Keywords

1 Introduction

The Carbon Limit Adjustment Mechanism (CBAM), a component of the “Fit for 55” climate policy package, is one of the organizations that make up the European Green Consensus, which was founded by the European Union (EU) to reduce greenhouse gas (GHG) emissions by 2050. This technology, implemented as a part of the EU Emissions Trading System, will attempt to minimize the danger of carbon leakage (EU ETS). As a measure of compensating for the price disparity between EU-made and non-EU-made goods, CBAM stipulates the introduction of import levies. This disparity is caused by the differing environmental regulations in exporting nations [1]. Although CBAM is applied to energy-intensive industries such as steel, cement, electricity, fertilizer, and aluminium, it is evident that it will significantly impact the industries and policies of countries. It is because CBAM will cover the subsectors involved in the supply chain and the enterprises that supply these industries with products. Consequently, nations and industries must establish plans for this transformation and settle on substantial courses of action.

The number of studies on CBAM is growing daily in the academic literature. Most research has addressed CBAM analysis, regulatory compliance, and the carbon leakage link. Mehling et al. (2019), for example, conducted a thorough review of CBAM and recommended a new approach after analyzing the legal, administrative, and environmental aspects of regulation [2]. Using data from the Global Trade Analysis Project (GTAP) and pollution from carbon leakage, Naegele and Zaklan (2019) evaluated the issue of carbon leakage in European manufacturing of EU ETS emission costs [3].

The literature also investigates, albeit in a limited capacity, the effects of CBAM on countries [4]. Tang et al. (2015) analyzed the impact of carbon-based border tax adjustments on the Chinese trade and concluded that they would be detrimental to Chinese exports [4]. Zhong and Pei (2022) evaluated the effects of CBAM on changes in competitiveness and welfare, including in various regions of China. It was concluded that the EU's growth in local supply-demand could motivate China to work on climate targets [5]. Overland and Sabyrbekov determined which nations whose industries rely on fossil fuels might effectively withstand CBAM [6]. According to the CBAM Opposition Index, the most significant competitors are Iran, Ukraine, the United States, the United Arab Emirates, Egypt, China, India, Kazakhstan, Russia, and Belarus. Perdana and Vielle (2022) assessed the impact of EU-CBAM and explored eight countermeasures for the Least Developed Countries [7]. It was stated that welfare would decline due to the decline in exports of less developed nations. Most CBAM research has focused on studying the system and how it will affect countries, while only a few have surveyed all nations. In addition, the literature lacks methods, such as data mining, for incorporating the research outcomes into decision-making.

Knowledge discovery, also known as data mining, involves a variety of approaches that may be generally categorized into two groups: descriptive and predictive methods [8, 9]. The goal of applying these approaches to studying massive datasets is to extract knowledge. Typically, descriptive data mining techniques involve statistical summarizing, in which the features of a database are statistically summarized using metrics such as the median, mean, and standard deviation. Although these approaches are less computationally costly, statistics summaries such as “the average daily sales for the firm are $10,425.4” might be challenging to recall. In comparison, statements such as “the average daily sales of the company are around a hundred thousand dollars” are more thorough and simpler to grasp. To construct such summaries, Yager's notion of linguistic summarization has received considerable attention under several names [10], including fuzzy quantification [11, 12], semi-fuzzy quantifier [13, 14], fuzzy association rules [15, 16], and fuzzy rules [17, 18]. Fuzzy sets may be used to represent linguistic words such as “about a hundred thousand dollars,” which is a key step in linguistic summarization. In other words, fuzzy sets can be used to designate database properties during the linguistic summarizing process. Linguistic summarization has applications in diverse fields, such as computer systems [19], investment funds [20], Kansei Engineering [21], human behavior modeling [22], sensor data for elderly care [23], human resources [24], oil price forecasting [25], fall detection [26], social networks [27], and transportation [28], due to its effectiveness in extracting knowledge from large datasets. Please refer to references for further information [29, 30].

Although it has been applied in various fields, its application in sustainable development and CBAM is not yet documented. This study utilized linguistic summarization to investigate the CBAM, which was implemented to assure sustainable development, one of the most pressing concerns facing exporting countries to the EU today. It has contributed to the development of the approach by providing decision-makers and policymakers with relevant linguistic summaries and by adapting the linguistic summary method to a different subject.

2 Materials and Methods

2.1 Preliminaries

A fuzzy set \(A\) on universe \(X\), is defined \(A=\{\langle x,{\mu }_{A}\left(x\right)\rangle |x\in X\}\) where \({\mu }_{A}\left(x\right):X\to \left[\mathrm{0,1}\right]\) is the membership degree of \(x\). Let \(Y\) be the set of objects, \(Y=\left\{{y}_{1},{y}_{2},\dots ,{y}_{M}\right\}\), \(S\) be the set of attributes \(S=\{{s}_{1},{s}_{2},\dots ,{s}_{K}\}\) and \({X}_{k}\) be the domain of \({s}_{k}(k=1,\dots ,K)\). A linguistic summary in the form of “\(Q\) \(B\) \(Y\) s are/have \(A\). . [\(T\)]” consists of four components as (i) a linguistic quantifier \(Q\), , (ii) a linguistic summarizer \(A\), , (iii) a linguistic pre-summarizer \(B\), , and (iv) truth degree of the summary \(T\). Following the extraction of potential summaries from a dataset, the truth degree of each sentence is calculated by Eq. (1).

$$T=Q\left(\frac{{\sum }_{m=1}^{M}({\mu }_{A}\left({y}_{m}\right)\wedge {\mu }_{B}\left({y}_{i}\right))}{{\sum }_{m=1}^{M}{\mu }_{B}({y}_{i})}\right)$$
(1)

2.2 Implementation

This research aims to provide linguistic summaries for decision-makers by employing several variables for countries exporting to EU member states under the scope of CBAM. The steps of the investigation are outlined below.

Step 1: Determining Countries

The initial step is to decide which countries will be included in the study. As a result, 177 countries selling products to any EU country were identified as input. There were 86 countries with exports worth more than $1 billion from these countries. Due to a lack of data, nine countries were dropped from the study, leaving 77 exporting countries as the basis for the research.

Table 1 lists the 27 EU nations included in the study and the 77 countries that export to the EU.

Table 1. Countries
Full size table

Step 2: Determining Attributes

This step identified the attributes of network data employed in the study. Nodes have the attributes of the Environmental Performance Index (EPI) index and population, while relations have the attributes of import value (USD thousands) and distance (km). The first attribute is the EPI, which reveals the environmental performance of countries to safeguard human health and support ecosystem vitality [31]. EPI was developed by Yale University and Columbia University in collaboration with the World Economic Forum and the European Commission Joint Research Center to compare countries' environmental performance. The second attribute determined was the country's population. In studies of countries, the population is a crucial factor [32]. The third attribute is the import value of exporting countries to the EU nations. The last attribute was selected as the distance between countries (km). Also, the current CBAM restriction targeted carbon-intensive industries primarily. To evaluate CBAM with the LS method, GTIP code 68: Objects with stone, plaster, cement, asbestos, mica, and similar materials were selected. Therefore, import value values were collected for this sector.

Step 3: Data Preparation with Fuzzy Sets

A fuzzy c-means approach was used before the modelling phase to automatically identify fuzzy sets and membership degrees of features and quantifiers by referencing fuzzy sets [33]. MATLAB was employed in the identification process [34]. Low, medium, and high are linguistic summarizers used to label fuzzy collections. Fuzzy sets extracted by the fuzzy c-means algorithm are depicted in Fig. 1.

Fig. 1.
figure 1

The membership degrees of the variables: (a) Distance, (b) Import value, (c) Population, (d) EPI

Full size image

Step 4: Modelling

During the modelling phase, polyadic quantifiers were utilized to obtain summary forms [35]. The generated summaries were assessed using the semi-fuzzy quantifiers-based evaluation method [36]. Genç et al. (2020) proposed semi-fuzzy quantifiers to evaluate the summaries in the form of polyadic quantification [37]. The semi-fuzzy iteration operator in the summary form “\(Q\) \(A\) \(Y\) s are in \(R\) with \(Q{\prime}\) \(B\) \(Y\) s” is defined as \(It\left(Q,{Q}{\prime}\right)[A,B,R]\Leftrightarrow Q[A,\{a|{Q}{\prime}[B,{R}_{(a)}]\}]\), where \(Q\) and \(Q{\prime}\) are semi-fuzzy quantifiers, \(A\) and \(B\) are the fuzzy subsets of the universe \(X\) for the attributes \({v}_{1}\) and \({v}_{2}\), \(R\) is a fuzzy relation, \({R}_{({x}_{i})}=\{{x}_{j}|R({x}_{i},{x}_{j})\}\), and \(F\) is a Quantifier Fuzzification Mechanism (QFM). When QFM is applied to \({F}^{I}\) for the finite case, the fuzzy value of the linguistic summary's truth is delivered. For more information, please refer to Genc et. al (2020) [37]

For example, “Most countries export small quantities of the products to EU countries.” is a polyadic quantification, and iteration may be used to express the meaning in terms of its constituents. The “export” is a relation between the sets of “ECs” and “products”. The sentence is valid under one interpretation if and only if a set contains most ECs, each of whom exports a few products.

3 Results and Discussion

Linguistic summaries for all possible combinations of quantifiers and summarizers were generated and evaluated with MATLAB. A total of 1944 linguistic summaries were generated between ECs and EU countries. Thirty-five summaries have a truth degree greater than or equal to the threshold value of 0.9, which is considered reasonable. In the obtained summaries, important features were examined. Summaries with a truth degree higher than 0.9 are given in Table 2.

Table 2. Generated Summaries
Full size table

In this section, the results were illustrated using graphs. Figure 2 highlights the relationship between ECs and EU nations based on linguistic summaries with an accuracy level exceeding the criteria. According to the findings, comparisons were made between exporter nations and only EU nations with a high EPI and a low population. These summaries indicated that EU countries have a higher EPI and a smaller population than exporter nations. As a result, it was anticipated that implementing CBAM would boost the sustainability of exporting countries and, thus, the value of EPI.

Fig. 2.
figure 2

Acquired Network with the LS

Full size image

Figures 3, 4, and 5 depict the networks of the EU countries' qualifiers concerning those of the exporting countries, as determined by the study. In addition, the importance of the accuracy degrees between 0.9 and 1 in the constructed networks ranges from 1 to 9. One represents the highest degree of precision, while 9 represents the lowest degree of precision. This application supported both visualization and analysis.

Fig. 3.
figure 3

Low Population EU Countries and ECs Network for Import Value

Full size image

The statement in Figure 3 with the best level of accuracy among the summaries derived for the ECs' population is that a few ECs with low populations export a substantial proportion of their goods to a few EU countries. However, no summary is given for EU countries with the most inclusive linguistic qualifier “most.” The only summary for ECs with “Most” is that numerous low-population ECs under-export to around half of the EU countries with low populations. According to the summary produced with the second inclusive qualifier “half,” ECs with medium or low populations export very little to half of the EU nations with a low population. The summary containing EPI values reveals that only a few ECs with high sustainable value export to EU nations. According to the collected results, countries selling to half of the EU countries with a low population tend to have low or moderate EPI levels. For exporting nations with high EPI values, no summary with the qualifiers “most” or “half” was obtained. A sample of the EC nations with high export levels for the given product code. High EPI EU countries are average or low EPI countries. According to the collected results, the EPI values for EC are generally low. The majority of ECs have a large population.

Fig. 4.
figure 4

High EPI EU Countries and ECs Network for Import Value

Full size image

Figure 4 illustrates the derived summaries of import value between EC countries and EU members. A small number of ECs with a low or medium EPI export to a small number of EU members with a high EPI. A few CEs with a high EPI export fewer goods to a few EU countries with a high EPI. The conclusion that can be drawn from these summaries is that EC countries must enhance their sustainability. Half of the population-dense EC nations export mainly to a few EU nations. Few EU nations with significant EPI export the most, whereas those with low populations export the least.

Fig. 5.
figure 5

High EPI EU Countries and ECs Network for Distance

Full size image

Figure 5 illustrates the derived summaries of distances between EC countries and EU members. A few ECs with high or moderate EPI are distant from most EU members with low populations. Moreover, ECs with a large population have a shorter distance than ECs with a small population. It illustrates that even if ECs with a large distance have high EPI values, they will face adverse conditions in CBAM.

4 Conclusion

This study examined the application of LS, which aims to give decision-makers supplementary data through the use of datasets, in the fields of the green deal and CBAM, which have recently been on the agenda of many countries. Uniquely, the LS discipline has adapted its usage to CBAM, which has only recently begun to be explored in the literature. Consequently, 1944 unique linguistic summaries were obtained, and 35 of those above the threshold were examined.

It can be expanded in future research by examining all product groupings within the purview of CBAM and integrating the regions, development levels, and energy mixes of these countries as attributes to facilitate efficient decision-making and provide an overview of exporting nations.