Quantifying Causality between Climate Change and Credit Risk: A Bibliometric Study and Research Agenda (2024)

2.1. Bibliometric Analysis

Bibliometric analysis was first applied in 1969 by Pritchard [7] and has been widely used in academic research. It is commonly used for investigating and analysing large amounts of scientific data [8]. In general, research on the impact of climate change focusing specifically on the analysis of credit risk has been conducted independently of other climate change-related topics [9,10,11]. We used the R software package, Bibliometrix, developed by Aria and Cuccurullo [12], to analyse and visualise the data or information from the data mining of the literature. The R software was used because it offers the bibliometrix package, which has all the tools to complete bibliometric analysis. Vosviewer was used due to its ability to easily interpret relationships between publications in a graphical format [13]. It is another version of bibliometric software that we used to visualise maps or networks. The data mining approach applied in this study was divided into four main steps (as per Figure 1): (i) select the database to query, namely, the Web of Science (WoS), (ii) define the search string, (iii) query the database using the defined search string, and (iv) select the relevant documents (articles, conference proceedings, and reviews) that have undergone the peer review process to ensure quality.

The string that was used to query or search the full records of all the documents in the WoS database was:

(“credit” or “credit risk*” or “credit default” or “default risk” or “default”) and (“climate” or “climate change” or “climate risk*” or “weather changes” or “global heating”).

The search string was run on 12 May 2022 and gave an output, after applying the inclusion and exclusion criteria, of 3138 documents.

The inclusion criteria led to a selection of documents in the English language only and document types labelled as articles, reviews, book chapters, and conference proceedings. The selected documents had had undergone a peer review process prior to their publication thus ensuring their quality [14]. Documents that were less likely to have undergone a peer review process such as letters and unpublished material were excluded from the final dataset used for the bibliometric analysis.

2.2. Systematic Review

A systematic review makes use of studies to collect secondary data and existing evidence to systematically analyse the information [15].

3.1. Bibliometric Results

The summary of the final dataset used in the bibliometric analysis is in Table 2. The table provides a summary of the main information including the timespan, document citations, document types, references, document contents, authors’ information including collaborations. The documents were published between 1991 and 2022, were from 1018 sources, had 11,606 authors, 136,661 references, and a collaboration index of 3.98.

The scientific publications on climate change and credit risk began in 1991 with an annual an publication increase rate of 11.94%. The number of papers published yearly is plotted in Figure 2. A clear upward trend of publications can be seen during the sample period. Noticeably, as opposed to the slow and steady growth trend before 2015, the number of relevant publications increased sharply since 2015 (the year of the aforementioned Paris Agreement), which indicates a substantial increase in interest on the part of academics in climate change and credit risk. The number of publications in year 2022 was lower because it is not full year; the string was run on 12 May 2022; 12 May represents 131 of 365 days.

Table 3 provides a list of the top 20 institutions with the highest number of publications or articles. The Nanjing University of Information Science and National Center for Atmospheric Research are the leading institutions with 63 and 59 articles, respectively, followed by the University of California with 54 articles. These institutions are from China and the USA, and these countries have high production outputs and citations, as reflected in Table 4.

Table 4 shows the top 20 countries contributing to research on climate change and credit risk. The USA is the leading country in this regard with 2402 publications, followed by China with 1121. The USA has thus produced almost double the number of publications than China. Publications from the USA also attract a high number of citations, 5.5 times more than that of China. The World Bank, the United Nations (UN), and many other influential international organisations which are focused on climate change are headquartered in the USA organisations are likely to facilitate and encourage research in the topic area.

Figure 3 lists the most frequently used words based on the authors keywords and keywords plus. “Climate change”, “impact” and “model” are the top three keywords used by authors and keywords plus. “Policy”, “risk” and “carbon tax” are among the used keywords and this clearly indicates the relation to policy around climate change financing and the risks associated. Climate change seems to have attracted much attention.

The country co-authorship network on Figure 4, shows the interrelation between countries. From the figure, the USA is at the heart or center of network, which is not surprising as the USA has the highest number of publications with 2402 publications and highest citations with 25,324 total citations. It is also noticeable that there is regional collaboration with the US and Europe working together, while the the other regions work together as continents.

The most local cited sources on Table 5, shows the journals with the most cited number of articles. These journals have a common scope, which is the science and the environmental affairs.

3.2. Systematic Review

This paper places a special emphasis on quantitatively measuring causality between climate change and credit risk. We therefore selected those papers which employed quantitative methods to measure the causality. Table 1 and Table 6 show a summary of the papers and a classification method to analyse the papers.

The classification groups were created by categorising articles according to their focus or study subject. The first category is climate focus where the majority of the papers quantified causality of climate change and emission rates against credit risk. Second is the credit focus, where the majority of articles examined credit risk factors, followed by the distance to default. Transition risk, index insurance, access to credit, credit rating, and cost of debt credit were not examined much. Third is the credit organisation which offers financial services to those who require them. Banks are leaders in the financial service providers, followed by bond markets and microfinance. Fourth is the sector on which the articles focused, and organisations (firms) comprised the majority sector, followed by the agricultural sector and non-financial enterprises. The fifth category comprises the quantitative modelling methods that were adopted by each study, and the majority of papers in this category applied supervised regression models followed by classification and unsupervised models. This is no surprise, as traditionally, credit risk modelling employs regression methods. The last classification category was the type of data that was applied in the study. Findings show that panel data and credit data were equally used. This validates our systematic method of selection, whereby the inclusion and exclusion criteria involved selecting papers that quantify climate change risks together with credit risk.

The threat of climate change to financial stability may be felt in various ways. Extreme weather and other natural disasters, such as floods and droughts, are causing financial assets to lose value. Asset price adjustments in the transition to a low-carbon economy may pose transition hazards. The International Energy Agency (IEA) and an increasing number of nations are taking steps to reduce greenhouse gas emissions and encourage private investment in cleaner alternatives [16]. As a result, all carbon-emitting assets [17] may be liable to regulatory penalties, affecting investors’ perceptions of future profitability, company sustainability, and creditworthiness. Unexpected and rapid regulatory changes might lead to a fire sale, precipitating a financial catastrophe [18].

Agricultural productivity can be negatively affected by extreme weather conditions, putting smallholder farmers at risk of failing to repay their debts on time [21]. Vanelli and Kobiyama [30] found that a bank’s vulnerability to high default rates is heightened by natural catastrophes, but that the bank’s exposure may be reduced by assigning more green loans to its overall loan portfolio [20]. People, businesses, towns, and other organisations are all affected by the implications of climate change on financial risk because of physical dangers that destroy actual assets such as infrastructure and property. The increasing frequency of harsh weather can lead to losses or damage to assets, devaluation of land and homes and higher default risks in the mortgage portfolio, for example, extreme weather may also affect agricultural productivity and cause farmers to fall behind on their loan payments if the amount they owe exceeds their farm revenue [31].

Previous studies suggest that climate-related risks such as extreme weather events, draughts, floods, etc., impact the stability of lending intermediaries and corporate bond markets [2]. Further, exposure to the climate-related risks negatively affects the ability of small business owners, farmers and retail customers to obtain loans and bonds [24,28]. This highlights the importance of financial regulators and policymakers to begin to evaluate and take into account the impact of climate changes on credit risk [29]. The analysis has shown that the optimization of a loan portfolio should take account of temperature and climatic change outcomes. The fact that climate change is ignored increases credit risk, but this study acknowledges the need for further research to investigate impacts of more severe natural weather patterns due to climate change [19]. The link between a financial institution’s commitment to addressing climate change and credit risks is analysed on this study. These findings suggest that the associated risk of borrowing is reduced as a result of increased attention to climate issues. Financial institutions which have medium to high climate change commitments experience lower credit risks and emphasise the importance of earmarking resources for issues related to climate change [20,21]. Other results show that during the transition to a rapidly evolving energy source, industry sectors such as utilities, minerals, energies and consumer staples are more vulnerable to default. The analysis highlights the potential consequences of failure by large revenue corporations to fulfil their financial obligations and calls for more accurate estimates based on global data [27].

It is essential to understand the factors that are used to quantify climate change risks and credit risks. A majority of the studies employ the following proxies to quantify climate change: Möllmann [21] used remotely sensed vegetation health indices which are vegetation condition index, temperature condition index, vegetation health index to explain climate risk. Other studies employed carbon emissions such as [2,21,32,33], draughts [24,34], floods [28,34], extreme weather and extreme temperatures [20,34], climate change score [20]; to quantify the impacts of climate change on credit risk. Credit risk is widely quantified by using the measures such as distance to default, loan (debt) amount, mortgage refinancing, credit ratings, probability of default [2,19,23,26].

There are at least five quantitative approaches that the reviewed articles applied to model the effects of climate change on credit risk assessment and losses. For example, (i) Ref. [21] used sequential logit models and quantile regressions to investigate the explanatory power of remotely sensed vegetation health indices for credit risk at the aggregated bank level, (ii) Ref. [23] employed support vector machine, random forests ensemble, and artificial neural networks to forecast the credit ratings of eco-friendly firms; and (iii) Ref. [28] applied the logistic regression to examine how lenders account for extreme weather by comparing the mortgage transactions before and after severe floods. These approaches have interesting implications that can be taken into consideration when assessing credit risk.

Climate change events may reduce the value of properties affected, disrupt the price and mortgage rates of neighboring properties, increase borrowers’ default rates, and worsen banks recovery rates from collateral fundamental underlying loans. One of the primary policy tools considered by governments to facilitate the transition to a low-carbon economy is the implementation of a carbon tax that penalises firms that emit greenhouse gases. Companies in high-emission sectors, such as energy (coal, oil, and gas), airlines, and steel manufacturing, may face higher operating costs or significant shifts in asset values, depending on the steepness of the curve, the new (or incremental) taxation regime, as well as their rate of adoption of green technology and/or making use of renewable energy sources.

4.1. Limitations

• This study analysed the documents that were mined (matched) by the search string. Other papers related to this study could have been missed because they used synonyms that were not included on the search string.

• Other studies were excluded because they were not applying or employing any quantitative methods, but they are also exploring the same topic.

4.2. Research Agenda and Future Work

Composite climate scoring for Multiple climatic events: The studies did not consider the possibility of more than one extreme weather event as a result of climate change. Multiple climate change factors, e.g., extreme rain, wind, temperature, floods, etc., can occur at the same time. We therefore consider the following question for further research:

Retail side of credit risk: Most of the studies focused on the corporate side of credit risk, but did not pay attention to the retail customers. Future research should focus on quantifying retail credit risk with respect to climate change effects. The following research questions are therefore proposed: