Evaluating the Environmental Phillips Curve Hypothesis in the STIRPAT Framework for Finland

Abstract

In the context of increasing concerns about environmental sustainability and economic growth, this study evaluates the Environmental Phillips Curve hypothesis within Finland’s STIRPAT framework from 1990 to 2022. Finland is renowned for its commitment to environmental policies and renewable energy innovations, yet it faces challenges in balancing economic growth with environmental protection. The identified problem is the need to understand the trade-offs between economic growth and environmental impact in this specific context. Using the ARDL model, we analyze the effects of GDP per capita, renewable energy consumption (RENC), urbanization (URB), and unemployment rates (UR) on greenhouse gas emissions (GHG). Our findings show that while GDP and urbanization increase GHG emissions, renewable energy significantly reduces them. The Error Correction Model highlights quick adjustments toward equilibrium, reflecting the effectiveness of Finland’s environmental policies. Short-term results confirm the limited impact of urbanization on GHG emissions, possibly due to advanced urban planning. The FMOLS, DOLS, and CCR techniques further support these findings, emphasizing the importance of renewable energy in mitigating environmental impacts. This study provides crucial insights for policymakers seeking to balance economic growth with environmental sustainability in Finland.

1. Introduction

In the context of global climate change and the need to incorporate sustainability considerations into economic policies, the Environmental Phillips Curve hypothesis emerges as a significant area of interest for researchers. This theoretical concept posits a potential interaction between inflationary pressures and the ecological performance of an economy, offering a new perspective on the trade-off between economic growth and environmental protection. Our study aims to assess the validity of the Environmental Phillips Curve hypothesis (EPC) within the STIRPAT framework (Stochastic Impacts by Regression on Population, Affluence, and Technology), applied specifically to Finland. To achieve this, we will employ the Autoregressive Distributed Lag (ARDL) model to examine both short-term and long-term dynamic relationships between greenhouse gas emissions (GHG) and a set of economic and demographic variables.

Human activities have changed the global environment in ways never seen before. Through the release of GHG and substances that deplete the ozone layer, humans have significantly changed the atmosphere’s chemical structure. Large portions of the Earth’s surface have seen dramatic changes in land cover, significant alterations have been made to key biogeochemical cycles, and the rate of species extinction has sharply increased. The factors behind the environmental impact can be better understood by discussing the STIRPAT model and its foundation, the IPAT model [1].

Finland is known for its strong environmental policies and commitment to sustainability. Finland is a global frontrunner in bioenergy production and among the European countries making significant contributions to energy transition and climate change mitigation. Presently, Finland is striving towards the target of achieving carbon neutrality by 2035, positioning itself about a decade ahead of the timeline advocated by the European Union for 2045 [2]. Since Finland has been able to achieve economic growth alongside environmental improvements, it is an interesting case study for analyzing the EPC and STIRPAT together. The STIRPAT model could be used to identify the factors influencing pollution levels in Finland. If the EPC holds true, there might be a quantifiable relationship between Finland’s environmental policies and its unemployment rate, which could be explored by incorporating EPC principles into the STIRPAT analysis. By incorporating the unemployment rate alongside environmental indicators, one can gain insight into the complex dynamics between economic policies and environmental sustainability.

The potential contributions of this study to existing research are highlighted as follows: It represents the first investigation into the EPC hypothesis with a focus on Finland. Distinct from the majority of prior research, this study employs the STIRPAT environmental model, facilitating an objective analysis of the impact of economic and socioeconomic variables on the environment. Additionally, it explores the EPC hypothesis, introducing a novel environmental pollution curve. As one of the limited investigations verifying the EPC hypothesis, this study seeks to fill this gap in the literature. At the same time, the study proposes to check the following four research hypotheses:

Hypothesis ${\mathbf{H}}_{\mathbf{1}}$: 

Renewable energy consumption is negatively correlated with GHGs.

Hypothesis ${\mathbf{H}}_{\mathbf{2}}$: 

Per capita GDP is positively correlated with GHGs.

Hypothesis ${\mathbf{H}}_{\mathbf{3}}$: 

There is a negative relationship between unemployment and pollution. This is the EPC hypothesis.

Hypothesis ${\mathbf{H}}_{\mathbf{4}}$: 

Urbanization increases GHGs.

These hypotheses will be tested by means of the ARDL model. This study adds to the limited empirical research within the emerging concept of the EPC in the field of environmental economics by investigating the case of Finland.

In the current context of climate change and growing concerns about environmental sustainability, investigating the relationship between economic activity and environmental impact becomes crucial. The Environmental Phillips Curve hypothesis proposes an analysis of this delicate balance, suggesting that there is a trade-off between economic growth and environmental protection. The use of the STIRPAT framework provides a robust methodology for examining these complex dynamics. By integrating these perspectives, this study focuses on evaluating the validity of the EPC hypothesis in the context of Finland, a country known for its commitment to sustainability and technological innovation. Next, we will explore the relevant literature to better understand the theoretical foundations and empirical findings that have shaped this field of research.

Hacıimamoğlu [3] uses the Augmented Mean Group and Dynamic Common Correlated Effects to estimate the validity of the EPC in the Next-11 countries. The empirical analysis revealed the validity of the EPC hypothesis in the Next-11 countries, indicating that a rise in unemployment corresponds to a decrease in environmental pollution represented by the ecological footprint. Anser et al. [4] confirmed by the PMG-ARDL model the validity of the EPC hypothesis for BRICS countries. The findings by Tanveer et al. [5] confirm a long-term negative correlation between the unemployment rate and ${\mathrm{C}\mathrm{O}}_2$, ${\mathrm{C}\mathrm{H}}_4$ emissions, and ecological footprint, demonstrating the presence of the Environmental Phillips Curve in Pakistan.

The study by Oprea and Bâra [6] outlined a Big Data framework that processes data from smart meters and weather sensors into a NoSQL database for analyzing and predicting residential electricity use over 24 h. The authors introduced a machine learning approach, notably a Feed-Forward Artificial Neural Network (FF-ANN) optimized for Short-Term Load Forecasting (STLF), and compared its accuracy with six other algorithms. This methodology selected the best performing algorithm based on precision, facilitating efficient electricity consumption forecasting. The process was validated through a real-world case study in a smart residential building. The paper by Oprea and Bâra [7] presents a smart system to optimize rooftop solar panel use in homes, integrating IoT technology for real-time appliance control, significantly enhancing energy self-sufficiency, and reducing reliance on the grid.

Pattak et al. [8] investigate the impact of various energy types—nuclear, environmentally friendly, and environmentally unfriendly—alongside population and GDP on ${CO}_2$ emissions in Italy in their study. The authors employ the extended STIRPAT framework using the ARDL model. Additionally, they utilize FMOLS, DOLS, and CCR regression estimators to assess the study’s reliability. Their findings corroborate that, over the long term, increases in GDP and non-renewable energy consumption may lead to higher ${CO}_2$ emissions, whereas increases in alternative and nuclear energy could decrease these emissions.

Another study [9] examines the impact of adopting smart technologies on carbon emissions in China’s mining industry. It employs an expandable STIRPAT model and ridge regression to analyze annual data from 2001 to 2020, incorporating multiple variables such as energy intensity and industrial production value. The findings suggest that certain percentage increases in these factors significantly affect the intensity of carbon emissions. Moreover, the paper utilizes scenario-based predictions to forecast emissions intensity from 2021 to 2035, identifying scenarios that could achieve emissions reduction targets. This provides a developmental strategy for minimizing the ecological impact of China’s mining sector.

In another study [10], an enhanced STIRPAT model, the environmental Kuznets curve, and machine learning techniques such as ridge and lasso regression were used to analyze the impact of institutional quality on carbon emissions across 22 European Union countries from 2002 to 2020. The analysis differentiates between countries with strong and weak institutions to compare effects. The results show that in countries with robust institutions, improvements in government effectiveness lead to increased emissions, whereas greater voice and accountability result in reductions. Conversely, in nations with weaker institutional frameworks, political stability and effective corruption control are associated with decreased emissions. Key findings also reveal that factors like population density, urbanization, and energy consumption significantly influence carbon emissions more than institutional governance, emphasizing the need for comprehensive and consistent climate-aligned policies across the EU.

In another study that analyzes the extended STIRPAT model [11], the impact of demographic structure on carbon emissions is explored. The analysis also includes the influence of climatic factors such as precipitation, degree days, and temperature anomalies, using NASA’s night-time light data as a proxy for population density. The Seemingly Unrelated Regression (SUR) method is applied to manage cross-sectional correlation and endogeneity within an interconnected global context. The results indicate a significant impact of age structure on carbon emissions, with a general inverted U-shaped pattern in carbon consumption across age groups. The study confirms that an aging population tends to increase carbon emissions from heating and electricity, suggesting a need to enhance energy efficiency, particularly through upgrading building insulation and implementing green building standards.

In other research, socio-economic and environmental variables have been analyzed for Finland. For instance, one study [12] examines the effects of productivity, energy consumption, foreign direct investments, and urbanization on carbon dioxide emissions in Finland from 2000 to 2020, utilizing an Autoregressive Distributed Lag (ARDL) model. The results indicate that: there is cointegration among the variables, energy consumption significantly contributes to an increase in ${\mathrm{C}\mathrm{O}}_2$ emissions over the long term, labor productivity and urbanization are associated with reductions in ${\mathrm{C}\mathrm{O}}_2$ emissions over the long term, and foreign direct investments have an insignificant effect on ${\mathrm{C}\mathrm{O}}_2$ emissions. The findings are discussed with considerations for policy implications and directions for future research. Another study [13] focuses on analyzing the impact of renewable electricity production, GDP per capita, and urbanization on forest growth in Finland between 1990 and 2022. Using the ARDL model, the research highlights a long-term relationship between these factors and forest area growth, which is supported by renewable electricity and urbanization but hindered by economic growth. In summary, the study’s results suggest that forest conservation strategies should consider these economic and development dynamics.

The literature review emphasizes the importance of an interdisciplinary approach in evaluating the Environmental Phillips Curve hypothesis within the STIRPAT model for Finland. The studies examined explore various aspects, from the impact of institutional quality and demographic structure on carbon emissions to the influence of energy consumption and urbanization on the environment. These analyses demonstrate the complexity of the relationships between economic variables and their impact on the environment, highlighting the need for a rigorous and adaptive approach to understand the dynamics between economic growth and environmental sustainability in Finland. In the current context, it is essential to integrate these findings into policy formulation and sustainable development strategies, considering both the short-term and long-term effects of economic decisions on the environment.

Shastri et al. [14] investigate the applicability of the EPC by analyzing gender-segregated data to determine potential disparities for India during 1990–2019. The results showed the absence of the EPC for female unemployment and its existence for male unemployment. This disparity and the absence of a trade-off between female employment and environmental quality stem from the occupational distribution and segregation between men and women in India. Indian women have predominantly engaged in labor-intensive and informal work, primarily concentrated in low-productivity sectors. More recently, Indian women are often employed in professions such as teaching and nursing, which are considered to be cleaner sectors of the economy. As a result, Indian women’s employment does not directly contribute to environmental degradation. Tariq et al. [15] validated the EPC hypothesis for South Asian nations, employing both panel data and country-specific estimations. Durani et al. [16] found that employment levels surged CO₂ emissions, therefore invalidating the EPC for BRICST countries for the period 1990–2020.

Our study is organized into several sections. Section 2 is dedicated to identifying and describing the research gap, an essential section for any study as it helps clarify the reason for the necessity of the research. Section 3 details the methodological flow, which includes describing the chosen data set, the STIRPAT framework, and the Environmental Phillips Curve, as well as the steps involved in applying the ARDL model to our data set. Section 4 is dedicated to the results of the research, where we will present the analysis derived from the ARDL Model. Our study concludes with Section 5, which is devoted to the conclusions and discussions of the research, detailing the limitations of the study and future research directions.

2. Research Gap

Despite the growing body of literature on the relationship between economic activity and environmental impact, there are notable gaps that this study aims to address. First, while the EPC hypothesis has been explored in various contexts, there is a lack of comprehensive analysis integrating the EPC within the STIRPAT framework. This combination can provide a more nuanced understanding of the trade-offs between economic growth and environmental sustainability.

Secondly, existing studies often focus on broader geographic regions or specific countries with distinct economic structures, leaving a gap in understanding how the EPC hypothesis applies to Finland, a country known for its advanced environmental policies and technological innovations. The unique economic and environmental context of Finland provides an ideal case for examining the validity of the EPC hypothesis using the STIRPAT model.

Furthermore, while the STIRPAT framework has been widely used to analyze environmental impacts, its application in conjunction with the EPC hypothesis remains underexplored. This study aims to fill this gap by applying the STIRPAT framework to test the EPC hypothesis, thus offering a robust methodological approach to investigate the complex dynamics between economic growth, population, affluence, and technology.

By addressing these research gaps, our study not only contributes to the theoretical advancement of the EPC hypothesis and the STIRPAT framework but also provides valuable insights for policymakers seeking to balance economic development with environmental sustainability in Finland and beyond.

To identify the research gaps, we will review the current literature on Nordic countries, environmental-economic interactions in Finland, and comparable economies. We will consider comparative studies and assess whether comparative analyses between Finland and other countries with similar economic structures have been sufficiently explored. This could highlight unique features or common challenges in implementing environmental policies. The relationship between unemployment and GHG emissions could be explored further, especially under different economic conditions or in light of recent economic shifts due to global events like the COVID-19 pandemic. The selection of Finland is motivated by two distinct factors, according to Esposito [2]. Firstly, in 2021, Finland surpassed a 40% share in renewable energy consumption. Secondly, Finland stands out for having one of the highest energy consumptions globally, primarily due to its extreme climatic conditions.

Finland provides a compelling context for applying the STIRPAT model due to its unique environmental, economic, and demographic characteristics. Finland is renowned for its commitment to sustainability and environmental protection, making it an ideal case study for examining the interplay between human activities and environmental impacts. Finland’s high level of technological innovation, especially in renewable energy and waste management, adds a layer of depth to such analyses. Moreover, Finland’s policies and regulations aimed at reducing GHG and promoting energy efficiency provide additional variables that can be incorporated into the model to understand their effectiveness. The STIRPAT model can particularly benefit from Finland’s detailed environmental and economic data, allowing for precise adjustments and accurate predictions. The outcomes of such studies are essential for policymakers, helping to craft strategies that balance economic growth with environmental sustainability, which is a key goal on Finland’s national agenda. This makes the STIRPAT model especially relevant and potentially revealing in the Finnish context, offering insights into the mechanisms through which small-scale changes in population, affluence, or technology might translate into significant environmental outcomes.

Norway and Sweden, thanks to the utilization of hydropower, along with Sweden and Finland, which have an effective deployment of biomass in heat and power plants, have reached high levels of renewable energy usage [17]. Denmark boasts the highest proportion of wind energy globally, and Iceland is rich in geothermal energy resources. This study focuses on Finland as a Nordic country for several compelling reasons. Finland has set an ambitious goal to drastically reduce carbon emissions and enhance energy efficiency by 2050 [18], aligning with its climate objectives. Known for its high innovation ranking, Finland is well-positioned to develop innovative solutions and provide policy insights. Achieving its climate goals will require a comprehensive green transition and advancements in green technologies across all sectors of Finnish society and economy. The Nordic Council of Ministers (NCM) is a cooperative body that facilitates collaboration and cooperation between the Nordic countries: Denmark, Finland, Iceland, Norway, and Sweden, as well as their autonomous territories, including the Faroe Islands, Greenland, and the Åland Islands. The NCM was established in 1971 as a forum for Nordic governmental cooperation and is headquartered in Copenhagen, Denmark. The NCM plays a significant role in promoting environmental quality and sustainability in the Nordic region through various initiatives and collaborations. The NCM facilitates cross-border cooperation among the Nordic countries to tackle shared environmental challenges, such as pollution prevention, waste management, and ecosystem conservation. The Nordic countries have faced challenges in significantly advancing Sustainable Development Goals (SDGs) 7 and 13 [19]. SDG 7 is “Affordable and Clean Energy”, which aims to ensure access to affordable, reliable, sustainable, and modern energy for all. SDG 13 is “Climate Action”, which focuses on taking urgent action to combat climate change and its impacts. This goal includes targets related to reducing GHG, increasing resilience and adaptive capacity to climate-related hazards, and integrating climate change measures into national policies. This situation may stem from underlying issues related to financialization concerns and obstacles in implementing renewable energy generation [20].

Magazzino et al. [21] examined data from the Scandinavian countries spanning the period 1990–2018, finding a negative correlation between the use of green energy and CO₂ emissions. The study by Sharif et al. [22] provides a thorough examination of the contributions of green technology, environmental taxes, and the adoption of green energy to foster environmental sustainability in the Nordic countries for the period 1995–2020 by CS-ARDL in the framework of the STIRPAT model. The results showed that green technology, environmental taxes, and green energy are negatively correlated with ${\mathrm{C}\mathrm{O}}_2$ emissions, while income and population are positively correlated with ${\mathrm{C}\mathrm{O}}_2$ emissions. Alola and Onifade [23] studied the influence of various types of energy sources to ${\mathrm{C}\mathrm{O}}_2$ emissions in Finland during 1974–2019. The ARDL technique suggested that coal, natural gas, nuclear, and oil energy sources have been identified as harmful to the environment in Finland. Alola et al. [24] assess the influence of GDP, coal, oil, and environmental innovation on carbon emissions for Nordic countries in the STIRPAT framework during 2000–2019 using CS-ARDL. They observed that changes in the energy mix resulted in an increase in carbon emissions both in the short and long term. The introduction of innovative environmental practices significantly reduced carbon emissions, particularly over the long term. When the interaction between coal and innovation was introduced, although coal usage continued to contribute to an increase in carbon emissions, the magnitude of this impact was less pronounced in both the short and long term. Alola and Adebayo [25] use panel symmetric and asymmetric ARDL for the study of GHG determinants for a set of Nordic countries during 2000–2019. Both in the short and long term, the findings indicate that GDP continues to contribute to an increase in GHG emissions, while raw material productivity mitigates GHG emissions. This underscores the efficient utilization of raw materials in the Nordic region. Environmental-related technologies and export intensity also contribute to GHG emission reduction in the Nordic countries, albeit primarily in the long term.

The study by Wang et al. [26] investigates the influence of financial risk, political risk, and renewable energy on CO₂ emissions in Finland from 1990 to 2020 by means of wavelet-based quantile correlation. The main results are that renewable energy and improvements in oil and gas efficiency consistently lead to reductions in CO₂ emissions across all quantiles.

Several studies have studied the existence of the Environmental Kuznets Curve (EKC) for Finland. Most studies invalidated it [12,27,28,29]. The invalidation of the EKC hypothesis in most cases for Arctic countries, including Finland, could be attributed to several factors. Arctic regions experience distinct environmental dynamics compared to other parts of the world. Factors such as extreme cold temperatures, permafrost thawing, ice melting, and unique ecosystems create complex and unique environmental challenges that may not conform to the assumptions of the EKC. Arctic countries often rely heavily on resource extraction industries such as mining, oil, and gas extraction. These industries can lead to significant environmental degradation, including habitat destruction, pollution, and disruption of fragile ecosystems, which may not follow the inverted U-shaped curve predicted by the EKC [29]. Arctic regions are particularly vulnerable to the impacts of climate change, including rising temperatures, melting ice caps, and sea-level rise. These environmental changes can have profound and immediate consequences for the region’s ecosystems, wildlife, and indigenous communities, making it challenging to observe the turning point where environmental degradation decreases with economic development. Arctic countries have economies that are heavily reliant on a few key industries, such as resource extraction, fishing, and tourism. Limited economic diversification may limit the potential for economic growth to lead to environmental improvements, as predicted by the EKC.

3. Methodology, Dataset, and Model

3.1. Dataset and Model

The STIRPAT model introduced by Dietz and Rosa [30] and the IPAT identity introduced by Ehrlich and Holdren [31] are frameworks used in environmental studies to understand and analyze the impact of human activities on the environment. Both offer insights into how various factors contribute to environmental change, but they differ in their flexibility and specificity. The IPAT identity is a conceptual model that provides a simple way to understand the factors that drive environmental impact. It can be represented as:

$$I=P\times A\times T$$

(1)

In Equation (1), $I$ stands for environmental impact, P stands for population, A stands for affluence (often measured as GDP per capita), and T stands for technology (considered as the impact per unit of economic activity). The IPAT identity suggests that the total environmental impact (I) is the product of population size, affluence, and technology. It is a straightforward framework that highlights the importance of these three factors but is often criticized for its simplicity and the implied proportionality between each factor and impact, which does not account for complex interactions or non-linear effects. The STIRPAT model is an extension of the IPAT identity that allows for more flexibility and detailed analysis. It can be represented as:

$$I={\alpha }_0{P}^{{\alpha }_1}{A}^{{\alpha }_2}{T}^{{\alpha }_3}\epsilon$$

(2)

In Equation (2), ${\alpha }_0$ is a constant, P, A, and T are defined in the IPAT model, ${\alpha }_1$, ${\alpha }_2$ and ${\alpha }_3$ are coefficients that determine the elasticity of the environmental impact with respect to population, affluence, and technology, respectively. $\epsilon$ stands for the error term. The STIRPAT model does not assume a fixed proportional relationship between impact and its drivers; instead, it allows for empirical determination of how changes in population, affluence, and technology individually and collectively affect environmental impacts. This flexibility makes STIRPAT a powerful tool for empirical analysis, enabling researchers to account for non-linearities and interactions between factors. While the IPAT identity offers a foundational understanding of the factors contributing to environmental impacts, the STIRPAT model provides a more nuanced and flexible approach for analyzing these impacts in detail.

A regression model that uses logarithms for all variables simplifies the processes of estimation and testing hypotheses. In common uses of the standard STIRPAT model, the variable T is incorporated into the error term instead of being estimated on its own. This approach aligns with the IPAT model, in which T is calculated to achieve a balance among I, P, and A [1,8]. Thus, from (2), we obtain Equation (3):

$$ln{I}_t={\alpha }_0+{\alpha }_1\times ln{P}_t+{\alpha }_2\times ln{A}_t+{\alpha }_3\times ln{T}_t+{\epsilon }_t$$

(3)

In the STIRPAT model (Equation (3)), T represents a range of factors (essentially, anything influencing impact per production unit). We will remove it from (3), obtaining Equation (4):

$$ln{I}_t={\alpha }_0+{\alpha }_1\times ln P_t+{\alpha }_2\times ln{A}_t$$

(4)

According to York et al. [1], IPAT’s primary advantages include its parsimonious specification of the main factors driving environmental change and its accurate delineation of how these factors are interrelated with the impacts. It emphasizes that P, A, and T are interdependent, meaning changes in any one factor are compounded by the others.

Table 1. Variable specification.

Variable	Acronym	Log Form	Measurement Unit	Source
Greenhouse gas esmissions	GHG	$lnGHG$	kiloton of ${\mathrm{C}\mathrm{O}}_2$ equivalent	Macrotrends
Renewable energy consumption	RENC	$lnRENC$	(%) of total final energy consumption	World Bank
Gross domestic product	GDP	$lnGDP$	Constant 2015 USD	World Bank
Unemployment rate	UR	$lnUR$	(%)	Macrotrends
Urbanization	URB	$lnURB$	(%)	World Bank

describes the variables used in this study for the period 1990–2022. This study investigates the impact of GDP, RENC, UR, and URB on GHG in Finland during 1990–2022. At the same time, it attempts to study if the EPC hypothesis is fulfilled in the case of Finland.

In our study, the STIRPAT framework can be reframed from the following perspective: The GHG variable can be considered in the STIRPAT framework as an outcome of the interaction between population, wealth level, and technology. Essentially, we can investigate how variations in GDP, UR, and URB influence GHG in Finland. For example, an increase in GDP and urbanization could lead to higher GHG emissions due to increased energy demand and resource utilization, while an increase in the unemployment rate could reduce consumption and, consequently, emissions. The RENC variable can be linked to the level of technology and how it is influenced by socio-economic factors within STIRPAT. For instance, we can explore how GDP, UR, and URB might stimulate investments in renewable energy technologies, while lower unemployment rates could influence the financial capacity of individuals and governments to invest in green energy. The GDP variable can be considered a measure of wealth level and can influence both energy consumption and GHG emissions. In the STIRPAT framework, we can investigate how GDP growth affects total energy consumption and how this may influence GHG emissions. For instance, strong economic growth could lead to increased energy consumption, particularly of fossil fuels, exacerbating GHG emissions. The UR variable, as well as the URB variable, can also influence energy consumption and GHG emissions. For example, a higher unemployment rate may affect resource and technology utilization, while urbanization may affect how people access and use energy. Within STIRPAT, we can explore how these variables influence energy consumption and, indirectly, GHG emissions.

Thus, Equation (3) can be rewritten as follows:

$$lnGHG={\alpha }_0+{\alpha }_1\times ln{RENC}_t+{\alpha }_2\times ln{GDP}_t+{\alpha }_3\times ln{UR}_t+ln{UR}_t+{\epsilon }_t$$

(5)

There are several studies that take such an approach, using the STIRPAT framework to analyze environmental aspects by employing socio-economic and technological factors [8,32,33,34].

As employment levels rise, it causes a boost in income, which subsequently leads to heightened pollution in an economy. The environmental consequences of this increased income manifest in several ways. The proportion of income spent on essential items like food decreases, whereas spending on non-essential goods sees an increase. There is a surge in the consumption of products that are high in inorganic materials and carbon, leading to a broader range of goods being consumed, which in turn amplifies pollution levels. There is a prevailing concern that improving an economy’s environmental footprint might inadvertently raise unemployment by reducing national income. These issues led Kashem and Rahman [35] to introduce the Environmental Phillips Curve (EPC). The EPC hypothesis posits that, given the present state of technology, there exists an inverse relationship between environmental pollution and unemployment rates [35,36]. Economic activities intensify environmental stress and contribute to pollution. It is posited that a positive correlation exists between economic expansion and environmental degradation [35]:

$$P=a+bY$$

(6)

In Equation (6), P denotes environmental pollution, and Y represents economic growth or income.

Okun [37] examined the dynamics between unemployment rates and economic expansion in the USA from 1948 to 1960, finding a negative correlation between the two. This inverse relationship, where unemployment falls as income rises, came to be known as Okun’s Law, highlighting the linkage between unemployment (U) and economic growth or income (Equation (7)):

$$U=c-dY$$

(7)

The inverse correlation between environmental pollution (P) and unemployment (U) is shown in Equation (8):

$$P=g-hU$$

(8)

Equation (8) is known as the EPC.

3.2. Autoregressive Distributed-Lag (ARDL) Model

The dependence equation of the model is described in Equation (9):

$${GHG}_t=a_0+a_1{GDP}_t+a_2{RENC}_t+a_3{URB}_t+a_4{UR}_t+{\epsilon }_t$$

(9)

The time series data have been converted into natural logarithms, a process that smooths out sudden changes and stabilizes the variance throughout the series [38].

Equation (9) is written as an ARDL (n, p, q, r, s) model as follows:

$$\begin{gathered} \mathsf{\Delta }{GHG}_t=a_0+\sum _{k=1}^n{a}_1\mathsf{\Delta }{GHG}_{t-k}+\sum _{k=1}^p{a}_2\mathsf{\Delta }{GDP}_{t-k}+\sum _{k=1}^q{a}_3\mathsf{\Delta }{RENC}_{t-k}+\sum _{k=1}^r{a}_4\mathsf{\Delta }{URB}_{t-k}+\sum _{k=1}^s{a}_5\mathsf{\Delta }{UR}_{t-k} \\ +{\lambda }_1{GHG}_{t-1}+{\lambda }_2{GDP}_{t-1}+{\lambda }_3{URB}_{t-1}+{\lambda }_4{UR}_{t-1}+{\epsilon }_t \end{gathered}$$

(10)

In Equation (10), $\mathsf{\Delta }$ is the first difference operator, and n, p, q, r, and s are the lag lengths of the ARDL model.

The examination of the cointegration relationship among GHG and its determinants is checked by means of the joint cointegration test developed by Bayer and Hanck in 2013 [39]. This test delivers reliable results by amalgamating four distinct cointegration techniques: those introduced by Engle and Granger [40], Johansen [41], Boswijk [42], and Banerjee et al. [43], which are referred to as EG, JOH, BO, and BDM, respectively. It employs Fisher F-statistics to provide evidence of cointegration. Following the Fisher formula, the test’s formulations are presented in (11) and (12):

$$EG-JOH=-[\ln \left(PEG\right)+\ln \left(PJOH\right)]$$

(11)

$$EG-JOH-BO-BDM=-2[\ln \left(PEG\right)+\ln \left(PJOH\right)+\ln \left(PBO\right)+\ln \left(PBDM\right)]$$

(12)

PEG, PJOH, PBO, and PBDM denote the test probabilities for EG, JOH, BO, and BDM, respectively. If the computed Fisher statistic exceeds the critical value set forth by Bayer and Hanck [39], the null hypothesis, which suggests the absence of cointegration, can be dismissed.

Moreover, the findings of this study are further validated using the ARDL cointegration bounds testing approach introduced by Pesaran et al. [44]. This method’s null hypothesis asserts the absence of cointegration, whereas the alternative hypothesis indicates cointegration’s existence. The test employs F-statistics for evaluation, comparing them against critical values specified by Pesaran et al. [44]. Should the F-statistic exceed the critical upper bound, labeled as I (1), the null hypothesis is refuted, confirming cointegration. In the case of cointegration, the Error Correction Model (ECM) can be formulated as follows:

$$\begin{gathered} {∆GHG}_t=a_0+{\sum }_{k=1}^n{a}_1{∆GHG}_{t-k}+{\sum }_{k=1}^p{a}_2{∆GDP}_{t-k}+{\sum }_{k=1}^q{a}_3{∆RENC}_{t-k}+{\sum }_{k=1}^r{a}_4{∆URB}_{t-k}+ \\ {\sum }_{k=1}^s{a}_5{∆UR}_{t-k}+{\mathsf{\Gamma }ECM}_{t-1}+{\epsilon }_t \end{gathered}$$

(13)

The coefficient of the ECM in Equation (13) defines the short-term dynamics. The Error Correction Term (ECT) should be statistically significant and bear a negative value, not exceeding −2, as highlighted by Samargandi et al. [45]. A negative sign is indicative of the rate at which adjustments occur on both short-term and long-term scales.

FMOLS (Fully Modified Ordinary Least Squares), DOLS (Dynamic Ordinary Least Squares), and CCR (Canonical Cointegrating Regression) are three advanced econometric techniques used to estimate the long-run relationships among integrated variables—variables that become stationary when differenced. These methods are particularly useful in the context of cointegration analysis, where the goal is to explore the equilibrium relationship between two or more non-stationary time series in the long term. Each of these techniques has unique features and is designed to address specific issues in cointegrated time series analysis, such as serial correlation and endogeneity.

FMOLS, developed by Phillips and Hansen [46], is designed to correct for the problems of serial correlation and endogeneity in the error term that are common in cointegrated systems. FMOLS adjusts the OLS estimator to account for these issues, providing more reliable and consistent estimates of the long-run parameters. This method is particularly useful when dealing with small sample sizes, as it adjusts the OLS estimator to become super-consistent.

DOLS, proposed by Saikkonen [47] and Stock and Watson [48], addresses the issue of serial correlation and endogeneity by including leads and lags of the first differences of the independent variables in the regression model. By doing so, DOLS filters out the noise caused by these problems, leading to consistent and efficient estimates of the long-term coefficients. DOLS is advantageous because it is relatively simple to implement and tends to have good finite-sample properties.

CCR, developed by Park [49], is another approach to estimating the long-run parameters in cointegrated systems. CCR uses canonical correlation analysis to transform the data before estimation, aiming to remove the endogeneity present in the regressors and correct for serial correlation. This method provides consistent estimates under a wide range of conditions, including cases with highly persistent time series. CCR is particularly noted for its robustness in dealing with a variety of issues that can arise in cointegrated regression analysis.

The choice among FMOLS, DOLS, and CCR depends on the specific characteristics of the data and the issues present in the econometric model. Each method has its strengths and weaknesses. FMOLS is preferred for its simplicity and effectiveness in small samples but might be less efficient if there are large numbers of integrated regressors or if the system is subject to significant structural changes. DOLS is often chosen for its efficiency and simplicity, especially in cases where the lead-lag structure can be easily determined. It is particularly useful when dealing with higher levels of serial correlation. Another benefit of the DOLS method is its ability to incorporate factors of different integration orders within a unified, cointegrated framework [50]. CCR is robust in a wide range of scenarios, especially when dealing with endogeneity and persistent data, but it can be more complex to implement compared to FMOLS and DOLS.

Finally, a variety of diagnostic tests were applied, such as the normality test, the Breusch–Pagan–Godfrey test, the Breusch–Godfrey serial correlation test, the LM test, and the Ramsey reset test.

The model’s stability was further examined using the cumulative sum (CUSUM). The CUSUM test involves the plotting of the ECM’s residuals. If the plots of CUSUM stay within the 5% critical boundary, the null hypothesis, which suggests that the model’s parameters are stable, cannot be dismissed.

The STIRPAT model provides a conceptual and theoretical methodology for evaluating the impact of demographic, economic, and technological factors on the environment, offering a static understanding of the relationships between the variables used in our study. However, the relationships between economic variables and environmental impact evolve over time and can be dynamic. To complement the STIRPAT framework, the ARDL model captures the short-term and long-term dynamics between our variables, providing a holistic perspective of the temporal relationships. By combining the STIRPAT model with the ARDL model, our study benefits from a comprehensive approach that not only identifies the determinants of greenhouse gas emissions but also examines how these relationships change over time. This integration allows us to formulate better-founded policy recommendations and gain a deeper understanding of the impact of economic and technological interventions on environmental sustainability in Finland.

4. Results

Figure 1 and Figure 2 illustrate the annual development of five indicators for Finland between 1990 and 2022, showcasing rising trends in GDP, RENC, and URB, while observing significant reductions in GHG in recent years. Additionally, Figure A1, Figure A2, Figure A3, Figure A4 and Figure A5 in Appendix A present the histograms and statistical summaries for the analyzed variables.

In Figure 2, we observe the evolution of the unemployment rate. The unemployment rate in Finland displayed a notable evolution between 1990 and 2000. Initially, there was a substantial increase from 1990 to 1995, reaching a peak in 1995. This could be attributed to various economic factors such as recession, changes in industrial structure, or shifts in government policies affecting employment. Following the peak in 1995, there was a decline in the unemployment rate until around 2000. This decline might be indicative of economic recovery, the implementation of new employment strategies, or other favorable conditions that led to more job opportunities and reduced unemployment. However, in the last two years analyzed, there were fluctuations in the unemployment rate, with periods of both increases and decreases. These fluctuations suggest that the labor market might have experienced some instability or volatility during those years. Factors such as changes in global economic conditions, shifts in industry demands, or specific domestic policies could have contributed to these fluctuations.

Overall, the trend indicates a general stabilization in the unemployment rate towards the later years, albeit with periodic ups and downs, reflecting the dynamic nature of the labor market and the complexities involved in managing unemployment rates over time.

Table 2. Summary statistics.

	GHG	GDP	RENC	URB	UR
Mean	11.11041	10.57361	3.507588	4.419932	2.236184
Median	11.15115	10.65798	3.455686	4.419286	2.156403
Maximum	11.36117	10.75978	3.860519	4.450631	2.833213
Std. Dev.	10.77354	10.234	3.179303	4.374083	1.791759
Skewness	0.153369	0.172764	0.210702	0.023339	0.298504
Kurtosis	−0.541264	−0.81029	0.229508	−0.248276	0.610719
Jarque-Bera	2.295358	2.152304	1.803304	1.892912	2.340489
Probability	2.294033	4.599193	2.258818	2.024285	2.649443

offers descriptive statistics for the variables following logarithmic transformation. It shows URB with a mean of 4.41 and a maximum value of 445. GDP’s mean stands at 10.57, with a relatively low variability of 0.17. URB also presents minimal variability, marked at 0.02, alongside a mean value of 3.98. GHG, RENC, URB, and UR are characterized by platykurtic distributions, while GDP exhibits a leptokurtic distribution.

The application of the Augmented Dickey–Fuller (ADF) unit root test [51] leads to the conclusion that all variables are integrated of order 1 (see

Table 3. ADF Unit Root Test Results.

Variables	Level	First Difference	Order of Integration
	T-Statistics	T-Statistics
GHG	0.61 (0.987)	$-$6.00 *** (0.000)	I (1)
GDP	$-$0.77 (0.812)	$-$4.11 *** (0.003)	I (1)
RENC	$-$0.57 (0.862)	$-$7.96 *** (0.000)	I (1)
URB	$-$1.21 (0.654)	$-$1.88 * (0.057)	I (1)
UR	$-$0.34 (0.553)	$-$3.47 *** (0.001)	I (1)

* and *** indicate the significance of variables at 10% and 1% levels, respectively.

). Therefore, the ARDL model can be applied.

Based on

Table 4. VAR Lag order selection criteria.

Lag	LogL	LR	FPE	AIC	SC	HQ
0	252.83	NA	7.83 × 10−14	−15.98	−15.75	−15.91
1	422.21	273.18	7.23 × 10−18	−25.30	−23.91	−24.85
2	470.05	61.73 *	1.90 × 10−18 *	−26.77 *	−24.23 *	−25.94 *

* indicates the lag order selected by the criterion. LR represents the sequential modified LR test statistic (each test at the 5% level), FPE represents the final prediction error, AIC represents the Akaike information criterion, SC represents the Schwarz information criterion, and HQ represents the Hannan-Quinn information criterion.

, all criteria indicate that choosing a lag length of 2 is the optimal decision for the Vector Autoregression (VAR) model.

The obtained model is ARDL (2,1,1,1,0). Therefore, n = 2, p = 1, q = 1, r = 1, and s = 0.

Table 5. Bayer–Hanck cointegration test.

Tests	Engel–Granger (EG)	Johansen (JOH)	Banerjee (BDM)	Boswijk (BO)
Test statistic	$-$3.80	44.94	$-$1.65	32.43
p-value	0.18	0.00	0.80	0.00
$EG-JOH$	16.25	5% critical value, 10.57
$EG-JOH-BO-BDM$	30.93	5% critical value, 20.14

shows that the F-statistic values calculated using both the $EG-JOH$ and $EG-JOH-BO-BDM$ approaches surpass the critical thresholds at the 5% significance level. This result substantiates the rejection of the null hypothesis that suggests no cointegration at the 5% significance level. The conclusion drawn is that there is cointegration among the variables.

Table 6. Results of ARDL cointegration bounds test.

Test Statistic	Value	K (Number of Regressors)
F-statistic	12.29	4
Critical value bounds
Significance	I (0)	I (1)
10%	2.52	3.56
5%	3.05	4.22
1%	4.28	5.84

shows that the computed F-statistic is 12.29, exceeding the upper critical bound for I (1), indicating the presence of cointegration among the variables. The estimated long-term coefficients can be found in

Table 7. Long-run estimated results.

Variable	Coefficient	T-Statistics	Prob.
GDP	0.63	3.10	0.00 ***
RENC	$-$1.64	$-$8.03	0.00 ***
URB	6.47	2.65	0.01 **
UR	0.13	1.80	0.08 *
C	$-$18.79	$-$2.24	0.03 **

*, **, and *** indicate the significance of variables at 10%, 5%, and 1% levels, respectively.

.

From Table 7, it follows that GDP has a positive and statistically significant long-term influence on GHG, confirming ${\mathrm{H}}_2$. A 1% increase in GDP exerts a 0.63% increase in GHG in Finland. It suggests a relatively high elasticity of emissions with respect to economic growth, though it is less than a one-to-one relationship. Finland’s energy mix has traditionally been diverse, with a significant reliance on biomass, nuclear power, and imported fossil fuels. The country has been making efforts to shift towards more renewable energy sources. The extent to which GHG emissions increase with GDP depends on how much additional economic activity relies on fossil fuels versus cleaner energy sources. Finnish industries and the economy, in general, have been focusing on improving energy efficiency and reducing carbon intensity. As GDP grows, if the increase in economic activities is more oriented towards services or high-tech industries that are less energy-intensive, the GHG emissions would not rise proportionately with GDP. Finland has been a pioneer in environmental legislation and carbon pricing, implementing carbon taxes since the early 1990s. These policies incentivize businesses to adopt cleaner technologies and reduce emissions. The effectiveness of these policies can mitigate the growth of GHG emissions as the economy expands. The Finnish economy has been undergoing structural changes, with a shift away from heavy industry and manufacturing towards services and technology. Since the service sector generally has a lower carbon footprint compared to the industrial and manufacturing sectors, GDP can grow with a relatively smaller increase in GHG emissions.

The specific increase rate of 0.63% in GHG emissions for a 1% increase in GDP could be influenced by seasonal or yearly variations, such as changes in weather conditions affecting heating needs, which in turn impact fossil fuel use and GHG emissions. This result is consistent with the findings by Kirikkaleli and Adebayo [52] and Alola and Adebayo [25] for the Nordic countries. Hypothesis ${\mathrm{H}}_2$ was also confirmed by Alola and Adebayo [25] for Finland over the period 1990–2020 and by Maâlej and Cabagnols [53] for Finland and Italy over the period 2001–2016.

A 1% increase in RENC exerts a long-term 1.64% decrease in GHG in Finland, confirming $H_1$. Finland’s specific context, including its climate goals, energy mix, and policy framework, makes the impact of renewable energy on GHG emissions particularly pronounced. For example, Finland has a strong commitment to climate neutrality and has implemented policies to increase the share of renewables in its energy mix. The country’s energy infrastructure, technological capabilities, and policy measures are designed to efficiently integrate renewable energy sources and reduce GHG emissions across different sectors of the economy. By directly substituting energy produced from fossil fuels with energy generated from renewable sources, the immediate effect is a reduction in GHG emissions. This is because renewable energy sources, such as wind or solar power, do not emit carbon dioxide or other GHGs as a byproduct of energy production. Increasing the share of renewable energy within the energy mix can stimulate broader systemic changes. This includes advancements in energy efficiency, changes in energy storage, and the electrification of sectors such as transportation and heating that traditionally rely on fossil fuels. These transformations further reduce GHG emissions. The increased demand for renewable energy accelerates technological innovation. This can lead to more efficient renewable energy technologies and reductions in the cost of renewable energy production, making it increasingly competitive with fossil fuels. Over time, this can lead to a more substantial displacement of fossil fuels from the energy mix, further decreasing GHG emissions. A growing commitment to renewable energy can influence policy and economic signals, leading to the implementation of carbon pricing, subsidies for renewable energy projects, and stricter emissions standards for industries. These policies can significantly reduce GHG emissions by making renewable energy more attractive and penalizing carbon-intensive energy production. Hypothesis ${\mathrm{H}}_1$ was also validated for Finland in the study by Wang et al. [26] using wavelet-based quantile correlation.

A 1% increase in URB exerts a long-term 6.47% increase in GHG in Finland, confirming hypothesis ${\mathrm{H}}_4$. In Finland, the impact of urbanization on GHG emissions can be particularly pronounced due to its high standard of living, cold climate necessitating significant heating, and the structure of its urban areas. While Finland is known for its commitment to sustainability and has made considerable progress in incorporating renewable energy sources, the dynamics of urbanization still pose challenges for managing GHG emissions. Urban areas typically have higher energy demands due to dense populations and a concentration of commercial, industrial, and residential buildings. Heating, cooling, lighting, and powering devices in these buildings can lead to increased consumption of electricity and, depending on the energy mix, potentially higher GHG emissions. Urbanization often leads to increased demand for transportation, including personal vehicles, public transit, and freight to support businesses and residents. If the transportation infrastructure relies heavily on fossil fuels, this can significantly increase GHG emissions. Urbanization drives construction, which requires energy-intensive materials like steel and concrete, contributing to GHG emissions. Additionally, older buildings may be less energy-efficient, leading to higher energy consumption for heating and cooling. Urban areas can become significantly warmer than their rural counterparts due to human activities and the prevalence of surfaces that absorb and retain heat. This effect can increase energy use for cooling during warmer months, thereby raising GHG emissions. Urban lifestyles may lead to higher consumption of goods and services, which, in turn, increases the demand for energy in manufacturing, transportation, and waste management processes associated with these goods and services, leading to higher GHG emissions. Hypothesis ${\mathrm{H}}_4$ is contradicted by the study of Wang et al. [54] for OECD countries. URB leads to a slight decrease in ${\mathrm{C}\mathrm{O}}_2$ emissions, although this effect is relatively modest in OECD countries. Developed economies have managed to decouple the growth of urban areas from the rise in ${\mathrm{C}\mathrm{O}}_2$ emissions. Grodzicki and Jankiewicz [55] used a spatio-temporal approach for Europe for the period 1995–2018 and obtained that ${\mathrm{C}\mathrm{O}}_2$ emissions and URB decreased towards the east and increased towards the north, resulting in increased URB having a detrimental effect on air quality.

A 1% increase in UR leads to a 0.13% long-term increase in GHG, invalidating ${\mathrm{H}}_3$, and therefore, EPC. Finland has a robust social safety net and a strong commitment to environmental sustainability, which could mitigate some of the direct impacts of unemployment on GHG emissions. An increase in unemployment could lead to a shift in the types of energy consumed. For instance, individuals who are unemployed might spend more time at home, increasing the use of residential heating and electricity, which could be less efficient and more carbon-intensive than the energy used in industrial or commercial activities. High unemployment rates can strain government budgets and reduce consumer spending, potentially leading to decreased investments in renewable energy sources and energy efficiency measures. This could slow down the transition to a greener energy mix, leading to a relative increase in GHG emissions. In response to higher unemployment, governments might increase spending on public projects to stimulate the economy. If these projects are not environmentally friendly or if they prioritize short-term economic gains over long-term sustainability, the net effect could be an increase in GHG emissions. Higher unemployment could lead to changes in how people use transportation. For instance, individuals might opt for more affordable but less environmentally friendly transportation options, or public transportation systems might see reduced investment and therefore become less efficient and more polluting. Unemployment affects consumption patterns. While overall consumption might decrease, the consumption of goods with higher GHG emissions relative to their price or utility might increase if they are perceived as necessities or more affordable options. There are several situations where EPC was invalidated, either for Turkey [3,56] or for BRICST countries [16]. On the other hand, Shastri et al. [14] prove the existence of an EPC in India for male unemployment and its absence for female employment.

An ECT value of −1.00 indicates that deviations from equilibrium in GHG levels are corrected almost entirely (around 100%) in the following period. Such a high rate of adjustment suggests a swift return to equilibrium, which might point to either an over-correction or a highly sensitive interaction between GHG and its determinants. From

Table 8. ECM model for short-run estimated results.

Variable	Coefficient	T-Statistics	Prob.
D (GHG (−1))	$-$0.16	$-$2.36	0.02 **
D (GDP)	1.47	8.45	0.00 ***
D (RENC)	$-$1.13	$-$9.85	0.00 ***
D (URB)	$-$3.21	$-$1.43	0.16
CointEq (−1)	$-$1.00	$-$9.55	0.00 ***
R-squared			0.89
Adjusted R-squared			0.87
Durbin–Watson stat			1.99

** and *** indicate the significance of variables at 5% and 1% levels, respectively.

, it can be seen that URB does not significantly influence GHG in the short term. Finland is known for its meticulous urban planning and sustainability goals. Cities are designed to be compact, reducing the need for extensive transportation, one of the major sources of GHG emissions. This compactness, combined with a focus on mixed-use developments, can limit the short-term growth in emissions typically associated with urbanization. Finland’s energy mix is already relatively clean, with a significant portion of its energy coming from renewable sources and nuclear power, which have low GHG emissions. As urban areas expand or densify, the additional energy demand is more likely to be met through these cleaner sources, minimizing the impact on GHG emissions. Finnish cities invest heavily in public transportation systems and infrastructure for walking and biking. This focus helps to reduce reliance on personal vehicles, limiting the growth of transport-related GHG emissions. In the short term, these systems can accommodate the increased demand from urbanization without significantly raising GHG emissions. Finland has stringent building codes aimed at enhancing energy efficiency and reducing heating needs, which is particularly important in its cold climate. New urban developments are likely to adopt these high standards, ensuring that the increase in building stock due to urbanization does not proportionally increase GHG emissions. The Finnish population is generally environmentally conscious, with a strong preference for sustainable living practices. This cultural inclination towards sustainability can help mitigate the potential increase in GHG emissions typically associated with urbanization, as both individuals and businesses may prioritize lower-emission options and technologies. The Finnish government has ambitious climate goals, including becoming carbon-neutral by 2035. Policies and incentives designed to reduce GHG emissions are likely to counterbalance the emission increases due to urbanization. These include investments in green technologies, subsidies for electric vehicles, and schemes to reduce industrial emissions.

The Durbin–Watson statistic is designed to detect the presence of first-order autocorrelation in the residuals from a linear regression. Here it equals 1.99, close to 2, which indicates that there is unlikely to be any autocorrelation in the residuals of the ARDL model. Thus, the model appears to have successfully captured the underlying process without leaving behind patterns in the residuals that would compromise the standard errors of the estimates and, consequently, the validity of hypothesis tests involving the regression coefficients.

Next, FMOLS, DOLS, and CCR techniques substantiate the findings derived from the ARDL model. These methodologies validate the accuracy of the statistical analysis regarding the research variables, especially considering the previously identified cointegration relationships. As depicted in

Table 9. FMOLS, DOLS, and CCR long-term coefficients.

Variable	FMOLS Coefficient, (t-Statistics), [p-Value]	DOLS Coefficient, (t-Statistics), [p-Value]	CCR Coefficient, (t-Statistics), [p-Value]
GDP	0.86 (12.72) [0.00 ***]	0.84 (5.44) [0.00 ***]	0.63 (9.34) [0.00 ***]
RENC	$-$1.21 (35.10) [0.00 ***]	$-$1.26 ($-$24.43) [0.00 ***]	$-$1.17 ($-$35.50) [0.00 ***]
URB	1.31 (7.97) [0.00 ***]	1.34 (3.21) [0.01 **]	1.83 (11.65) [0.00 ***]
UR	0.17 (6.60) [0.00 ***]	0.30 (3.77) [0.00 ***]	0.15 (6.57) [0.00 ***]

** and *** indicate the significance of variables at 5% and 1% levels, respectively.

, the signs of FMOLS, DOLS, and CCR long-term coefficients coincide with those provided by the ARDL model in Table 7.

The null hypotheses of diagnostic and stability tests are shown in

Table 10. Results of diagnostic and stability tests.

Diagnostic Test	${\mathit{H}}_0$	Decision Statistics [p-Value]
Serial Correlation	There is no serial correlation in the residuals	$\mathrm{Accept} H_0$ 1.44 [0.26]
Heteroscedasticity	There is no autoregressive conditional heteroscedasticity	$\mathrm{Accept} H_0$ 0.92 [0.52]
Jarque-Bera	Normal distribution Skewness Kurtosis	$\mathrm{Accept} H_0$ 12.52 [0.18] 1.26 [0.26]
Ramsey reset	Absence of model misspecification	$\mathrm{Accept} H_0$ 0.22 [0.82]

.

The model’s stability is evaluated through the CUSUM test, with the findings illustrated in Figure 3. The CUSUM test asserts the stability of the model’s parameters, as evidenced by the paths of CUSUM staying within the 5% significance threshold, marked by a red dashed line.

5. Conclusions and Recommendations

In our study, we analyzed the dynamics of greenhouse gas emissions (GHG) in Finland over the period 1990–2022, utilizing the Autoregressive Distributed Lag (ARDL) model. The analysis incorporated key variables such as Gross Domestic Product (GDP), renewable energy consumption (RENC), urbanization (URB), and unemployment rate (UR) within the STIRPAT framework to assess the Environmental Phillips Curve (EPC) hypothesis in the specific context of Finland. The cointegration tests indicate a stable long-term relationship among the selected variables, demonstrating complex interdependencies between economic growth, resource use, demographic changes, and environmental impact. Our findings revealed that energy consumption had a directly proportional effect on GHG emissions. This outcome highlights that current energy consumption methods still rely on sources that intensify carbon emissions, reflecting an urgent need for a transition to cleaner alternatives. Additionally, the effects of labor productivity and urbanization demonstrated a long-term negative impact on GHG emissions. Enhancing work efficiency and smartly managing urban expansion can therefore significantly contribute to emission reductions. Furthermore, foreign direct investments did not show a statistically significant impact on GHG emissions, suggesting that local factors such as national energy policies and sustainable development initiatives might play a more decisive role.

Based on our results, the recommendations we can make underscore the importance of accelerating Finland’s transition to renewable energy sources. Government policies should support the development of infrastructure for clean energies, such as wind and solar, and gradually discourage the use of fossil fuels through carbon taxes and reduced subsidies. Moreover, urbanization should be strategically managed to minimize energy consumption. This includes improving building standards for energy efficiency, developing eco-friendly public transport, and supporting smart building technologies that reduce energy requirements. Given the positive impact of labor productivity on reducing emissions, investments in innovation and technology are crucial. These could include support for research and development in carbon capture and storage technologies, as well as in new materials that reduce the carbon footprint of industrial products. Since the unemployment rate has a visible impact on GHG emissions, employment policies should be aligned with sustainability goals. Training and retraining programs could focus on green sectors, thus promoting a circular economy.

Our analysis extends the STIRPAT framework, considering not only the effects of population, affluence, and technology on the environment but also how institutional and economic policies (reflected through the EPC) can influence these relationships. The confirmation of variable cointegration suggests that traditional economic models need to integrate ecological considerations to provide a complete picture of the impact of human activities on the environment.

Our study proposed testing four hypotheses. According to the results obtained, hypothesis ${\mathrm{H}}_4$ was not validated. The absence of an EPC in Finland can be due to several reasons. Finland’s economic and environmental characteristics may differ significantly from those of other countries where the EPC has been validated. Factors such as industrial structure, resource endowment, policy interventions, and environmental regulations may interact differently in Finland, leading to deviations from the expected relationship between unemployment and environmental quality. Finland’s environmental policies and interventions may be particularly effective in decoupling economic growth from environmental degradation. This could result in environmental improvements even in periods of economic downturn or high unemployment, contrary to the expectations of the EPC.

Overall, the absence of validation for the EPC in Finland highlights the need for further research and nuanced analysis to understand the complex dynamics between unemployment, economic growth, and environmental quality in the Finnish context. Several policy recommendations are proposed to mitigate environmental pollution while promoting economic growth and employment in Finland [3,56]. Policies aimed at fostering economic growth and employment should be aligned with sustainable environmental objectives. This entails ensuring that economic expansion does not come at the expense of environmental degradation. Industries with lower environmental footprints should be prioritized, and efforts should be made to promote employment and entrepreneurship within these sectors. Accelerated investments in renewable energy and carbon-neutral technologies include providing tax incentives to incentivize adoption within both direct and indirect industries, as well as streamlining bureaucratic processes to facilitate implementation. Enhancing society’s environmental consciousness is essential for fostering a sustainable environment. This can be achieved through educational training programs, seminars, and public awareness campaigns aimed at promoting environmentally responsible behavior.

As in any study, it is important to acknowledge certain limitations we may have. In our case, it is important to emphasize that our results and conclusions may only be valid for Finland for the analyzed time period. They may be extended to other countries or temporal contexts if the socio-economic development context allows for comparability. Additionally, the STIRPAT model is based on certain assumptions about human behavior and interactions between different aspects of societal development. For example, it assumes that an increase in income will always lead to an increase in energy consumption and consequently to greenhouse gas emissions, which may not always be the case.

In our future research directions, we will consider examining how changes in the behavior of the Finnish population can influence greenhouse gas emissions. This could involve assessing the impact of awareness campaigns or ecological education programs on consumer choices and lifestyles. Additionally, another research perspective may extend our analysis to include international comparisons between different countries and regions to evaluate the effectiveness of policies and identify best practices for reducing greenhouse gas emissions. Furthermore, methodological improvements could be explored to refine the STIRPAT model and better account for the various aspects of socio-economic development.

In conclusion, the study provides a solid foundation for formulating policies that not only stimulate economic growth but also protect the environment, highlighting the crucial role of innovation and energy transition in achieving these goals. Future strategies should consider the complex interactions between economic variables and their impact on carbon emissions, thus navigating towards a sustainable future for Finland. Another open problem is to conduct a gender-segregated analysis in the study of EPC.