government expenditure may increase the risk of poverty. Ngubane et al. (2023) find that while economic growth may reduce poverty in the long run in South Africa, unemployment and negative economic shocks may exacerbate the problem.

In terms of poverty reduction strategies, Fonseca et al. (2024) emphasize the role of microfinance as an effective tool if supported by appropriate entrepreneurship policies, while Kou and Yasin (2024) identify political instability and corruption as factors contributing to poverty in emerging economies. On the other hand, Copeland (2023) criticizes poverty management in the European Union, pointing out that the hybrid of intergovernmental and national measures limits the effectiveness of interventions and leads to slow and uneven progress across member states. These findings suggest that a more integrated and regionally oriented approach could increase the effectiveness of policies to combat poverty and social exclusion.

The study by Tafran et al. (2020) shows that in Malaysia, reducing poverty and unemployment contributes to increases in the life expectancy, but the effects vary depending on the economic structure and regional policies. Similarly, Umoh (2024) in Nigeria, shows that poverty exacerbates unemployment rates by limiting access to education and job opportunities, perpetuating the cycle of poverty and inequality. In Europe, Signoret et al. (2020) note that regions open to international trade have lower poverty rates, although the benefits are not evenly distributed. Dai et al. (2023) emphasize the importance of education and active labor force participation in reducing unemployment and poverty, especially in areas with low human capital.

Mansi et al. (2020) examine poverty in the European Union and the Western Balkans, identifying income inequality and unemployment as key factors, with regional differences in the impact of GDP growth and per capita income. The relevance of education and governance is emphasized as decisive for long-term poverty reduction.

While there are numerous studies dealing with the influence of individual factors on poverty (such as unemployment, government spending or corruption), the literature is less focused on the combined effects of economic and institutional factors, analyzed simultaneously within a unified methodological framework. In particular, studies that use dynamic panel models to capture these interactions in a way that distinguishes between short-run and long-run effects are lacking. There is also a need for comparative research applied on extensive European datasets, taking into account regional diversity and institutional context.

The present study therefore seeks to address these gaps by providing an integrated perspective on how economic and institutional factors influence poverty risk in the European Union, using an advanced econometric approach and relevant data from 2004 to 2023.

3. Data and Methodology

For the analysis of the relationship between poverty risk and economic and institutional factors in the European context, a two-stage econometric modeling strategy was adopted to capture both the dynamic relationships and to separate short-run and long-run effects. In the first stage, GMM (generalized method of moments) dynamic panel models were estimated, suitable to control for the endogeneity of the explanatory variables and to isolate the influence of country-specific effects. In the second stage, an ARDL/PMG (autoregressive distributed lag/pooled mean group) model was applied, suitable for analyzing long-run equilibrium relationships between variables in the presence of first-order integrated series. This combined approach is appropriate because it provides a robust framework for testing hypotheses on the influence of institutions and economic conditions on poverty in a context characterized by structural heterogeneity and complex time trends at the EU level.

The European countries included in this analysis are Belgium, Bulgaria, Czech Republic, Denmark, Germany, Estonia, Ireland, Greece, Spain, France, Croatia, Italy, Cyprus, Latvia, Lithuania, Luxembourg, Hungary, Malta, Netherlands, Austria, Poland, Portugal, Romania, Slovenia, Slovak Republic, Finland, Sweden, Norway and Switzerland.

The choice of the 29 European countries for this study is justified by several aspects related to data quality and availability, as well as the need to obtain a broad perspective on institutional and economic influences in different European contexts. By using the data available on EUROSTAT, this study benefits from access to standardized and high-quality information that is essential for the application of a robust methodology such as the dynamic panel econometric model. This type of model, although not allowing for direct cross-country comparative analysis (individual specific effects are removed to solve the endogeneity problem induced by the inclusion of the dependent variable with lag among the regressors), is very well suited to identify the short- and long-run effects of various economic and institutional policies and conditions on poverty.

By including a broad range of countries, the study captures the diversity of governance, economic and social conditions in Europe, allowing for a thorough understanding of how these conditions influence poverty. This approach is important in illustrating how different institutional and economic frameworks can have different impacts on poverty, highlighting the need for a differentiated approach to public policy formulation.

In addition, the use of an extensive and representative database ensures that the findings of the study are well grounded and can have significant implications for the development of poverty reduction strategies at the European level. Thus, the selection of these countries not only enriches the quality of the analysis, but also contributes to creating a more coherent and implementable framework for the development of effective public policies oriented towards combating social exclusion and promoting economic growth that benefits all segments of society.

3.1. Data

The indicators selected for this study are detailed in

Table 1. The symbol, the name, the source and the meaning of the variables used in the analysis.

Symbol	Name	Source	Meaning
CI	Corruption Control Indicator	Worldwide Governance Indicators (WGI) https://www.worldbank.org/en/publication/worldwide-governance-indicators (accessed on 18 February 2025)	It measures perceptions of the degree to which public authorities use government functions for private purposes, including bribery, embezzlement and elite influence over state institutions. Its range is from −2.5 to 2.5, where −2.5 is a perception of rampant corruption and 2.5 is a perception of clean and efficient governance and a more transparent business environment.
ExSP	Government Expenditure for Social Protection (% GDP)	Eurostat https://doi.org/10.2908/GOV_10A_EXP (accessed on 18 February 2025)	Represents government expenditure allocated to social protection (as % of total government expenditure), i.e., resources allocated to pensions, unemployment benefits, family subsidies and other forms of social support. This indicator reflects the government’s priorities in allocating resources for the well-being of the population.
HDI	Human Development Index	UNDP https://hdr.undp.org/data-center/human-development-index#/indicies/HDI (accessed on 18 February 2025)	The Human Development Index (HDI), calculated by the UN, measures the overall level of human development in a country based on indicators related to health, education and living standards. It ranges from 0 to 1. A score close to 1 indicates a high level of human development.
PR	Poverty Rate (%)	Eurostat https://ec.europa.eu/eurostat/databrowser/product/page/ILC_LI02 (accessed on 18 February 2025)	The percentage of the population living below the poverty line, showing the proportion of the poor in the total population.
RLI	Rule of Law Indicator	Worldwide Governance Indicators (WGI) https://www.worldbank.org/en/publication/worldwide-governance-indicators (accessed on 18 February 2025)	Reflects the degree to which citizens and economic entities trust and respect the rules of society, including the efficiency of the judicial system, the enforcement of contracts and the protection of property rights. It is scaled from −2.5 to 2.5. A higher score suggests greater respect for the rule of law, including the effective protection of property rights and equal enforcement of laws.
ShE	Shadow Economy (% GDP)	European Parliament https://www.europarl.europa.eu/RegData/etudes/STUD/2022/734007/IPOL_STU(2022)734007_EN.pdf (accessed on 18 February 2025)	The estimation of the shadow economy as a percentage of gross domestic product, according to European Parliament data, representing unrecorded and untaxed economic activities.
GINI	Gini Index (%)	World Bank https://data.worldbank.org/indicator/SI.POV.GINI (accessed on 18 February 2025)	It is an indicator of income inequality in a population. Its value varies between 0 and 1 (or between 0 and 100 if expressed as a percentage). A value of 0 indicates a perfectly equal distribution of income (all individuals have the same income), while a value of 1 (or 100%) indicates extreme inequality (one individual has all of the income and the others have none). It is widely used to assess economic inequality in a country and to compare income distribution.
UR	Unemployment Rate (%)	Eurostat https://ec.europa.eu/eurostat/databrowser/product/page/UNE_RT_M (accessed on 18 February 2025)	The unemployment rate as a percentage of all employed persons shows the share of the unemployed out of the active population.

Note: The authors’ overview.

. These indicators, collected for the 29 European countries, covered the period from 2004 to 2023.

The selection of the indicators included in our econometric model derives from a combination of their theoretical relevance and empirical support from previous research. Thus, the institutional variables—the Corruption Control Index and the Rule of Law Index—were chosen because they reflect the quality of governance, which recent studies (Aracil et al. 2022; Behnezhad et al. 2021) have shown to influence access to public services, income redistribution and financial inclusion, all of which have direct effects on the risk of poverty. Economic variables—the unemployment rate, the shadow economy, social protection expenditures and the Gini index—are also frequently used in the literature (Lakner et al. 2022; Cardoso et al. 2023; Mansi et al. 2020) as explanatory factors of poverty, reflecting dimensions such as income inequality, labor market exclusion and the effectiveness of social policies.

In general, GDP per capita could also be included in the explanatory variables. However, since in this study the variables include the Human Development Index (HDI), which takes into account the GDP per capita indicator, this indicator was not separately highlighted. As for HDI, it was chosen as a control variable because of its composite nature, which allows for capturing a more nuanced perspective on human development.

The nature of the data series is described in

Table 2. Unit root tests in panel for analyzed variables.

	CI	ExSP	HDI	PR	RLI	ShE	GINI	UR
Null: Unit root (assumes common unit root process)
Levin, Lin and Chu t	I(1)	I(0)	I(0)	I(0)	I(1)	I(0)	I(0)	I(0)
Breitung t-stat	I(1)	I(1)	I(1)	I(1)	I(1)	I(1)	I(0)	I(0)
Null: Unit root (assumes individual unit root process)
Im, Pesaran and Shin W-stat	I(1)	I(1)	I(1)	I(0)	I(1)	I(1)	I(0)	I(0)
ADF—Fisher Chi-square	I(1)	I(1)	I(1)	I(0)	I(1)	I(1)	I(0)	I(0)
PP—Fisher Chi-square	I(1)	I(1)	I(0)	I(0)	I(1)	I(0)	I(0)	I(1)
Conclusion	I(1)	I(1)	I(1)	I(0)	I(1)	I(1)	I(0)	I(0)

Notes: For Breitung t-stat test, exogenous variables: individual effects and individual linear trends. For all of the other tests, exogenous variables: individual effects. Automatic lag length selection based on AIC: 0 to 3. For all of the outcomes p(H₀) is evaluated at the 0.05 level. Source: EViews estimates.

. Although the tests are not fully concordant, the conclusions presented in the last row of Table 2 are accepted as hypotheses regarding the nature of the series.

The tests reject the unit root null hypothesis (as a joint process, or as individual processes) at the p < 10⁻⁴ threshold for the PR (poverty rate), Gini index and UR (unemployment rate) series. For the other data series, the tests do not reject the unit root null hypothesis, and all tests reject the unit root for the series calculated by differentiation.

The uncertainty generated by the non-concordance of the unit root tests in the panel does not generate problems for the econometric modeling, as long as all tests reject the non-stationarity hypothesis for the differentiated series, i.e., the series are not I(2), and the models used are of the PMG/ARDL (pooled mean group/autoregressive distributed lag models), or GMM/DPD (generalized method of moments/dynamic panel data) type. Some descriptive analysis elements for the series involved in the construction of the econometric models are detailed in

Table 3. Elements of descriptive analysis.

Series	Range	Overall Mean	Overall Variance	Between Variance		Within Variance
				Value	%	Value	%
PR	[8.6%, 26.4%]	16.099	14.580	12.996	89.1%	1.584	10.9%
GINI	[23.2%, 41.3%]	31.209	13.678	11.906	87.0%	1.772	13.0%
UR	[2%, 27.8%]	8.070	17.276	8.779	50.8%	8.497	49.2%
CI	[−0.38, 2.46]	1.045	0.649	0.622	95.8%	0.027	4.2%
d(CI)	[−0.34, 0.38]	−0.004	0.0076	0.0004	5.3%	0.0072	94.7%
ExSP	[7.5%, 27.1%]	16.408	16.244	13.791	84.9%	2.453	15.1%
d(ExSP)	[−3.8%, 4.9%]	0.047	1.068	0.028	2.6%	1.040	97.4%
HDI	[0.74, 0.97]	0.883	0.002	0.0017	85.0%	0.0003	15.0%
d(HDI)	[−0.01, 0.03]	0.003	1.65 × 10−5	7.74 × 10−7	4.7%	1.58 × 10−5	95.3%
RLI	[−0.19, 2.12]	1.131	0.379	0.361	95.2%	0.018	4.8%
d(RLI)	[−0.317, 0.248]	−0.0003	0.0051	0.0003	5.9%	0.0048	94.1%
ShE	[5.5%, 35.3%]	18.694	54.181	50.657	93.3%	3.614	6.7%
d(ShE)	[−1.6%, 3.2%]	−0.282	0.437	0.030	6.9%	0.407	93.1%

Source: EViews calculations by authors. Note: The formulas used are the following: Overall mean: $\overline{\mathrm{x}}={\displaystyle \frac{1}{\mathrm{N}\mathrm{T}}}{\sum }_{\mathrm{i}}{\sum }_{\mathrm{t}}{\mathrm{x}}_{\mathrm{i}\mathrm{t}}$. Between variance (variation between individuals): ${\mathrm{s}}_{\mathrm{B}}^2={\displaystyle \frac{1}{\mathrm{N}-1}}{\sum }_{\mathrm{i}}{\left({\overline{\mathrm{x}}}_{\mathrm{i}}-\overline{\mathrm{x}}\right)}^2,\mathrm{a}\mathrm{n}\mathrm{d} {\overline{\mathrm{x}}}_{\mathrm{i}}={\displaystyle \frac{1}{\mathrm{T}}}{\sum }_{\mathrm{t}}{\mathrm{x}}_{\mathrm{i}\mathrm{t}}$. Within variance (variation within individuals, over time): $s_W^2={\displaystyle \frac{1}{NT-1}}{\sum }_i{\sum }_t{\left(x_{it}-{\overline{x}}_i\right)}^2$. Overall variance (variation over time and individuals): $s_O^2={\displaystyle \frac{1}{NT-1}}{\sum }_i{\sum }_t{\left(x_{it}-\overline{x}\right)}^2$.

. The statistical description was done through indicators that provide a more detailed analysis of the structure of the variability of the data, especially in spatial or panel contexts (by years or countries), because through them one can distinguish between variance and within variance, one can highlight stability or instability over time and one can identify the dominant type of variation.

For the stationary series poverty rate (PR) and Gini index (GINI), almost 90% of the total dispersion is explained by between variance (i.e., it is determined by differences between European countries), and for the unemployment rate (UR), the total dispersion is split approximately equally between individuals and within variance (50.8% vs. 49.2%).

For non-stationary series, descriptive analysis may lack statistical significance because the mean and dispersion computed for within-level series do not reflect stable characteristics of the analyzed processes. For this reason, Table 3 presents elements of descriptive analysis and dispersion decomposition for both in-level (non-stationary) and first-difference (stationary) series. In the case of non-stationary series transformed through simple differencing, most of the total dispersion is accounted for by the within-country variance.

3.2. Methodology

The impact of institutional and economic factors on the risk of poverty, measured by the poverty rate (%), was evaluated using econometric models. The analysis employed dynamic panel data models as the methodological approach (Baltagi 2021).

The dynamic panel data model is as follows:

PR_i,t = φPR_i,t−1 + a₁x_1:i,t + … + a_kx_k:i,t + γ_i + δ_t + ε_i,t, i = 1, …, N; t = 1, …, T

(1)

where PR (i.e., poverty rate) is the endogenous variable, such that, PR_i,t denotes the observation t for the country i, x₁, …, x_k are the explanatory variables (x_j:i,t means the value of variable x_j at time t, for the country i), γ_i is the individual-specific effect (time-invariant), δ_t is the time-specific effect (invariant for cross-section) and ε_i,t is the idiosyncratic error variable. The number of countries (cross-sectional units) is denoted by N = 29, and the number of time periods (spanning from 2004 to 2023) is denoted by T = 20.

The other coefficients are symbolized by φ (the coefficient of the lagged dependent variable) and a₁, …, a_k (coefficients of exogenous regressors).

Let us assume that the idiosyncratic error (ε_i,t) is IID(0, σ_ε²) and γ_i is the individual random effect, also IID(0, σ_γ²), independent from ε_i,t. The model (1), for PR_i,t−1 is written as follows (Equation (2)):

PR_i,t−1 = φPR_i,t−2 + a₀ + a₁x_1:i,t−1 + … + a_kx_k:i,t−1 + γ_i + δ_t−1 + ε_i,t−1,

(2)

which means that PR_i,t−1 is a function of γ_i. Then, in Equation (1), if γ_i are individual random effects, then PR_i,t−1 is correlated with the error variable so that the OLS estimators are biased and non-consistent. If γ_i are individual fixed effects, then, according to Baltagi (2021), the OLS estimators are also biased and non-consistent. Under these conditions, the estimation of Equation (1) was carried out using the technique developed by Arellano and Bond (1991), which involves the elimination of individual effects (γi\gamma_i) through the first differencing of Equation (1), followed by the application of the generalized method of moments (GMM) using PR_i,t−2PR_{i,t−2} as instrumental variables.

4. Results and Discussions

This study on the evolution of the poverty rate between 2004 and 2023 for the 29 European countries was carried out by grouping the selected variables into two categories: institutional (Corruption Control Index and Rule of Law Index) and economic (unemployment rate, shadow economy, government expenditures on social protection, and the Gini index).

4.1. The Impact of Institutional Variables on the Poverty Rate

The Corruption Control Index and the Rule of Law Index were included as institutional variables, while the Human Development Index was introduced as a control variable in the model. This is because the Human Development Index is a complex composite index that is based on three basic dimensions of human development: life expectancy at birth, education (expected and mean years of schooling) and economic development (gross national income per capita).

To assess potential collinearity among the explanatory variables in the model, linear correlation coefficients (Pearson) were computed. The results are presented in

Table A1. Pairwise correlation (Pearson) between explanatory variables. All of the variables in the level.

	CI	ExSP	HDI	RLI	ShE	GINI	UR
CI	1	0.4639	0.7907	0.9527	−0.7671	−0.4097	−0.3616
ExSP	0.4639	1	0.4820	0.4284	−0.4876	−0.2895	0.1046
HDI	0.7907	0.4820	1	0.7914	−0.7903	−0.4585	−0.3379
RLI	0.9527	0.4284	0.7914	1	−0.7876	−0.4316	−0.3714
ShE	−0.7671	−0.4876	−0.7903	−0.7876	1	0.3992	0.2989
GINI	−0.4097	−0.2895	−0.4585	−0.4316	0.3992	1	0.3890
UR	−0.3616	0.1046	−0.3379	−0.3714	0.2989	0.3890	1

Note: All of the coefficients are statistically significant, at the 1% level, according to t-test.

. Only one coefficient exceeds 0.95, indicating a high linear correlation between the Corruption Control Index (CI) and the Rule of Law Index (RLI). Additionally, the Human Development Index (HDI) shows relatively strong linear correlations with both CI and RLI, each at a value of 0.79. Since all variables are non-stationary and integrated of order 1, linear correlation coefficients were also calculated for the first-differenced series.

The correlations between d(CI), d(RLI) and d(HDI) do not exceed 0.25 (according to

Table A2. Pairwise correlation (Pearson) between explanatory variables. Non-stationary variables in first difference.

	dCI	dExSP	dHDI	dRLI	dShE	GINI	UR
dCI	1	0.0239	−0.0159	0.2505	0.0380	0.0412	−0.0363
dExSP	0.0239	1	−0.3432	−0.0647	0.4690	0.0687	0.0499
dHDI	−0.0159	−0.3432	1	0.1357	−0.4164	0.0479	0.0309
dRLI	0.2505	−0.0647	0.1357	1	−0.0980	−0.0086	−0.0708
dShE	0.0380	0.4690	−0.4164	−0.0980	1	−0.0135	−0.0075
GINI	0.0412	0.0687	0.0479	−0.0086	−0.0135	1	0.3890
UR	−0.0363	0.0499	0.0309	−0.0708	−0.0075	0.3890	1

Note: Coefficients in italics are statistically significant at the 5% level, according to the t-test. Source: Calcule în EViews pe baza datelor descrise mai sus.

).

Under these conditions, the dynamic panel data model, for the relationship between poverty rate and institutional variables is the following:

$$\mathrm{P}{\mathrm{R}}_{\mathrm{i},\mathrm{t}}=\underset{\left(0.0286\right)}{0.6013}\cdot \mathrm{P}{\mathrm{R}}_{\mathrm{i},\mathrm{t}-1}-\underset{\left(0.5743\right)}{0.9817}\cdot \mathrm{d}\left(\mathrm{R}\mathrm{L}{\mathrm{I}}_{\mathrm{i},\mathrm{t}}\right)-\underset{\left(0.3433\right)}{1.2118}\cdot \mathrm{d}\left(\mathrm{C}{\mathrm{I}}_{\mathrm{i},\mathrm{t}}\right)-\underset{\left(3.4121\right)}{9.1235}\cdot \mathrm{d}\left(\mathrm{H}\mathrm{D}{\mathrm{I}}_{\mathrm{i},\mathrm{t}}\right)$$

(3)

(under estimators, in the round brackets, are the standard errors).

The estimating output is detailed in Appendix B.1. In Equation (3), the symbols are the following:

PR_i,t—poverty rate for country i, in year t;

RLI_i,t—Rule of Law Index for country i, in year t;

CI_i,t—Corruption Control Index for country i, in year t;

HDI_i,t—Human Development Index for country i, in year t;

d—the differentiation operator, so that dy_i,t = y_i,t − y_i,t−1.

The coefficient of the d(RLI) variable is statistically significant at the 10% level (according to a two-sided Student’s t-test). The other coefficients are significant at 1%. The probability attached to the null hypothesis in the over-identifying test (H₀: the over-identifying restrictions are valid) is p(J-stat = 23.073) = 0.5733, considerably above the standard threshold of 5%. This means that the instrument variables (higher in number than the estimated variables) are valid and uncorrelated with the other variables in the model.

The two step Arellano–Bond estimators are unbiased and consistent, if the errors ε_i,t are serially uncorrelated (Arellano and Bond 1991, p. 278). For the model (3), the Arellano–Bond test (Appendix C.1) does not reject the hypothesis of first-order serial correlation (at 0.0035 level) and rejects second-order serial correlation (at 0.2836 level). This means that the model in the specification described by Equation (3) is well founded.

The size of the estimators presented in Equation (3) is influenced by the mean and standard deviation of each variable, so they cannot be directly compared. Specifically, all series were transformed to have a mean of zero and a standard deviation of one. The estimators are presented in

Table 4. Standardized coefficients for the model of the institutional variables’ impact on the poverty rate.

Variables	Coefficient	Standardized Coefficient
PR(−1)	0.601335	0.603095
D(RLI)	−0.981680	−0.018399
D(CI)	−1.211813	−0.027808
D(HDI)	−9.123493	−0.009723

Note: For standardized coefficients, all of the variables were transformed to have a mean of zero and a standard deviation of one. Source: EViews-12 estimates.

.

The model reveals a strong inertial structure of the poverty rate. The autoregressive coefficient is 0.601, statistically significant, positive and sub-unitary, which reveals the stability of the model.

For the 29 European countries analyzed, the poverty rate is negatively associated with the Rule of Law Index. This is a theoretically anticipated result, as more effective governance has positive effects on economic growth and social equity and a better ability to implement effective economic and social policies. Also, economic, political and legislative stability encourages investment, which leads to increased employment opportunities, with a positive effect on poverty rate reduction. This result is also in agreement with Kaufmann et al. (2010) and Gramatikov et al. (2021).

Poverty rate is also negatively correlated with Corruption Control Index. This is because corruption is associated with the inefficient and preferential allocation of public resources, reduces the access of the poor to economic opportunities and discourages private investment. Similar results have been obtained by Gupta et al. (2002) and Khan (2022), among others.

Poverty rate is negatively associated with the Human Development Index. The Human Development Index aggregates information on life expectancy, education and economic development. Better access to quality health care results, in addition to increased life expectancy, in an improvement in the general health status of the population, with effects on labor force participation, productivity and income, and, derived from these, on the reduction of the poverty rate. Education contributes to increasing labor market opportunities, raising incomes and reducing economic and social inequalities. Economic development generates jobs and other economic opportunities, increasing budget revenues which allows the application of redistributive policies to reduce poverty. The result is in line with the UNDP, United Nations Development Programme (UNDP) (2024), Beja (2021), Hasan and Putri (2022), Lestari et al. (2022), Cashin et al. (2001) and so on.

As a dimension (standardized coefficients, in Table 4), the impact of the Corruption Control Index change is double the impact of the dynamic Rule of Law Index and triple the impact of the Human Development Index. The probability that this size ratio between the negative effects induced by the mentioned factors on the poverty rate is 0.9864 (according to the Wald methodology for testing restrictions between coefficients).

4.2. The Impact of Economic Variables on the Poverty Rate

As economic variables, we used the unemployment rate, shadow economy, government expenditure for social protection and Gini Index. We also included the Human Development Index as a control variable in the model. As in the previous model (Equation (3)), in order to avoid potential multicollinearity problems, non-stationary variables were included in the model by simple differentiation. The first dynamic panel data model for the relationship between poverty rate and economic variables is as follows (Equation (4)):

$$\begin{gathered} {PR}_{i,t}=\underset{\left(0.0501\right)}{0.6050}\cdot \mathrm{P}{\mathrm{R}}_{\mathrm{i},\mathrm{t}-1}+\underset{\left(0.0137\right)}{0.0366}\cdot {\mathrm{U}\mathrm{R}}_{\mathrm{i},\mathrm{t}-1}+\underset{\left(0.0499\right)}{0.1205}\cdot {\mathrm{G}\mathrm{I}\mathrm{N}\mathrm{I}}_{\mathrm{i},\mathrm{t}} \\ -\underset{\left(0.0931\right)}{0.2118}\cdot {\mathrm{d}\left(\mathrm{S}\mathrm{h}\mathrm{E}\right)}_{\mathrm{i},\mathrm{t}}-\underset{\left(0.0448\right)}{0.0029}\cdot {\mathrm{d}\left(\mathrm{E}\mathrm{x}\mathrm{S}\mathrm{P}\right)}_{\mathrm{i},\mathrm{t}}-\underset{\left(3.5667\right)}{40.5655}\cdot {\mathrm{d}\left(\mathrm{H}\mathrm{D}\mathrm{I}\right)}_{\mathrm{i},\mathrm{t}} \end{gathered}$$

(4)

(under estimators, in the round brackets, are the standard errors).

The estimating output is detailed in Appendix B.2. In Equation (4), the symbols are the following:

PR_i,t—poverty rate for country i, in year t;

UR_i,t—unemployment rate for country i, in year t;

GINI_i,t—Gini index for country i, in year t;

ShE_i,t—shadow economy for country i, in year t;

ExSP_i,t—government expenditure for social protection for country i, in year t;

HDI_i,t—Human Development Index for country i, in year t;

d—the differentiation operator, so that dy_i,t = y_i,t − y_i,t−1.

The coefficient of d(ExSP_i,t), i.e., the dynamic of government expenditure for social protection, is not statistically significant (the probability attached to the null hypothesis H0, that the parameter does not differ significantly from zero, is 0.9481). The other coefficients are statistically significant and have the expected sign. Also, the probability attached to the null hypothesis in the over-identification test is p(J-stat = 24.382) = 0.3829, considerably above the standard threshold of 5%.

To explain the non-intuitive result regarding the lack of impact of the dynamics of government expenditure for social protection on the poverty rate, an ARDL/PMG model was estimated to distinguish between long-run and short-run relationships between the variables.

The long-run equilibrium relationships are as follows (Equation (5)):

$$\begin{gathered} {PR}_{i,t}=-\underset{\left(0.0503\right)}{0.1092}\cdot {\mathrm{E}\mathrm{x}\mathrm{S}\mathrm{P}}_{\mathrm{i},\mathrm{t}} \\ {\mathrm{E}\mathrm{x}\mathrm{S}\mathrm{P}}_{\mathrm{i},\mathrm{t}}=\underset{\left(0.0951\right)}{0.1990}\cdot {PR}_{i,t} \end{gathered}$$

(5)

(under estimators, in the round brackets, are the standard errors).

The cointegration coefficients are statistically significant at a threshold of 0.03, and the error correction coefficients are close in value (−0.43 and −0.40, respectively) and significant at a threshold of less than 0.0001. This means that, in the long run, for the countries analyzed, the increase in government expenditure for social protection has the effect of decreasing the poverty rate, and the increase in the poverty rate has the effect of a (positive) scaling down of government expenditure for social protection. But in the short-run dynamics equations, the impact coefficients are not significant. The probability attached to the null hypothesis H₀, that the coefficient of the d(ExSP) variable in the short-run equation of d(PR) does not differ significantly from zero, is 0.4447, and for the coefficient of the d(PR) variable in the d(ExSP) equation, p(_H0) = 0.9855. This could mean that social policies do not react quickly to shocks arising in the poverty rate.

An econometric model of the type specified in Equation (4) was constructed, with government expenditure for social protection excluded from the specification.

The results (detailed in Appendix B.3) are summarized in the following equation:

$$\begin{gathered} {PR}_{i,t}=\underset{\left(0.0290\right)}{0.6087}\cdot \mathrm{P}{\mathrm{R}}_{\mathrm{i},\mathrm{t}-1}+\underset{\left(0.0139\right)}{0.0370}\cdot {\mathrm{U}\mathrm{R}}_{\mathrm{i},\mathrm{t}-1}+\underset{\left(0.0477\right)}{0.1219}\cdot {\mathrm{G}\mathrm{I}\mathrm{N}\mathrm{I}}_{\mathrm{i},\mathrm{t}} \\ -\underset{\left(0.0635\right)}{0.2133}\cdot {\mathrm{d}\left(\mathrm{S}\mathrm{h}\mathrm{E}\right)}_{\mathrm{i},\mathrm{t}}-\underset{\left(3.8941\right)}{40.3326}\cdot {\mathrm{d}\left(\mathrm{H}\mathrm{D}\mathrm{I}\right)}_{\mathrm{i},\mathrm{t}} \end{gathered}$$

(6)

(under estimators, in the round brackets, are the standard errors).

In Equation (6), the symbols are the same as in Equation (4). Technically, all of the coefficients are statistically significant, and the probability attached to the null hypothesis in the over-identification test (H₀: the over-identifying restrictions are valid) is p(J-stat = 24.47) = 0.435 >> 0.05. This means that the instrument variables (superior in number to the estimated variables) are valid and uncorrelated with the other variables in the model.

For the model (6), the Arellano–Bond test (Arellano and Bond 1991) does not reject the hypothesis of first-order serial correlation (at the 0.0008 level) and rejects second-order serial correlation (at the 0.6510 level) (Appendix C.2). This means that the model in the specification described by Equation (6) is well founded.

For the model described by Equation (6), standardized coefficients were also calculated. The corresponding estimators are presented in

Table 5. Standardized coefficients for the model of the economic variables’ impact on the poverty rate.

Variables	Coefficient	Standardized Coefficient
PR(−1)	0.608662	0.610444
UR	0.037016	0.040328
GINI	0.121942	0.118108
D(ShE)	−0.213330	−0.036507
D(HDI)	−40.33261	−0.042984

Note: For standardized coefficients, all of the variables were transformed to have a mean of zero and a standard deviation of one. Source: EViews-12 estimates.

.

The dynamic panel model, described by Equation (6), identified, as in model (3), an inertial effect in the dynamics of the poverty rate: the dynamic autoregression coefficient is 0.609, almost equal to the one obtained (0.601) in the estimation described by Equation (3). The autoregression coefficient is statistically significant, positive and sub-unitary, which reveals the stability of the model.

The poverty rate (PR) is positively associated with the unemployment rate (UR) and the Gini index (GINI) and negatively associated with the dynamics of the underground economy, d(She) and the dynamics of the Human Development Index, d(HDI).

The link identified between unemployment and poverty can be explained by the fact that an increase in the unemployment rate has the effect of reducing disposable income, with a direct impact on poverty (relative and absolute). As an indirect effect, higher unemployment weakens the bargaining power of workers in the wage-bargaining process, leading to lower incomes, especially in vulnerable sectors. Similar results have been obtained previously. The study by Corcoran and Hill (1980) showed that eliminating unemployment could reduce the poverty rate by 10% in the US. Other studies with similar results include Bžanová and Kováč (2024) and Oktaviani and A’yun (2021); they found a positive relationship between unemployment and poverty, but the correlation coefficient was not statistically significant (Lusiarani et al. 2023; Tudorache 2022). On the other hand, Agénor (2004) shows that expanding employment by creating low-paid jobs and increasing the number of working poor can lead to lower unemployment and higher poverty.

The Gini index measures income inequality, and an increase in the value of the index is associated with an increase in inequality in income distribution and consequently an increase in relative poverty. The results described in Equation (6) are consistent with Chen et al. (2014) and Gornick (2024).

An interesting result was obtained on the relationship between the dynamics of the underground economy, d(She) and the poverty rate (PR): the growth of the shadow economy is associated with a decrease in the poverty rate. This means that, although the shadow economy is a disruptive factor in the overall economy (with negative effects on taxation and economic growth in the long run), under certain conditions it can act as a social shock absorber (at least in the case of subsistence level incomes) (Iacobuta-Mihaita et al. 2022). Among the aspects taken into account are income coverage through self-employment (such as domestic services, subsistence farming, and casual or seasonal work) and unregulated labor (performed without an employment contract)—both as a means of generating income and as a strategy to circumvent excessive taxation and rigid labor market regulations—as well as the avoidance of economic exclusion and barriers to formal labor market access for certain groups, such as the elderly and immigrants (Ifrim et al. 2022).

Model (6), estimated for the economic explanatory factors, identifies a negative relationship between the poverty rate (PR) and the Gini index (GINI), as in model (3) specified for the institutional explanatory factors. As a dimension (assessed by standardized coefficients), inequality in income distribution (measured by the Gini index) had the strongest factor effect on the poverty rate. This result highlights the importance of economic and social policies aimed at reducing income inequality. The effects induced by the variables unemployment rate, the dynamics of the shadow economy and the dynamics of the Human Development Index on the poverty rate had approximately equal intensities (with a positive sign for UR and a negative sign for d(ShE) and d(HDI)). The probability that the respective effects (the standardized coefficients) are equal in absolute value is 0.6650 (in line with the Wald methodology for testing restrictions between coefficients).

This study therefore makes an original contribution by using a dynamic panel econometric model applied on an extensive dataset for 29 European countries, simultaneously analyzing the impact of institutional and economic variables on the poverty rate. This type of integrative approach is relatively rare in the literature, which often treats these categories of factors separately. The novelty also lies in the separate analysis of the short-run and long-run effects of government spending on social protection by combining the Arellano–Bond technique with ARDL/PMG models, which allows a more nuanced assessment of the dynamics of causal relationships.

In contrast to some research that argues for an immediate impact of social spending on poverty reduction, the present study shows that this effect is only relevant in the long-run and statistically insignificant in the short-run dynamics (Naveed and Gordon 2024; Gallego-Álvarez et al. 2021). Also, the findings on the role of the shadow economy as a cushioning factor of poverty in times of crisis provide a perspective that is less addressed in the current literature, where the focus is more on the negative effects of this phenomenon (Dessie 2014; Ngubane et al. 2023).

These results have important implications for public policy making in the EU member states. Findings on the positive influence of the quality of governance and the rule of law on poverty reduction support the need to strengthen institutions, not only as a democratic objective, but also as an effective tool for social cohesion. The finding that the shadow economy can act as a buffer in times of economic instability suggests the need for gradual formalization policies that do not exclude vulnerable people from access to minimum income.

The findings partly confirm recent studies (e.g., Lakner et al. 2022), but also provide original contributions, in particular by clearly distinguishing between short- and long-term effects and by extending the analysis to a comparative European dimension. Divergences from other papers can be explained by differences in the period analyzed and datasets and econometric methods used.

5. Conclusions

This study analyzing the evolution of the poverty rate between 2004 and 2023 for 29 European countries was structured along two main directions, addressing both institutional and economic variables. Focusing on the institutional variables, we have included in the analysis the Corruption Control Index, the Rule of Law Index and, as a control variable, the Human Development Index. The latter, being a composite indicator, provided a broad perspective on human development, including life expectancy at birth, education and the level of economic development.

Correlation analysis revealed a strong relationship between the Corruption Control Index and the Rule of Law Index, as well as a significant correlation between them and the Human Development Index, suggesting that improvements in governance are essential for raising living standards. The dynamic panel data models confirmed that improvements in the rule of law are directly correlated with lower poverty rates, emphasizing that effective governance stimulates economic growth and social equity while facilitating the implementation of effective social and economic policies. This result is supported by the literature that links effective governance with investment and increased employment opportunities, thus contributing to poverty decrease.

On the other hand, the poverty rate was negatively associated with the Corruption Control Index, reflecting the negative impact of corruption on resource allocation and access to economic opportunities. Moreover, the poverty rate was also inversely correlated with the Human Development Index, showing that improved access to quality health care, education and economic opportunities contributes significantly to improved living standards.

In terms of economic variables, the study showed that the unemployment rate, the shadow economy, government expenditure on social protection and the Gini index are all significant determinants. However, the dynamics of government expenditure on social protection did not show a statistically significant impact in the short run, suggesting that the effects of social protection policy on poverty reduction may be time lagged. However, the long-run relationships indicate a positive effect, showing that, over time, increases in government expenditure on social protection contribute to a decrease in the poverty rate.

One surprising finding was the positive impact of the shadow economy on poverty reduction, which may indicate its role as a support mechanism in the absence of formal employment opportunities. This suggests that, despite long-term negative effects on taxation and economic growth, the shadow economy can provide important temporary support for vulnerable populations.

The findings of this study highlight the complexity of the relationships between institutional and economic variables and their impact on poverty. The results suggest that an integrated approach, combining improvements in governance with well-directed economic and social policies, is essential to effectively combat poverty. This perspective provides a thorough basis for the formulation of public policies that promote inclusive and sustainable economic growth capable of benefiting all sectors of society.

Thus, based on the empirical evidence on the importance of governance, it is recommended to strengthen public institutions through measures that increase transparency, administrative efficiency and the capacity to implement social policies. Reducing corruption and strengthening the rule of law must become priorities for governments seeking to sustainably reduce poverty rates, as they create a favorable framework for attracting investment, creating jobs and ensuring the equitable delivery of public services.

On the economic front, social protection policies need to be designed for the long term, with predictable and efficient budgetary allocations, as their impact on poverty reduction is not immediate but manifests itself over time. It is also essential to tackle inequality through well-targeted redistributive measures, especially in the context where the Gini index has been identified as a major determinant of poverty.

Another innovative element of the study is the highlighting of the ambivalent role of the shadow economy. While in the long run it can negatively affect the formal economy and tax revenue collection, in the short run it can function as a shadow support mechanism for the vulnerable population. Public policies therefore need to facilitate the gradual transition of workers from the shadow to the formal economy through tax incentives, more flexible regulations and increased access to social protection.

At the academic level, the research contributes to expanding the literature on the determinants of poverty through an integrated approach and an advanced econometric methodology applied comparatively at the European level. As directions for future research, we consider it appropriate to deepen the analysis at the sub-national level to capture regional disparities, as well as to test alternative indicators of the quality of institutions, such as the efficiency of justice or trust in government.

All of these strands outline an integrated policy framework where institutional and economic reforms need to be linked to achieve a real and sustainable impact on reducing risk and poverty rates in Europe.

6. Limitations and Future Research Directions

This study provides a detailed analysis of the impact of economic and institutional factors on the risk of poverty in most European countries. However, there are some limitations to be considered. First, the data used covers the period 2004–2023. While a broad perspective on the evolution of poverty is provided, this period does not allow a full capture of the long-term effects of institutional reforms. Changes in the quality of governance or in the effectiveness of social policies may produce effects that become visible only after several years, and the present analysis cannot fully capture these dynamics. Also, the impact of major economic events, such as the 2008 financial crisis, the COVID-19 pandemic and the war in Ukraine, could have consequences on poverty that manifest themselves beyond the time horizons of the study. Also, the quality of institutional variables such as the control of corruption and the rule of law is influenced by the subjective nature of the indicators used, which are often based on perceptions rather than objective measurements.

Another limitation comes from the econometric methods applied. Although the dynamic panel models used allow a robust assessment of the relationships between variables, they do not fully capture the heterogeneity across member states. This is because the technique developed by Arellano and Bond (1991)—and used in this study—avoids obtaining biased and non-consistent estimators in the dynamic panel by eliminating individual specific effects (by a first-order differentiation). Moreover, including period dummy variables (period fixed effects) in either Equations (3) or (6) results in coefficients that are not statistically significant.

In addition, variables such as the shadow economy or social expenditure may have different effects depending on the economic structure and level of development of each country.

Future research directions could aim to enlarge the analysis by integrating additional variables, such as education and health policies, which have a high significance in decreasing poverty in the long run. Also, a more detailed approach at the level of regions, rather than just at the national level, could provide a more detailed insight into disparities across the European Union. Another possible direction would be to explore alternative ways of measuring institutional quality, based on objective indicators of government effectiveness, to reduce dependence on perception data. Thus, future research can contribute to a more detailed understanding of the mechanisms through which institutional and economic factors influence the risk of poverty and to the development of better calibrated policies to combat poverty.