Assessing industrialized countries’ environmental sustainability performances using an integrated multi-criteria model and software

Abstract

This study seeks to identify and highlight the differences in the environmental sustainability and environmental protection practices of industrialized countries. For this purpose, thirty industrialized countries were examined using a multi-criteria integrated decision model based on fourteen environmental sustainability performance criteria. An open-source software based on the Python that allows different multi-criteria decision-making methods to be used through a user-friendly interface was developed for evaluations. The results from the developed software rank Sweden first and India last in their environmental sustainability performances. Sweden stands out notably in terms of negative greenhouse gas emission technologies, afforestation, reforestation, environmental laws, carbon tax, bioenergy, and national consciousness. India, on the other hand, does not appear to handle environmental problems in a timely and appropriate manner considering its large population and development needs. Development, utilization, and expansion of renewable energy sources, waste recycling, waste reduction, proper waste disposal, and prioritization of the development of technologies with zero or negative greenhouse gas emissions have emerged as significant factors for environmentally friendly industrialization.

1 Introduction

Industrialization and technical developments make our lives easier and expand our comfort zone. However, whether these facilities and comforts should be prioritized at the expense of irreversible damage to the environment and the livable world for both humans and non-human living beings and to nature, in general, is among the most important debates of the last few decades. A lot of research has been conducted in this area including topics like ecological balance, environmental protection, and environmental sustainability.

The definition of ecological sustainability emphasizes the elimination of adverse effects on ecosystem health while seeking to meet human needs. On the other hand, environmental sustainability, a subset of ecological sustainability, focuses on people interacting with the environment and is defined as carrying out human activities without negatively impacting ecological systems. In other words, environmental sustainability refers to meeting the resource and service needs of current and future generations without endangering the health of the ecosystems that provide them (Morelli, 2011). Without a doubt, the reality of irreversible environmental pollution began with the industrial revolution, and this continues to manifest strongly in today’s industries. At this point in history, our desire to expand the level of comfort we have achieved puts a further burden on the environment. It is thus incumbent on industrialized countries to reduce global damage and set a positive example for developing countries by maintaining environmental sustainability when implementing development programs. This study seeks to identify and highlight the differences in the environmental sustainability and environmental protection practices of industrialized countries. The study evaluates thirty of the world’s largest industrialized countries based on World Bank data using an integrated multi-criteria decision model based on environmental sustainability criteria. The decision model will incorporate both well-known and new multi-criteria decision-making (MCDM), an effort geared toward benefiting from the positive attributes of the various methods while eliminating the negative aspects.

The study’s primary objective is to assess industrialized countries’ environmental sustainability performance and highlight their strengths and weaknesses in this regard. It also aims to provide implications for solving the conflict between industrial development and environmental protection. Hence, it is intended to contribute to the search for industrial development that will allow for environmental protection and sustainability. To accomplish these goals, an easy-to-use Python-based software called ka-decision has been developed. With the ka-decision software, the solutions of different MCDM methods on the same problem can be obtained simultaneously or integrated. Considering its features, the ka-decision software can be used to problems in a wide range of fields. In general, it is expected that the study will make important contributions to the literature and present different perspectives to important stakeholders such as researchers, policymakers and non-governmental organizations. Literature review is provided in the next section of the study.

2 Literature

Many studies on the environment and sustainability are available. It is not possible to include all of these studies here. In this context, the literature review was completed by doing two separate searches on Google Scholar and Scopus for the years 2013 to 2022. The databases examined were chosen because they include a comprehensive collection of scientific papers from various interdisciplinary fields (Negi et al., 2022). The query applied to Scopus is “TITLE-ABS-KEY ("industr*" AND "countr*" AND "environment*" AND ("sustainab*" OR "qualit*" OR "pollution" OR "performance") AND ("multi-criteria*" OR "mcdm" OR "mcda" OR "multi criteria*")) AND (LIMIT-TO (SRCTYPE, "j")) AND (LIMIT-TO ( DOCTYPE, "ar")) AND (LIMIT-TO (LANGUAGE, "English")).” 160 papers were found by this query. On the other hand, the Google Scholar search results obtained by entering the keywords “industrial countries”, “environmental sustainability,” “environmental quality,” and “environmental pollution” were examined for those that included countries’ performance evaluations. After a detailed screening, gray literature such as reports, conference proceedings, notes, and manuscripts unrelated to the studied subject were eliminated. Nine publications discovered through a Scopus search and twenty-three papers found via a Google Scholar search were comprehensively reviewed since they were related to the subject and countries studied. As a result, Table 1 provides succinct summaries of the key findings and practices of these studies.

Table 1 Literature review on environmental sustainability

In the studies in Table 1, country population growth, industry value added, economic growth, environmental performance in the context of urbanization factors, the impact of renewable energy on the environment, the relationship between the clean manufacturing industry and environmental quality, and environmental sustainability analysis are all discussed. Furthermore, energy policies, the interaction between energy resources and the environment, the influence of total industrial reserves on the environment, and the relationship between financial development and environmental pollution have all been studied. At this point, Table 1 shows the use of structurally different methodologies such as artificial neural networks, econometric analysis, and MCDM methods. Among these, MCDM methods are widely used for dealing with complex problems, and they assist managers and decision makers in dealing with various aspects of difficult decisions at the same time, and then selecting the best among them (Gedam et al., 2023). MCDM methods are also commonly used to solve sustainability and environmental problems (Perez-Gladish et al., 2021). This study focuses on the environmental performance and sustainability evaluation of industrial countries, as well as a user-friendly software proposal, with the aim of filling a significant gap for decision makers, practitioners, and the literature in this context. The software, called ka-decision, was created using the Python programming language to achieve the stated aims.

In this context, a multi-criteria integrated decision model will be used to evaluate the environmental sustainability performance of thirty industrialized countries in terms of energy, waste management, air quality, water management, soil and agricultural pollution, biodiversity, and protection. The evaluation will be carried out using the ka-decision software. Thus, unlike many studies in the literature that use MCDM methods, an integrated decision model will be created with user-friendly software support. It is predicted that this methodology will benefit from the positive characteristics of numerous methods while avoiding the problems generated by their undesirable aspects. In sum, this study proposes a methodology for assessing environmental sustainability performances as well as a new software to fill the literature gaps. The following sections will provide information on the MCDM methods that will be used in the study and the proposed software.

3 Methods

There are a lot of methods that may be employed to solve MCDM problems. While these methods produce successful results in a variety of contexts, they are ineffective in solving every decision problem. At this point, the concept of combining solutions obtained by many methods with different structures with aggregation techniques is becoming more common (Aytekin & Orakçı, 2020). The Multi-Attributive Border Approximation Area Comparison (MABAC), Multi-Objective Optimization on the basis of Simple Ratio Analysis (MOOSRA), Preference Selection Index (PSI), Nearest Solution to References-I (REF-I), Nearest Solution to References-II (REF-II), Range of Value (ROV), TOPSIS, and Weighted Aggregated Sum Product Assessment (WASPAS) methods were chosen to solve the decision problem in the study because they have different characteristics such as ease of application, not include the rank reversal problem, and handling criteria with different structures. The results of the MABAC, MOOSRA, PSI, REF-I, REF-II, ROV, TOPSIS, and WASPAS methods will be combined with the Borda technique in this study.

When the literature is examined, the MABAC method application examples include the sustainability assessment of OPEC countries (Ecer et al., 2019), green supplier selection (Xu et al., 2019), green composite selection for clean sustainable energy (Bose et al., 2020), sustainability energy source selection (Rahim et al., 2020), and various field related to sustainable development and circular economy (Torkayesh et al., 2023). The problems of assessing biomedical waste disposal systems (Narayanamoorthy et al., 2020), evaluation of innovative success criteria for hotels (Demirdağ et al., 2021), and solar power plant location selection (Ateş & Topal, 2021) were examined in some studies where the MOOSRA method was applied. It has been seen that the PSI method is applied in transportation company selection (Ulutaş et al., 2021), material selection (Emovon & Oghenenyerovwho, 2020), and selecting a cleaning method for solar panels (Obeidat et al., 2020) problems. REF-I was used in the second-hand vehicle selection problem (Aytekin & Durucasu, 2021), and REF-II was used in the performance evaluation of food companies (Aytekin, 2021a). Among the problems to which the ROV method is applied are cutting fluids selection (Madić et al., 2016), green supplier selection in the textile industry (Karamaşa et al., 2021), and measuring the success of countries in the COVID-19 pandemic (Türkoğlu & Tuzcu, 2021).

TOPSIS has recently been applied in the following areas: smart cities ranking (Hajduk & Jelonek, 2021), evaluation of sustainable development of islands (Zhao et al., 2022), assessment of sustainable urban development (Foroozesh et al., 2022), ranking renewable energy alternatives (Bilgili et al., 2022), and evaluating the sustainable performance of dairy plants (Kumar et al., 2022). Among the problems addressed by WASPAS are the sustainability assessment of renewable and conventional energy sources (Saraswat et al., 2021), landfill site selection (Ghoushchi et al., 2021), evaluation of countries’ sustainable governance levels (Aytekin & Gündoğdu, 2021), and ranking of wastewater reuse allocation alternatives (Azbari et al., 2022). Evaluations of the financial performance of tourism companies (Aytekin, 2019), 5G sector assessments (Peng et al., 2020), and blockchain evaluation system are among the studies that use the Criteria Importance Through Inter-Criteria Correlation (CRITIC) technique to weigh the criteria (Zafar et al., 2021). The application processes of these methods will be briefly detailed in the subheadings that follow.

3.1 MABAC

Pamučar and Ćirović created the MABAC method in 2015, in which ideal and anti-ideal solutions are represented in distinct regions and the best solution is obtained by assessing these regions (Pamučar & Ćirović, 2015). The MABAC method’s processing steps are outlined below (Pamučar & Ćirović, 2015).

Step 1: Construct the initial decision matrix using Eq. (1), where ${\mathrm{x}}_{\mathrm{ij}}$ depicts the ith alternative’s performance, attribute, utility, or preference value in the j-th criterion, where $i = 1, \ldots ,m$ and $j = 1, \ldots ,n$.

$$X = \left[ {\begin{array}{*{20}c} {x_{11} } & \cdots & {x_{1n} } \\ \vdots & \ddots & \vdots \\ {x_{m1} } & \cdots & {x_{mn} } \\ \end{array} } \right]$$

(1)

Step 2: Normalize the decision matrix using Eq. (2).

$$n_{ij} = \left\{ {\begin{array}{*{20}c} {\frac{{x_{ij} - \begin{array}{*{20}c} {\min } \\ i \\ \end{array} x_{ij} }}{{\begin{array}{*{20}c} {\max } \\ i \\ \end{array} x_{ij} - \begin{array}{*{20}c} {\min } \\ i \\ \end{array} x_{ij} }},} & {} & {j \in J^{ + } } \\ {\frac{{x_{ij} - \begin{array}{*{20}c} {\max } \\ i \\ \end{array} x_{ij} }}{{\begin{array}{*{20}c} {\min } \\ i \\ \end{array} x_{ij} - \begin{array}{*{20}c} {\max } \\ i \\ \end{array} x_{ij} }},} & {} & {j \in J^{ - } } \\ \end{array} } \right.$$

(2)

In Eq. (2), ${\mathrm{J}}^{+}$ depicts the benefit-oriented criteria, while ${\mathrm{J}}^{-}$ represents the cost criteria.

Step 3: Construct the weighted normalized decision matrix V:

$$v_{ij} = w_{j} \left( {n_{ij} + 1} \right)$$

(3)

Step 4: Construct the border approximation matrix $G = \left[ {g_{1} , \ldots ,g_{n} } \right]$:

$$g_{j} = \left( {\mathop \prod \limits_{i = 1}^{m} v_{ij} } \right)^{{{\raise0.7ex\hbox{$1$} \!\mathord{\left/ {\vphantom {1 m}}\right.\kern-0pt} \!\lower0.7ex\hbox{$m$}}}}$$

(4)

Step 5: Calculate the distance from G using Eq. (5).

$$q_{ij} = v_{ij} - g_{j}$$

(5)

The position of the alternative i relative to the border approximation matrix is determined using the $q_{ij}$ values calculated using Eq. (5). G border approximation area, $G^{ + }$ upper approximation area, and ${\mathrm{G}}^{-}$ lower approximation area are all referred to as Eq. (6) in this context. Adoption of an alternative as the best solution is proportional to its score in the ${\text{G}}^{ + }$ region of the criteria.

$$A_{i} \in \left\{ {\begin{array}{*{20}l} {G^{ + } ,} \hfill & {} \hfill & {q_{ij} > 0} \hfill \\ {G,} \hfill & {} \hfill & {q_{ij} = 0} \hfill \\ {G^{ - } ,} \hfill & {} \hfill & {q_{ij} < 0} \hfill \\ \end{array} } \right.$$

(6)

Step 6: Rank alternatives using Eq. (7).

$$S_{i} = \mathop \sum \limits_{j = 1}^{n} q_{ij}$$

(7)

The problem is solved by sorting the alternatives in descending order based on the ${S}_{i}$ values.

3.2 MOOSRA

Das, Sakar, and Ray (2012) reintroduced ratio analysis as MOOSRA, based on the basic knowledge provided by Brauers (2003). At this point, it is possible to state that MOOSRA and MOORA have a very close relationship. MOOSRA differs from MOORA in that it adds a separate evaluation for benefit–cost criteria after normalization. The following section summarizes MOOSRA’s implementation steps (Aytekin, 2022; Ray, 2015; Sarkar et al., 2015; Ulutaş & Topal, 2020):

Step 1: Create the decision matrix in the form prescribed in Eq. (1), where the benefit criteria are $j=1,\dots ,g$, while the cost criteria are $j=g+1,\dots ,n$.

Step 2: The vector normalization technique is used to normalize the decision matrix. Vector normalization and 0–1 interval Max–Min normalization, on the other hand, are ineffective in negative and 0-valued decision matrices (Aytekin, 2021b). In such cases, a positive value decision matrix will be generated in this study and the software it contains by applying T-score transformation to the relevant criteria in the decision matrix. The normalization procedure will be conducted over the transformed data in this situation.

$${t}_{ij}=10\frac{\left({x}_{ij}-{\mu }_{j}\right)}{{\sigma }_{j}}+50$$

(8)

The decision matrix is normalized using Eq. (9).

$$r_{ij} = \left\{ {\begin{array}{*{20}c} {\frac{{x_{ij} }}{{\sqrt {\mathop \sum \nolimits_{i = 1}^{n} x_{ij}^{2} } }},} & {} & {{\text{If}}\; T - {\text{score}}\;{\text{ transformation}}\; {\text{is}}\; {\text{not}}\; {\text{employed}}} \\ {\frac{{t_{ij} }}{{\sqrt {\mathop \sum \nolimits_{i = 1}^{m} t_{ij}^{2} } }},} & {} & {{\text{If}}\; T - {\text{score}}\; {\text{transformation}} \;{\text{is}}\;{\text{ employed}}} \\ \end{array} } \right.$$

(9)

Step 3: Construct the weighted normalized decision matrix V:

$$v_{ij} = w_{j} r_{ij}$$

(10)

Step 4: Calculating performance scores based on benefit criteria:

$$b_{i} = \mathop \sum \limits_{j = 1}^{g} v_{ij}$$

(11)

Creating performance scores based on cost criteria.

$$c_{i} = \mathop \sum \limits_{j = g + 1}^{n} v_{ij}$$

(12)

Step 5: Calculate the alternatives’ overall performance score.

$${p}_{i}=\frac{{b}_{i}}{{c}_{i}}$$

(13)

The alternatives are ranked in descending order based on the ${p}_{i}$ values.

3.3 PSI

PSI was developed in 2010 by Maniya and Bhatt as a basic structured method that may be used instead of complex MCDM methods in material selection problems (Maniya & Bhatt, 2010). According to Maniya and Bhatt (2010), the PSI method can handle decision problems without the need for determining criteria weights, unlike other methods that use weighting techniques or subjective judgments. It has also been suggested that PSI can be used to calculate criteria weights where there is disagreement. The PSI method’s processing steps are outlined below (Aytekin, 2022; Maniya & Bhatt, 2010):

Step 1: Construct the decision matrix using Eq. (1).

Step 2: Construct the weighted normalized decision matrix R.

$$r_{ij} = \left\{ {\begin{array}{*{20}c} {\frac{{x_{ij} }}{{\begin{array}{*{20}c} {{\text{maks}}} \\ i \\ \end{array} x_{ij} }}} & , & {j \in J^{ + } } \\ {\frac{{\begin{array}{*{20}c} {\min } \\ i \\ \end{array} x_{ij} }}{{x_{ij} }}} & , & {j \in J^{ - } } \\ \end{array} } \right.$$

(14)

If transformation to the decision matrix is performed prior to normalization due to negative and 0 values, ${\text{t}}_{{{\text{ij}}}}$ values are used in Eq. (14) instead of ${\text{x}}_{{{\text{ij}}}}$ values.

Step 3: Calculate preference variability values using Eq. (15), where ${\overline{\text{r}}}_{{\text{j}}}$ is the mean value of criterion j.

$${\text{PV}}_{j} = \mathop \sum \limits_{i = 1}^{m} \left( {r_{ij} - \overline{r}_{j} } \right)^{2} = \mathop \sum \limits_{i = 1}^{m} \left( {r_{ij} - \left( {\frac{1}{m}\mathop \sum \limits_{i = 1}^{m} r_{ij} } \right)} \right)^{2}$$

(15)

Step 4: Calculate the overall preference value (${\uppsi }_{{\text{j}}}$), for each criterion using the $\phi_{{\text{j}}}$ indicating departures from the $PV_{j}$ values.

$$\psi_{j} = \frac{{\phi_{j} }}{{\mathop \sum \nolimits_{j = 1}^{n} \phi_{j} }} = \frac{{1 - {\text{PV}}_{j} }}{{\mathop \sum \nolimits_{j = 1}^{n} \left( {1 - {\text{PV}}_{j} } \right)}}$$

(16)

Step 5: Create a preference selection index.

$$I_{i} = \mathop \sum \limits_{j = 1}^{n} \left( {r_{ij} \psi_{j} } \right)$$

(17)

The alternatives are ranked in descending order based on the ${I}_{i}$ values.

3.4 REF-I

Aytekin (2020) proposed REF-I to solve decision problems including the criteria and preferences measured with different scales. The criteria measured by nominal (binary and multinomial), ordinal, interval, or ratio scales can be used together, and the decision maker can use these criteria to determine a specific point, range, or category as the reference in REF-I. The REF-I application steps are listed below (Aytekin, 2020, 2022; Aytekin & Durucasu, 2021).

Step 1: Construct the initial decision matrix using Eq. (1)

Step 2: Each criterion’s reference points/ranges are defined by the decision maker. In the context of decision criterion $j$, the reference is a specific value or range of values that the decision maker uses to evaluate the alternatives. If the reference is specified as a specific value for criterion $j$, this value is denoted by ${R}_{j}$. If the reference (${R}_{j}$) is specified as a range, the lower limit is expressed as (${{}_{1}R}_{j}$), and the upper limit is expressed as (${{}_{2}R}_{j}$). In the decision problem, the decision maker may think that the distances from the reference point should be evaluated differently at certain ranges/points or in a specific direction. Values to the right of a reference point/range may appear to be more acceptable than values to the left, or the preferability levels of values to the right and left of different ranges may differ. In this case, the value/range deemed less important than the reference is referred to as the successor value/range. ${{}_{v}V}_{j}$ denotes the successor value/range of criterion j. As a result, it is possible to determine many successor ranges/values for criterion j, where $v = 1,\dots , q$. The decision maker determines whether successor ranges have a symmetric or asymmetric effect on the decision problem. An unacceptance value ($\beta$) is assigned for each successor value/range. $\beta$ denotes the unacceptable level of the value within the successor range in comparison with the reference point/range. Also, $\beta$ has a value between 0 and 1. The following REF-I implementation steps will be detailed by considering the case of determining the reference as a single value.

Step 3: Calculate the distances of alternatives from reference values in qualitative criteria, which has a binary structure, using Eq. (18).

$$d_{ij} = \left\{ {\begin{array}{*{20}c} {0{,}} & {x_{ij} = R_{j} { }} \\ {1{,}} & {{\text{in }}\;{\text{other }}\;{\text{cases }}} \\ \end{array} } \right.$$

(18)

The distances of alternatives from reference values in multi-nominal qualitative criteria are computed using Eq. (19).

$$d_{ij} = \left| {\hat{x}_{{j_{c} }} - \hat{x}_{{j_{R} }} } \right|$$

(19)

In Eq. (19), $\hat{x}_{{j_{c} }}$ is the relative frequency value of $c_{th}$ category in criterion j, and $\hat{x}_{{j_{R} }}$ denotes the reference value. The relative frequency ($\hat{x}_{{j_{c} }}$) is calculated using Eq. (20), where $y_{{j_{c} }}$ denotes the number of observations of the $c$ categorical value in the j criterion, and $m$ is the number of alternatives.

$$\hat{x}_{{j_{c} }} = \frac{{y_{{j_{c} }} }}{m}$$

(20)

The distances of alternatives from reference values in ordinal criteria are computed using Eq. (21), where ${s}_{j}$ denotes the number of ranks in criterion j.

$${d}_{ij}=\frac{\left|{x}_{ij}-{R}_{j}\right|}{{s}_{j}-1}{{}_{v}\beta }_{j}$$

(21)

The unacceptance value of the interval/point v is shown by ${{}_{v}\beta }_{j}$, where $v=1,\dots ,q$.

The distance of alternatives from reference values in cardinal criteria is calculated using Eq. (22).

$${d}_{ij}=\left|{x}_{ij}-{R}_{j}\right|{{}_{v}\beta }_{j}$$

(22)

If the successor point/range is not used, the unacceptance value cannot be included or considered equal to 1 in the calculations. If the reference range is specified, instead of ${R}_{j}$, the closest of ${{}_{1}R}_{j}$ or ${{}_{2}R}_{j}$ according to the ${x}_{ij}$ value is used in Eqs. (19–21-22). If the alternative is within the reference range, the distance from the reference is equal to zero.

Step 4: Construct the normalized decision matrix using Eq. (23).

$${f}_{ij}=\frac{{d}_{ij}}{{\sum }_{i=1}^{m}{d}_{ij}}$$

(23)

Step 5: Construct the weighted normalized decision matrix using Eq. (24).

$${h}_{ij}={f}_{ij}{w}_{j}$$

(24)

Step 6: Calculate general performance values using Eq. (25).

$$U_{i} = \sum\limits_{{j = 1}}^{n} {h_{{ij}} }$$

(25)

The alternatives are ranked in ascending order based on ${U}_{i}$ values.

3.5 REF-II

Aytekin (2020) proposed the REF-II to solve rank reversal problems as well as the requirement to recalculate for current alternatives when a new alternative is added to or removed from the decision problems. On the other hand, REF-II, like most of the MCDM methods, needs criteria having a cardinal structure. The application steps for REF-II are detailed below (Aytekin, 2020, 2021a, 2022).

Steps 1–2 are applied as defined in REF-I.

Step 3–4: Calculate the normalized distance values using Eq. (26).

$$f_{ij} = \frac{{\left| {x_{ij} - R_{j} } \right|}}{{\left| {R_{j} } \right| + 10^{{\rho_{j} }} }}{}_{v}^{{}} \beta_{j}$$

(26)

On the other hand, if the reference is specified as a range, Eq. (27) is used for normalization.

$$f_{ij} = \left\{ {\begin{array}{*{20}c} {\frac{{\left| {x_{ij} - {}_{1}^{{}} R_{j} } \right|}}{{{\text{maks}}\left( {\left| {{}_{1}^{{}} R_{j} } \right|,\left| {{}_{2}^{{}} R_{j} } \right|} \right) + 10^{{\rho_{j} }} }}{}_{v}^{{}} \beta_{j} ,} & {} & {x_{ij} < {}_{1}^{{}} R_{j} } \\ {0, } & {} & {{}_{1}^{{}} R_{j} \le x_{ij} \le {}_{2}^{{}} R_{j} } \\ {\frac{{\left| {x_{ij} - {}_{2}^{{}} R_{j} } \right|}}{{{\text{maks}}\left( {\left| {{}_{1}^{{}} R_{j} } \right|,\left| {{}_{2}^{{}} R_{j} } \right|} \right) + 10^{{\rho_{j} }} }}{}_{v}^{{}} \beta_{j} ,} & {} & {x_{ij} > {}_{2}^{{}} R_{j} } \\ \end{array} } \right.$$

(27)

Steps 5–7 are applied as defined in REF-I. Finally, the alternatives are ranked in ascending order based on ${\mathrm{U}}_{\mathrm{i}}$ values.

3.6 ROV

Yakowitz, Lane, and Szidarovszky proposed the ROV method in 1993, which considers the highest and lowest utility levels that each alternative will provide. The ROV method’s application steps are outlined below (Aytekin, 2022; Ulutaş & Topal, 2020; Yakowitz et al., 1993).

Step 1: Construct the decision matrix using Eq. (1).

Step 2: Construct the normalized decision matrix R.

$$r_{{ij}} = \left\{ {\begin{array}{*{20}c} {\frac{{x_{{ij}} - \begin{array}{*{20}c} {\min } \\ i \\ \end{array} x_{{ij}} }}{{\mathop {\max }\limits_{i} x_{{ij}} - \mathop {\min }\limits_{i} x_{{ij}} }},} & {j \in J^{ + } } \\ {\frac{{\mathop {\max }\limits_{i} x_{{ij}} - x_{{ij}} }}{{\mathop {\max }\limits_{i} x_{{ij}} - \mathop {\min }\limits_{i} x_{{ij}} }},} & {j \in J^{ - } } \\ \end{array} } \right.$$

(28)

Step 3: Calculate the highest utility value for each alternative, where ${\mathrm{J}}^{+}$ denotes benefit criteria.

$$\begin{array}{*{20}c} {u_{i}^{ + } = \mathop \sum \limits_{j = 1}^{n} w_{j} r_{ij} ,} & {} & {j \in J^{ + } } \\ \end{array}$$

(29)

Calculate the lowest utility value for each alternative, where ${\text{J}}^{ - }$ denotes cost criteria.

$$\begin{array}{*{20}c} {u_{i}^{ - } = \mathop \sum \limits_{j = 1}^{n} w_{j} r_{ij} ,} & {} & {j \in J^{ - } } \\ \end{array}$$

(30)

Step 4: Rank the alternatives in ascending order based on ${\text{u}}_{{\text{i}}}$ values.

$$u_{i} = \frac{{u_{i}^{ - } + u_{i}^{ + } }}{2}$$

(31)

3.7 TOPSIS

TOPSIS is one of the most widely used reference-based methods. According to Yoon and Hwang (1980), the solution to the multi-criteria decision problem can be reached by selecting the alternative that is closest to the ideal and farthest from the negative (anti) ideal. TOPSIS implementation steps are summarized below (Hwang & Yoon, 1981; Yoon & Hwang, 1980).

Step 1: Construct the decision matrix using Eq. (1).

Step 2: Construct the normalized decision matrix R using Eq. (9).

Step 3: Construct the weighted normalized decision matrix V using Eq. (10).

Step 4: Create ideal solution values ($A^{*}$) using Eq. (32).

$$A^{*} = \left\{ {\left( {\begin{array}{*{20}c} {{\text{maks}}} \\ i \\ \end{array} v_{ij} |j \in J^{ + } } \right),\left( {\begin{array}{*{20}c} {\min } \\ i \\ \end{array} v_{ij} |j \in J^{ - } } \right)|i = 1, \ldots ,m} \right\} = \left\{ {v_{1}^{*} , \ldots ,v_{j}^{*} , \ldots ,v_{n}^{*} } \right\}$$

(32)

Create anti-ideal solution values ($A^{ - }$) using Eq. (33).

$$A^{ - } = \left\{ {\left( {\begin{array}{*{20}c} {\min } \\ i \\ \end{array} v_{ij} |j \in J^{ + } } \right),\left( {\begin{array}{*{20}c} {{\text{maks}}} \\ i \\ \end{array} v_{ij} |j \in J^{ - } } \right)|i = 1, \ldots ,m} \right\} = \left\{ {v_{1}^{ - } , \ldots ,v_{j}^{ - } , \ldots ,v_{n}^{ - } } \right\}$$

(33)

Step 5: Calculate the distances of alternatives from the ideal solutions.

$$S_{{i^{*} }} = \sqrt {\mathop \sum \limits_{j = 1}^{n} \left( {v_{ij} - v_{j}^{*} } \right)^{2} }$$

(34)

Calculate the distances of alternatives from the anti-ideal solutions.

$$S_{{i^{ - } }} = \sqrt {\mathop \sum \limits_{j = 1}^{n} \left( {v_{ij} - v_{j}^{ - } } \right)^{2} }$$

(35)

Step 6: Determine the closeness values for each alternative.

$$C_{{i^{*} }} = \frac{{S_{{i^{ - } }} }}{{\left( {S_{{i^{*} }} + S_{{i^{ - } }} } \right)}}$$

(36)

The alternatives are ranked in descending order based on ${\mathrm{C}}_{{\mathrm{i}}^{*}}$ values.

3.8 WASPAS

Zavadskas et al. (2012) introduced the WASPAS method by combining the WPM and WSM methods. The WSM method is essentially the same as the SAW method (Mulliner et al., 2016). WPM, on the other hand, incorporates the multiplicative model rather than the additive model and is often referred to as dimensionless analysis (Triantaphyllou, 2000). The WASPAS application steps are defined below (Aytekin & Gündoğdu, 2021; Zavadskas et al., 2012).

Step 1: Construct the decision matrix using Eq. (1).

Step 2: Normalize the decision matrix using Eq. (14). If transformation to the decision matrix is performed prior to normalization due to negative and 0 values, ${\mathrm{t}}_{\mathrm{ij}}$ values are used in Eq. (14) instead of ${\mathrm{x}}_{\mathrm{ij}}$ values.

Step 3: Calculate the WSM priority values for each alternative.

$$Q_{i}^{\left( 1 \right)} = \mathop \sum \limits_{j = 1}^{n} r_{ij} w_{j}$$

(37)

Step 4: Calculate the WPM priority values for each alternative.

$$Q_{i}^{\left( 2 \right)} = \mathop \prod \limits_{j = 1}^{n} \left( {r_{ij} } \right)^{{w_{j} }}$$

(38)

Step 5: Calculate the integrated relative priority values of alternatives.

$$Q_{i} = \lambda Q_{i}^{\left( 1 \right)} + \left( {1 - \lambda } \right)Q_{i}^{\left( 2 \right)} , \lambda = 0, \ldots ,1$$

(39)

The alternatives are ranked in descending order based on ${Q}_{\mathrm{i}}$ values.

3.9 CRITIC

CRITIC will be used to weigh the criteria in the study. Diakoulaki et al. (1995) developed CRITIC for objective weighting that considers the relationship between criteria. The following are the steps in applying the CRITIC technique (Aytekin, 2019; Diakoulaki et al., 1995):

Step 1: Construct the decision matrix using Eq. (1).

Step 2: Normalize the decision matrix using Eq. (40).

$$n_{{ij}} = \left\{ {\begin{array}{*{20}c} {\frac{{x_{{ij}} - \mathop {\min }\limits_{i} x_{{ij}} }}{{\mathop {\max }\limits_{i} ~x_{{ij}} - \mathop {\min }\limits_{i} x_{{ij}} }}} & {j \in J^{ + } } \\ {\frac{{\mathop {\max }\limits_{i} x_{{ij}} - x_{{ij}} }}{{\mathop {\max }\limits_{i} x_{{ij}} - \mathop {\min }\limits_{i} x_{{ij}} }}} & {j \in J^{ - } } \\ \end{array} } \right.$$

(40)

Step 3: Determine the information values of the criteria, where ${\text{r}}_{{{\text{jk}}}}$ shows the correlation coefficient of criterion j with the criterion k, and ${\text{j}},{\text{k}} = 1, \ldots ,{\text{n}}$.

$$C_{j} = \sigma_{j} \mathop \sum \limits_{k = 1}^{n} \left( {1 - r_{jk} } \right)$$

(41)

Step 4: Calculate the criteria weight.

$$w_{j} = \frac{{C_{j} }}{{\mathop \sum \nolimits_{k = 1}^{n} C_{j} }}$$

(42)

In this study, the CRITIC method was used because it considers the relationships between the criteria. As a result, the goal was to reflect the effect levels of interrelated or conflicting criteria on the problem solution objectively.

3.10 Borda Technique

The Borda technique, developed by Jean-Charles de Borda (1781), is commonly employed in social science and/or voting problems. Also, the Borda technique is used for aggregation in MCDM problems. In this context, ${\alpha }_{ik}$ denotes the rank value of the alternative $i$ in the $k$ order. Then, the Borda points for each alternative are computed using Eq. (43), where m shows the number of alternatives (Aytekin, 2022; Aytekin & Orakçı, 2020; Lansdowne & Woodward, 1996).

$$B_{i} = \mathop \sum \limits_{k = 1}^{r} m - \alpha_{ik}$$

(43)

The alternatives are ranked according to their ${B}_{i}$ points in descending order. Borda technique has an understandable and simply applicable structure that produces effective results in most decision problems (Aytekin, 2022; Aytekin & Orakçı, 2020; Lansdowne & Woodward, 1996).

4 Development and introduction of a new decision-making software

This section includes ka-decision, a software developed with the Python programming language to solve decision problems using MCDM methods.

4.1 Software for multi-criteria decision-making

With a limited number of alternatives and criteria, solving decision problems is usually not difficult. However, given the numerous alternatives and criteria, as well as the several methods that can be employed in evaluation, calculations by hand are nearly impossible, and calculations conducted with software not specially designed for MCDM methods are impracticable. Aside from these reasons, the need for MCDM-oriented software also arises from the need to enter the formulas for decision problems into a spreadsheet and repeat the procedure for each method.

There are various software solutions specifically developed for MCDM methods. Thomas Saaty contributed to the development of Super Decisions Software, which is software for AHP and ANP (analytic network process) (Adams & Saaty, 2003). InfoHarvest Inc’s Criterium Decision Plus (CDP) supports AHP and SMART (Simple Multi-Attribute Rating Technique) (InfoHarvest, 2003). The Visual Promethee software is developed to apply PROMETHEE (Preference Ranking Organisation Method) (Mareschal & De Smet, 2009). Another MCDM software Analytica®, which is developed by Lumina Decision Systems, Inc, offers solutions MAUT (Multi-Attribute Utility Theory) and SMART methods (Lumina Decision Systems, 2010). 1000Minds software, a web-based application, provides the implementation of the PARPIKA (Potentially All Pairwise RanKings of all possible Alternatives) method (Ombler & Hansen, 2012). PyTOPS is an open-source software written in Python that solves problems using the TOPSIS (Yadav et al., 2019).

The software presented in this study is a tool that allows for the simultaneous application of MCDM methods and the generation of outputs for the solution of multi-criteria decision problems. The software is expected to appeal to a wide range of users because all necessary operations are performed through the interface and no coding knowledge is required. It is intended to enhance this tool, which is made available to researchers, over time by adding new methods and features.

4.1.1 Architecture of the ka-decision software

The ka-decision software was built in the Python programming language. Python is a popular programming language for data science, although it is not restricted to that. Many code libraries and Frameworks for the Python programming language have been produced and are still being developed by the Python community. Python users can benefit from libraries such as “PyQt” and “WxPython” for interface design while developing desktop apps, “Django” for website infrastructure, “Pandas” for datasets and data mining operations, and “Numpy” for matrix operations. The software is compatible with both Windows and Mac operating systems. The third version of Python must be installed on the PC for this. Furthermore, the Python libraries required by the software must be installed. These libraries are Pandas, NumPy, Math, and Collections used for data operations, and PyQt libraries used for interface design.

4.1.2 ka-decision features

In order to appeal to a wide range of users, the ka-decision software was developed for users who do not know any programming language. As a result, the application’s interface provides visual controls for all processes, from data entry to recording results. The primary functions of the applications are as follows:

• i. The form into which users can enter data can be constructed by entering parameters such as the number of alternatives and criteria.

• ii. Using the file browser, data in the xlsx format can be simply loaded.

• iii. There are error checks for the data loading process. When an error occurs, users are guided by customized error messages.

• iv. Users can access the data entered into the system whenever they want.

• v. MCDM methods to be employed in the application can be selected by clicking the box next to each method.

• vi. CRITIC, Equal, or Entropy objective weighing techniques can be used to weight criteria. Besides this, the user can manually set the weight values if desired.

• vii. Once the previous processes have been completed, all calculations can be performed with the click of a button.

• viii. Ranking results for all methods are displayed together when a button is clicked.

• ix. The tables created throughout the computation phases are combined into a single file. The data set provided in the first worksheet and the method outputs are included in the following worksheets in the Excel workbook.

• x. It has the capability of combining the ranking results received from all methods. After selecting the integration method, this function can be accessed with a single button click.

• xi. There are parts in the software that allow REF-I and REF-II methods to be used with all of their capabilities. To reach these areas, click the tabs at the top.

• xii. There is an option to export all presented data sheets to the xlsx format.

4.2 Usage information of the ka-decision software

When the software is run, an interface that consists mostly of data processing, method selection, weighting method selection, analysis, reporting, and integration components is displayed. Figure 1 depicts the aforementioned interface.

Fig. 1

User interface of the ka-decision software

The diagram in Fig. 2 depicts the ka-decision software’s operation steps.

Fig. 2

Flowchart of the ka-decision software

4.2.1 Data import

The methods under the “Methods and Aggregation” page require data in a specific format. A template form can be constructed first in order to load data in this format. To do so, enter criteria and alternative numbers in the top two fields before clicking the “Generate Data Form” button. This results in the creation of a Microsoft Excel (*.xlsx) file. After the data have been saved in this file, the relevant file is imported into the software.

Following the creation of the data set, the “Import Data” button is activated, and the data file created with the opened file browser is selected and accepted. When the data are successfully loaded, a notification appears indicating that the operation was successful. In the event of a data error, the relevant information is provided to the user. The entered data can be examined and controlled using “Show Data” if required.

The optimization aspects/orientations of the criteria are assigned one of two values: one for maximization and zero for minimization. If data are entered outside of specified ranges, prompt messages appear to allow users to make the appropriate changes. For the criterion weight, one of the various techniques can be chosen in the software. In addition, the user can enter the criteria weights into the data file as he or she considers appropriate. In this case, the criteria weights should be between 0 and 1, with a sum of 1.

4.2.2 Selection of MCDM methods

After the data importing process is finished, the desired MCDM methods should be chosen. To select a method, click the box to the left of the selected option. For the WASPAS method, the λ parameter value should be defined. Also, the default λ value is assigned as 0.5 for the user not being able to determine. A parameter entry has been placed next to the WASPAS method so that users can modify it as they see appropriate.

4.2.3 Criteria weighting function

While the criteria weights entered during the data loading phase can be directly used, one of the weight techniques can also be chosen. The weighting techniques included by the software are as follows:

• Equal weighting

• Entropy weighting

• CRITIC weighting

When one of these techniques is chosen, the criteria weights are calculated, and all methods are processed using these weights. More than one weighting method can be employed with the same data set. There is no need to reload the dataset or reselect the MCDM methods for this. That is, the capability of preserving the results, which will be detailed further below, allows users to preserve all of the results generated by modifying the weighting choice. In other words, the steps of modifying the weight technique, conducting the computation, and saving the results can all be repeated many times.

4.2.4 Calculation process

The calculating step begins when the data have been loaded, the MCDM methods have been selected, and the weighting preference has been made. To complete all computations, users simply click the “Calculate” button. All operations are automated, and the user is notified when the results are available.

4.2.5 Obtaining results

When the computations are complete, the user can click the “Show Ranking” button to show the combined ranking results of all methods. Figure 3 depicts an example screenshot. The notations of the alternatives are on the left side, as well as the names of the methods in the column headings of the result table. The ranking results for the relevant method are displayed in each column. Because its values are 1 in all columns, A3 is found as the best alternative in all methods in the example in Fig. 3. Because of the nature of the methods, the ranking results obtained by MCDM methods may differ. For example, the rankings of the A4 alternative range between 3 and 5 depending on the methods.

Fig. 3

Presentation of the results

Aside from the ranking results, the software allows you to generate additional detailed reports. The output file is stored when the “Save Output Tables” button is clicked after the calculation is finished. Figure 4 shows an example output table. The first page of the worksheet provides the data set employed in the computations as well as the criteria weight values, which have been updated with the weighting technique. The following spreadsheets have outputs based on the attributes of all the selected methods.

Fig. 4

File containing detailed outputs

4.2.6 Integration process

Integration yields a combined ranking based on the rankings acquired by all the methods that were chosen. In the software, there is Borda technique for the integration process. When the computations are finished, the integration result is presented by selecting Borda from the “Aggregation” drop-down list and clicking the “Aggregate” button. Figure 5 depicts an example of the integration process.

Fig. 5

Aggregation results

4.2.7 Detailed REF-I and detailed REF-II

The ka-decision software allows users to choose the newly developed REF-I and REF-II methods, as well as the well-known MCDM methods. In the processes outlined above, it is not possible to use all the features of these methods. Nominal variables, for example, can be used in the REF-I method, but this feature is discarded in the integrated computation. The application pages of the detailed REF-I and detailed REF-II sections in the software’s interface provide guidance on how to use these two methods.

5 Results

Criteria for evaluating countries’ environmental sustainability were determined by utilizing numerous research in the literature. Table 2 lists the criteria, as well as the sources from which the data were gathered and brief explanations. In this context, Table 2 includes criteria for various aspects of environmental sustainability, such as energy, waste management, air quality, and pollution, water quality and pollution, land use and agriculture, biodiversity, forests, and soil degradation.

Table 2 Brief explanations and weights for criteria

CRITIC, one of the objective weighing techniques, was used to weight criteria listed in Table 2. The order of importance of the criteria was obtained as C14≻C11≻C4≻C5≻C8≻C13≻C9≻C1≻C10≻C6≻C7≻​C12≻C2≻C3. “Endangered species” was found to be the most important criterion in environmental sustainability assessment. The Borda aggregation technique was used to combine the solutions provided by the MABAC, MOOSRA, ROV, PSI, REF-I, REF-II, WASPAS, and TOPSIS methods. On the other hand, the World Bank’s “Industry (including construction), value added (current US$)” statistics were used to determine the first thirty industrialized countries. Table 3 shows the countries’ environmental sustainability rankings.

Table 3 Ranking countries based on environmental sustainability performances

It is well known that different MCDM methods can produce different results when applied to the same decision problem (Mahmoud & Garcia, 2000; Mulliner et al., 2016; Zanakis et al., 1998). The main reason for this situation is that the MCDM methods differ in their fundamental principles, the type of data normalization process used, and how the criteria values and weights are included in the evaluation process (Mulliner et al., 2016). For example, when the normalization processes included in the methods are reviewed, it is seen that MABAC and ROV include Weitendorf’s linear normalization, MOOSRA and TOPSIS contain vector normalization, PSI and WASPAS involve 0–1 interval normalization based on max–min values, REF-I normalizes distances from references using sum-based linear normalization, and REF-II employs reference-based normalization (Aytekin, 2021b). Table 3 shows that the methods except MABAC and ROV produced different ranking results while solving the problem. When the discrepancies in the rankings provided by the methods are studied, it is observed that Brazil, which is placed in the top six in general, is ranked sixteenth by REF-I. The main reason for REF’s distinct ranking is the direct processing of distances from references. As a result, Brazil, which ranks twenty to thirty in criteria including C2, C3, C7, C10, C11, and C14, does not benefit as much from the compensatory impact of REF-I solutions as other methods. It is also clear that PSI and WASPAS rankings for Ireland, the Netherlands, and Poland differ from those derived by other methods. PSI and WASPAS both employ the same normalization technique. Another important point is that PSI takes into account distances from the mean. While WASPAS has the highest Spearman rank correlation with PSI, it has relatively lower rank correlation values when the rank correlations of all methods are compared. As previously mentioned in the methods section, WASPAS includes WSM and WPM. WASPAS differentiates itself from other methods by employing exponential weighting in WPM, as well as combining additive and multiplicative models. Moreover, it can be seen that TOPSIS and MOOSRA’s rankings for Russia and Venezuela differ from those produced by other methods. The same normalizing technique is used by the TOPSIS and MOOSRA methods. The use of Euclidean distances from ideal and anti-ideal values in TOPSIS and the ratio of two separate additive values for benefit and cost criteria in MOOSRA produced results that were different from those of other methods, besides the normalizing process. This study aims to find a balanced solution to these disparities by implementing an integrated solution while benefiting from the positive features of methods that do not have permanent advantages over one another.

Sweden takes first place, according to the aggregated results in Table 3. Austria, Brazil, Germany, and Canada are among the top five countries after Sweden. Saudi Arabia, Malaysia, Indonesia, China, and India are the last five countries on the list. The findings are generally consistent with previous research (Chandrasekharan et al., 2020; García-Álvarez & Moreno, 2018; Nash & Steurer, 2021; Sangeetha & Amudha, 2016; Sarkodie, 2021; Tutak et al., 2021; Udemba & Tosun, 2022).

Sweden, Austria, Germany, the Netherlands, and the UK are among the leading countries that have enacted laws to tackle climate change. The goal of these countries’ climate change laws is to drastically reduce greenhouse gas emissions (Nash & Steurer, 2021). Furthermore, the success of Sweden, Germany, and Austria in terms of environmental performance is partly due to the adoption of alternative energy sources in the energy sector (García-Álvarez & Moreno, 2018). Brazil is one of the leading countries in environmental performance. Brazil outperforms other countries in renewable energy generation and renewable water resources. Brazil’s performance in renewable resource output looks to compensate for its comparatively poor performance in some areas, such as greenhouse gas emissions (Udemba & Tosun, 2022).

In the following section, the best performer, Sweden, and the worst performer, India, are evaluated in detail for pairwise comparison of their environmental performances. Figure 6 depicts the criteria scores of Sweden and India using radar graphs based on median values. The reason for choosing the median values is that when computing the mean value, the extreme values misleadingly affect the mean. The graph depicts Sweden in green, India in blue, and the median values in orange. It can be shown that Sweden does not fall below the median in any of the criteria. As shown in Fig. 6, Sweden has a balanced and successful environmental management strategy across all criteria. Sweden is also well above the median score for the C14 (endangered species) criterion, which has the highest importance level determined by the CRITIC. On the other hand, India is significantly below the median values with regard to the C2 (energy consumption), C3 (greenhouse gas emissions), C6 (pollution by particulate matter), C7 (household solid fuels), and C14 (endangered species) criteria.

Fig. 6

Radar charts for Sweden and India

Sweden outperforms other countries in greenhouse gas emissions, particulate matter pollution, wastewater treatment, and forest area criteria. Sweden’s environmental policy is defined by four basic characteristics: an academic information platform that includes natural sciences and technology, an independent legal system, a national consensus on environmental and human health protection, and a desire to be a pioneer in economic modernization (Lönnroth, 2010). In Sweden, there is a comprehensive domestic waste management system, as well as an effort to reduce waste (Andersson & Stage, 2018). Sweden prioritizes the development of climate change mitigation technologies (Tanner et al., 2019). Sweden has successfully reduced greenhouse gas emissions while developing its economy. Swedish negative greenhouse gas emission generating technologies, afforestation, reforestation, carbon tax, and bioenergy that enable carbon capture and storage in energy systems are all prominent. Furthermore, a carbon tax in Sweden is said to be an incentive to reduce energy use, boost energy efficiency, and encourage the adoption of renewable energy sources (Brännlund & Lundgren, 2010; Christiansen & Carton, 2021; Government Offices of Sweden, 2022). Sweden conducts planting and afforestation operations for trees that can be used for a variety of purposes. They employ specific solutions, such as planting willow trees capable of collecting nutrients from wastewater (Börjesson & Berndes, 2006; Dimitriou & Aronsson, 2004; Rosenqvist et al., 2000). In general, Sweden’s environmental awareness is seen in its environmental practices in a variety of fields. Taking all of these practices into account, it is reasonable to say that Sweden’s ecologically sensitive management vision, which has existed for many years, is one of the most essential causes driving its success.

Developing countries prioritize economic development and employment creation over the protection of the environment. It is argued that as income increases, so will ecological consciousness, as well as awareness and desire for environmental protection (Chen et al., 2020). On the other hand, the reality that the environmental degradation we are causing is irrevocable may render economic advancement meaningless. India’s poor environmental performance as a developing country has frequently been linked to excessive poverty and unrestrained population growth (Bhandari, 2013). Furthermore, it has been claimed that Indian municipal governments lack the capacity to collect, handle, and recycle solid and hazardous waste (Chandrasekharan et al., 2020). The Kuznets curve, which depicts the inverse relationship between environmental degradation and economic growth, has been declared valid for India (Aydin & Turan, 2020).

According to the British Petroleum Statistical Review, India is the third greatest CO2 emitter in the world, behind China and the USA (Sangeetha & Amudha, 2016). However, if the country does not focus on cutting unnecessary coal usage, the dream of transitioning to a low-carbon economy would fail. Currently, the country’s economy is still dependent on fossil fuels (Orhan et al., 2021). A strong emphasis on growing GDP to overcome challenges has led India to the verge of exceeding its ecological carrying capacity (Khursheed et al., 2014). India’s new economic liberalization, privatization, and globalization policies, implemented since 1991, have resulted in a rise in foreign direct investments. However, one of the major factors influencing foreign investment in emerging countries such as India is that they maintain environmental standards that are lower than productivity levels (Sajeev & Kaur, 2020). To meet its energy needs, India relies on traditional fuels (gasoline, diesel, natural gas) and crude oil imports. Crude oil imports climbed from 57.8 million tons in 1999–2000 to 140.4 million tons in 2009–10, accounting for almost 81 percent of total oil consumption (Singh et al., 2016).

The most important factor for poor functioning and inadequate utilization of sewerage systems is a lack of motivation, knowledge, operation, and maintenance by technical personnel. In India, just 31% of the population is served by wastewater treatment plants, and many of these plants appear to be running at or beyond their intended capacity or not at all (Khursheed et al., 2014). Small-scale farms and rainfed agriculture dominate Indian agriculture, which employs around 78 percent of the people. Indian agriculture is particularly sensitive to climate change due to a lack of market orientation and a weak ability to cope with extreme weather. More losses happen as a result of unpreparedness, a lack of adequate storage facilities, and the cultivation of a single crop with a single variety. Traditional agriculture is resistant to these effects, but low productivity makes it vulnerable (Patel et al., 2020). It is clear that India suffers from several disadvantages when it comes to ecologically conscious practices. For many years, ignoring environmental issues in favor of focusing solely on economic development has led to huge challenges that cannot be easily overcome in the end. Unfortunately, economic development always takes precedence over environmental challenges in emerging countries.

China is one of the countries that differed negatively from the other countries studied. China has the worst record in terms of energy consumption, greenhouse gas emissions, and pesticide use. However, it is said that China has adjusted its environmental policy concerning clean energy, green agriculture, and sustainable urbanization under the umbrella of “ecological civilization.” It has been stated that Chinese cities have lost their top spot on the list of the world’s most polluted cities to countries such as India and Pakistan (Gürcan, 2021). On the other hand, it is said that the Chinese economy is sustained by polluting trade. Indeed, it has been said that an overdependence on fossil fuel energy and natural resources for economic development, carbon-intensive trade, and carbon-embedded human capital impedes efforts to mitigate climate change and its consequences (Sarkodie et al., 2020).

According to the criteria, the countries considered in the study have varying levels of success. To put it another way, it is not accurate to argue that the top countries are more successful in all criteria, while the others are less successful in all aspects. Every country has a unique approach to environmental issues. The vision and/or imperative requirements of management are responsible for these differences. There are some remarkable points when the ranking results are reviewed on the basis of criteria. The C4 criterion (per capita municipal solid waste ton year) is one of them. India, Indonesia, and China, which are in the bottom five in the overall ranking, look to be performing well in criterion C4. This data only reveals the quantity of waste created per capita; therefore, more study into waste-recording research is required. Austria, Germany, and Canada, which are among the top five countries in the overall ranking, are ranked last out of thirty countries in criterion C4. In this respect, the aforementioned results show that many issues in the leading countries need to be improved. When it comes to criterion C5 (waste management), it is clear that the top three countries (Sweden, Germany, and Austria) have the highest scores, whereas the bottom five countries are Egypt, India, Nigeria, Russia, and Venezuela. As for criterion C7 (household solid fuels), the last three countries, namely India, Indonesia, and Nigeria, have very high values. These countries must act immediately to address the issue of household solid fuel management. Similarly, the bottom-ranked countries such as Malaysia, Indonesia, and Nigeria perform poorly on criterion C8 (wastewater treatment rate). Although China and Malaysia are at the bottom of the overall ranking, they might be considered successful in criterion C9 (renewable water resources). Finally, analysis of criterion C14 (endangered species) reveals that the countries in the first place such as Netherlands, Sweden, and Switzerland are successful, while some countries including Brazil, Indonesia, and Malaysia are unsuccessful. Therefore, it is worth noting that Brazil does not put forth enough effort in this regard.

6 Conclusions and Implications

It is an obvious fact that different countries have different geographies, cultures, and communities. On the other hand, industrialized countries, which achieved economic development and relatively high welfare levels before the rest of the world, serve as role models for developing and underdeveloped countries. However, the environmental degradation resulting from the pursuit of better development standards threatens to leave a planet that future generations may find impossible to inhabit. And that is because, regardless of how developed an economy is, there will be no better world to live in the absence of quality air, clean water, biodiversity, forests, and flora. The negative environmental impact of industrialization considered a critical part of economic development and welfare is undeniable. Industrialized and developing countries thus require a model that can secure both environmental sustainability and economic development. Despite various deficiencies, Sweden can be considered to be the best model among the countries evaluated in the study. Accordingly, the following recommendations were developed based on the characteristics shared by countries at the forefront of environmental sustainability in industrialized countries:

• Practices such as carbon tax have a favorable impact on environmental conservation. Although such taxes may have a detrimental impact on the industry initially, they can have a good long-term impact by encouraging industries to pursue efficiency and modernization.

• Developing policies that increase the proportion of renewable energy in total energy contributes to energy security to the extent that it maintains countries’ environmental sustainability. Given the irreversibility of fossil fuels and the fact that suppliers are frequently in conflict zones, there is an urgent need for all governments to move to low-cost renewable, and ecologically friendly energy sources.

• Waste management systems should be handled through taxation, reduction, and recycling. Waste can be taxed based on volume and weight. Furthermore, activities such as uncontrolled waste release into the environment, burning, and exporting to foreign countries should have serious consequences. Domestic wastewater rehabilitation should receive special attention, and harsh penalties should be implemented to avoid the uncontrolled discharge of wastewater into streams, rivers, seas, and oceans. Recycling water for drinking and domestic purposes is crucial for resource efficiency.

• It would be a good practice to share technologies that produce negative emissions for free with all countries. It should be remembered at this point that environmental damage in one country can have a worldwide influence, and commercial considerations should be put in the background.

• The global action plan for endangered species must be rigorously implemented. It is impossible for a single country or a group of countries to protect the environment and living things by themselves. It is critical to put differences aside and build emergency response measures on a global scale.

• Harmful fertilizers and pesticides can contaminate soil, water, and air, resulting in unfavorable outcomes. In the usage of fertilizer and pesticides, ecologically friendly alternatives must be developed and implemented on a global basis.

• Deforestation has reached alarming proportions in several countries. Reforestation and the establishment of new forests will give dwelling areas for living things and will have a good impact on environmental protection.

• Particulate matter pollution is caused by industrialization and environmental degradation. Measures addressing the particulate matter, which is hazardous to the health of living creatures, should be considered alongside environmental protection policies.

• The global spread, adoption, and actual implementation of the laws of leading countries that mitigate climate change and ensure environmental sustainability will yield good benefits.

• Increasing societal awareness of environmental protection is critical for the laws and policies that will be implemented. Society should be educated and made aware of environmental protection and sustainability challenges, and their corporation obtained.

• Policies that prioritize the environment should be implemented instead of traditional approaches to industrialization, economic development, welfare, and sustainability. The development, use, and expansion of renewable and environmentally sound energy sources, recycling, waste reduction, and waste rehabilitation, as well as prioritizing the development of technologies with zero or negative greenhouse gas emissions, will provide an important framework for industrialization.

While industrialization has caused severe environmental damage, it has also significantly increased societal welfare. The industrialized countries that have aggressively pursued industrialization over the last few centuries have caused tremendous environmental damage. Discussions that began with climate change and global warming are now being carried under the umbrella of sustainability. Developing countries such as India, Saudi Arabia, Malaysia, Indonesia, and China, without a doubt, lag far behind developed countries. Future generations will not have a green, clean, and healthy planet if developing countries continue on the traditional development line that developed countries have taken, and developed countries continue on the path that they have gone on in the past. For developing countries, finding a balance between economic well-being and environmental protection is a difficult task. With the power of technology, it may be possible to resolve such a dilemma. Countries and multi-national enterprises should put aside commercial considerations to ensure that proper waste management, renewable energy, zero-negative greenhouse gas emissions production, and environmentally friendly pesticide and fertilizer manufacturing technologies are widely used around the world. Furthermore, social awareness of environmental protection and climate change must be enhanced, and national and international consensus assured. For example, the degradation or loss of the Amazon forests will have a negative impact not only on Bolivia, Brazil, Ecuador, French Guiana, Guyana, Colombia, Suriname, Peru, and Venezuela but on the entire world. In that situation, environmental protection is an idea that should be shared by all countries and people, not just a few.

Another significant contribution of the study is the development of ka-decision, an open-source software, that will allow the methods to be implemented in the solving of multi-criteria decision problems. The integrated decision model used in the solution of the examined problem, as well as the methods included in the program, can be used to solve problems in a variety of fields. The study, on the other hand, is limited by the number of countries covered, the use of recent data, and environmental sustainability. Additionally, only one technique that uses data structure was used to weigh the criteria.

Appendices

Appendix A. The ka-decision software

The ka-decision software files can be accessed online at https://github.com/kadirkirda/ka-decision.

Appendix B. Supplementary data

Countries	C1	C2	C3	C4	C5	C6	C7	C8	C9	C10	C11	C12	C13	C14
Australia	8.57	5.57	619,260,000	0.56	77.30	107.30	0.94	1.00	492.00	85.87	2.00	19.27	17.42	348.00
Austria	33.70	1.38	67,850,000	0.59	97.20	295.93	1.18	0.94	77.70	135.06	3.30	28.40	47.25	44.00
Brazil	45.02	12.01	1,420,580,000	0.38	65.80	542.74	196.02	0.75	8233.00	304.66	6.00	29.42	59.42	906.00
Canada	27.64	13.63	763,440,000	0.71	84.70	142.22	0.35	0.80	2902.00	111.82	2.40	9.69	38.70	104.00
China	12.67	145.46	11,705,810,000	0.28	51.80	1743.36	471.13	0.29	2840.00	393.22	13.10	15.45	23.34	898.00
Egypt	4.56	3.65	329,400,000	0.24	16.30	3993.22	3.46	0.57	57.30	569.12	2.20	13.14	0.05	98.00
France	11.73	8.70	361,370,000	0.55	94.80	221.07	0.66	0.88	211.00	172.68	3.60	25.79	31.51	125.00
Germany	17.49	12.11	776,610,000	0.61	97.90	334.52	0.40	0.97	154.00	166.48	4.00	37.75	32.68	76.00
India	7.78	31.98	3,346,630,000	0.14	16.10	2706.53	1837.97	0.19	1.91	175.02	0.30	5.97	24.27	809.00
Indonesia	6.09	8.10	1,703,860,000	0.25	49.80	1449.93	1169.11	0.01	2.02	236.44	0.03	12.17	49.07	975.00
Ireland	15.45	0.62	62,290,000	0.60	81.70	162.59	0.67	0.93	52.00	1544.89	6.50	14.44	11.35	45.00
Italy	16.29	5.86	386,780,000	0.50	83.70	357.49	2.08	0.94	191.30	130.59	6.10	21.54	32.13	179.00
Japan	9.31	17.03	1,154,720,000	0.34	86.50	256.02	0.66	0.96	430.00	253.74	11.80	29.39	68.41	205.00
Malaysia	6.01	4.11	388,110,000	0.43	81.40	1101.21	13.44	0.16	580.00	2106.45	8.10	19.12	58.18	948.00
Mexico	7.22	6.48	695,260,000	0.42	74.30	901.62	254.08	0.50	457.20	102.93	1.80	14.50	33.79	832.00
Netherlands	6.54	3.37	179,990,000	0.51	100.00	307.87	0.41	1.00	91.00	265.95	7.90	11.24	10.97	29.00
Nigeria	12.50	4.90	357,520,000	0.18	7.70	1869.21	3505.58	0.02	286.20	19.74	0.00	13.93	23.75	331.00
Poland	6.18	4.01	356,740,000	0.34	91.10	947.92	97.27	0.65	61.60	177.60	2.20	39.65	30.97	34.00
Russia	5.87	28.31	1,992,080,000	0.42	3.20	764.50	25.84	0.25	4508.00	20.81	0.20	9.73	49.78	186.00
Saudi Arabia	0.14	10.56	638,120,000	0.51	61.40	2791.99	7.99	0.25	2.40	81.23	1.50	4.76	0.45	74.00
South Korea	2.54	11.79	673,080,000	0.40	96.70	598.74	0.47	0.84	69.70	369.74	12.40	11.67	64.50	104.00
Spain	16.97	4.97	313,060,000	0.48	89.00	213.02	4.91	0.93	111.50	157.70	3.60	28.07	37.17	366.00
Sweden	42.24	2.20	30,050,000	0.45	99.80	72.74	0.36	1.00	174.00	100.35	0.60	14.88	68.69	32.00
Switzerland	30.64	1.08	43,780,000	0.71	99.00	181.69	0.22	0.97	53.50	187.79	4.70	9.67	32.12	25.00
Thailand	6.19	5.12	431,220,000	0.39	32.90	935.49	240.54	0.50	438.60	148.94	1.70	18.81	38.90	386.00
Turkey	18.47	6.29	473,870,000	0.42	48.50	1274.59	7.99	0.43	211.60	109.73	2.30	0.22	28.87	283.00
United Arab Emirates	0.77	4.19	263,240,000	0.57	26.80	2430.08	0.14	0.98	0.15	745.35	0.00	17.95	4.47	41.00
UK	14.45	6.89	441,130,000	0.46	92.90	280.51	0.24	0.99	174.00	245.62	3.20	28.68	13.19	106.00
USA	8.71	87.79	5,794,350,000	0.81	48.30	249.43	1.02	0.78	3069.00	128.77	2.50	12.99	33.87	892.00
Venezuela	25.39	1.44	277,260,000	0.33	0.00	1213.01	8.97	0.27	1233.00	177.27	1.20	54.14	52.41	218.00