Bank lending and environmental quality in Gulf Cooperation Council countries

The degradation of the natural environment over the last 100 years, primarily due to increased greenhouse gas (GHG) emissions from anthropogenic activities, has had lasting adverse effects on the climate, threatening both terrestrial and aquatic life on our planet (Basha et al., 2017; IPCC, 2007; IPCC, 2014a, 2014b). Recognizing these threats, governments and businesses around the world have taken serious initiatives to adopt eco-friendly economic models. A major global initiative in this direction is the Paris Agreement of 2015, which was signed by 194 countries and the European Union. This landmark agreement commits its signatories to combat the threats of climate change by limiting greenhouse gas emissions to keep the rise in earth’s temperature to less than 2 ^oC of pre-industrial levels while striving to restrict the increase to 1.5 ^oC. The agreement stipulates several remedial measures to combat the threats of climate change, including channeling finances to low GHG emission economic activities for climate-resilient development (The Paris Agreement, 2015, Article 2.1.c). Out of 159 signatories, 140—responsible for 88% of global GHG emissions—have committed to achieving net-zero emissions. However, the existing strategic initiatives fall short of achieving the goal of 1.5 ^oC, underscoring the urgent need for concerted actions from all the signatories (Climate Action Tracker, 2022; Throp, 2023).

The Gulf Cooperation Council (GCC) consists of six geographically proximal, culturally and politically similar countries: Bahrain, Kuwait, Oman, Qatar, Saudi Arabia, and the United Arab Emirates (UAE). All member countries have pledged to the Paris Agreement, committing to pursue green growth and transition to net-zero economies. However, their economies are heavily reliant on fossil fuel exports. According to the GCC Statistical Center (GCC-STAT, 2024), in 2022, oil sector exports contributed 30% to 54% of the GDP of five out of six GCC countries, the exception being Bahrain, where oil exports contributed around 17% to the GDP. In the same year, the GCC countries contributed 25% to the world oil production and 24% to the world oil exports (OPEC, 2024). Besides, the GCC countries are home to 21% of proven natural gas reserves and contributed 11% to the global natural gas exports in 2022 (OPEC, 2024). Despite that, the policymakers in the region unanimously advocate shifting away from heavy reliance on the hydrocarbon sector to diversify their economies (Al-Saidi, 2021).

The adoption of green growth strategies by GCC countries is expected to boost their economic diversification significantly and steer their collective GDP to over $13 trillion by 2050, more than double the projected $6 trillion GDP under current business practices (World Bank, 2022b). The report emphasizes the leading role of the private sector in broader economic activities such as investment, employment generation, and value creation, all of which align with the pursuit of green growth (World Bank, 2022a; 2022b). The private sector, along with the traditionally dominant public sector, relies heavily on the banking sector for financing. As a result, banks, through green financing, have a critical role to play in combating the threats of environmental degradation in the region.

GCC economies’ current reliance on hydrocarbons, the potential economic gains from green growth strategies, and the expected role of banks in this transition underscore the importance of studying how bank lending to public and private sectors intersects with environmental quality. Past studies have examined the relationship between carbon dioxide (CO2) emissions and financial development and banking sector dynamics, such as banking structure, banking competition, and bank concentration (Kim, Wu, & Lin, 2022; Kim, Wu, & Lin, 2020), and banking development (Zafar, et al., 2019). On the lending front, Kim, Wu, & Lin (2020) studied the impact of bank lending to households and businesses on CO2 emissions, using a panel of 86 developed and developing countries, including four from GCC. Moreover, Wu & Xu (2022) explored how environmental information disclosure affects heavy polluters’ access to bank loans, suggesting that banks are integrating environmental issues into their lending decisions.

The objective of this study is to investigate the impact of bank lending on environmental quality in the GCC. We use two measures of bank lending—lending to the government and lending to the private sector, and two proxies for environmental quality—carbon dioxide (CO2) emissions and greenhouse gas (GHG) emissions. The specific objectives of our study are to estimate the long- and short-term effects of bank lending to the government and private sector on CO2 and GHG emissions at both regional and individual country levels within the GCC.

The rationale for using both CO2 and GHG emissions is twofold: 1) CO2 constitutes about three-fourths of the GHG emissions globally, and researchers have largely taken it as a representative of the environmental quality. Studying CO2 emissions in the GCC will make our findings compatible with those of past studies; and 2) Gases other than CO2 makeup one-fourth of the GHG, but their impact is more severe. Methane, nitrous oxide, and industrial fluorinated gases are 25, 300, and 1000 times more potent than carbon dioxide (United States Environmental Protection Agency, 2022). Furthermore, the GCC member countries’ commitments to mitigate GHG emissions differ as they are tied to the nature of the gases emitted locally (Naber, 2016). We believe that a holistic portrayal of the sources of GHG emissions is imperative for devising robust strategies to counter their threats and achieve a set target of net-zero emissions.

This study contributes to sustainable finance literature by adopting a novel approach to examining how bank lending to public and private sectors impacts environmental quality in the GCC region using two drivers of environmental degradation. Secondly, the study compares the behaviors of CO2 and GHG emissions models at the regional and country levels featuring the complexity of the relationship between explanatory variables and the emissions. Finally, the outcomes of this study provide insights to GCC banks for devising effective strategies for green financing. This research provides recommendations for government and financial institutions to draft policies on overall GHG emissions, rather than focusing only on CO2 emissions. Furthermore, it suggests sharing the best practices to develop a common regional policy, which would facilitate more informed investment decisions for a constructive societal impact.

The remainder of the paper is organized as follows: section “Literature review” reviews related studies, section “The Models” presents the theoretical and econometric models of the study, section “Data and methodology” explains the data and methodology, section “Empirical results” presents the empirical results, section “Discussion” discusses the results, and section “Conclusions” concludes the paper.

The literature review offers insights into the intricate relationships among banking, economic growth, and environmental quality. It emphasizes the pivotal role of banking in facilitating economic development and its subsequent impact on GHG emissions. The review is divided into three sections: the first appraises literature related to banking and economic growth, the second explores the connection between banking and environmental quality, and the third focuses on macro factors affecting environmental quality.

Banking and economic growth

Banking plays a critical role in economic development as it serves as a key driver of liquidity management and facilitates the flow of credit in an economy. Keynes in his book “The General Theory of Employment, Interest, and Money” argues that government interventions through monetary or fiscal stimuli are essential for boosting economy-wide demand. He advocates for an increase in government spending to stimulate economic growth. The budget imbalance caused by increasing spending can be funded by the government through borrowings from banks and other financial institutions, which is one of the four methods used by governments for deficit financing (Fischer & Easterly, 1990). The other methods include printing money, using foreign reserves, and borrowing externally. Furthermore, effective domestic government borrowing is contingent on the existence of robust institutional and macroeconomic foundations, which widen the investor base, expand the market depth, and lower the foreign exchange risk for local currency-denominated debt (Abbas & Christensen, 2009; Hausmann, Panizza, & Rigobon, 2006).

The banking sector serves as a determining factor in the selection of a country’s pathway to economic growth (Gjelsvik, 2017). It acts as a catalyst for this growth by providing businesses access to finances, which is indispensable for the development of any economy (Shaista, 2021). A well-developed banking sector promotes economic growth by reducing financial constraints, directing savings to high net present value investments, and exerting prudence in business decisions (Tonguraia & Vithessonthi, 2018). Additionally, lower financial constraints promote economic growth by granting businesses access to funds at lower costs (Pradhan, et al., 2014; Islam and Mozumdar, 2007; Love, 2003). The stability and intense competitiveness of a well-developed banking sector foster financial innovation and enhance efficiency (Jayakumar, et al., 2018).

One aspect of the banking sector-economic growth relationship is that it is not homogeneous across regions or types of economies. This heterogeneity of the relationship is primarily attributed to the level of depth in banking (Barajas, Chami, & Yousefi, 2013). However, irrespective of the structural differences, a robust banking system is indispensable for the long-term economic growth of any country. The quality of intermediation in the allocation of funds is critical for maximizing factor productivity and minimizing the suboptimal use of the funds (Pagano, 1993).

The banking sector in the GCC is well-capitalized, with capital adequacy ratios (CAR) above the minimum required levels. The strong capitalization is due to high profitability; however, profitability has witnessed a significant drop in recent years due to increased lending and credit growth. One notable feature of the banking sector, which dominates the financial systems of GCC countries, is its substantial public and quasi-public ownership. This ownership shows considerable variation from 13% in Kuwait to over 52% in the UAE. The banks’ exposure to sovereign debt has remained consistent since 2014, albeit the level of this exposure varies among the member countries. Banks’ claims on public debts stood at 20% in Qatar, 13% in the UAE, 11% in Oman, 2% in Kuwait and 4% in Saudi (Al-Hassan et al., 2022).

Banking and environment

The relationship between banking and economic growth, and economic growth and the quality of the environment has prompted researchers to examine how banking affects the environment. However, most studies focus on how various dimensions of banking, such as banking development, banking structure, and bank financing, affect CO2 emissions.

Past studies have reported contradictory effects of baking on CO2 emissions across different countries. Zafar, et al. (2019) observed a negative impact of banking on CO2 emissions in the Group of 7 (G-7) countries, while a positive impact was seen in the Next-11 (N-11) countries. Samour, Isiksal, & Resatoglu (2019) found that banking development has a deteriorating effect on environmental quality in Turkey. Samour, Moyo, & Tursoy (2022) reported a negative impact of banking on the environment in South Africa, mainly due to its heavy reliance on fossil fuel (87%) for primary energy consumption. However, they also assert a crucial role of banking in the betterment of environmental quality through financing alternative renewable energy sources.

Kim, Wu, & Lin (2020) state that CO2 emissions are influenced by the banking structure that emerges in response to the financial choices of corporations. The banking structure, along with the power of banks, dictates the access of households and corporations to financial services and investments. In a subsequent study, Kim, Wu, & Lin (2022) argue that a more competitive and less concentrated banking system promotes green technology, while a moderately competitive and excessively concentrated banking system hampers its adoption, thereby affecting CO2 emissions mitigation.

Obiora, et al. (2020) studied the relationship between banking and environment quality across developing, emerging, and developed economies. They found that commercial banking and domestic credit to the private sector tend to increase CO2 emissions, regardless of the level of development. Ntarmah, et al. (2022) investigated the influence of bank funding on Saharan South Africa’s economic growth and environmental outcomes over the period of 1990 to 2018. They observed that bank financing enhanced economic development and carbon emissions across quantiles. It has a positive impact on economic prosperity and CO2 emissions in East and Central Africa but a negative impact in West Africa. Furthermore, bank funding lessened growth-GHG damage in low-emissions countries, but exacerbated it in median- and high-emissions countries.

Kim, Wu & Lin (2022) investigate the influence of banking lending policies and institutional quality on the level of CO2 emissions. They conclude that lending to enterprises induces the adoption of green technology, thus diminishing CO2 emissions, whereas lending to households tends to add to the atmospheric CO2, particularly in countries with low institutional quality. In addition, higher lending rates in developed economies and higher deposit rates in developing and emerging economies have a downward effect on CO2 emissions (Obiora, et al., 2020).

Macro factors and environment

The relationship between economic development and CO2 emissions has been extensively examined, mostly in the context of the inverse U-shaped Environmental Kuznets Curve (EKC). Several studies have validated this relationship (Holtz-Eakin & Selden, 1995, Bertinelli & Strobl, 2005, Aeknarajindawat et al., 2020), while many others reported inconsistency in EKC’s explanation of CO2 emissions (Wang, Zhou, Zhou, & Wang, 2011; Koc & Bulus, 2020). Dinda (2004) attributes the EKC relationship between economic growth and CO2 emissions to several factors, including structural changes in the economy, increased environmental awareness, income levels, and effective institutional policies.

Another string of research focuses on the relationship between the drivers of economic growth and environmental quality. Mitić, Munitlak, & Zdravković (2017) report that in seventeen transitional countries, GDP increases at the cost of environmental quality, resulting in an increase in CO2 emissions. Similar results were observed by Farhani & Ozturk (2015) in Tunisia, Zou (2018) in America, Malik et al. (2020) in Pakistan, and Wahab et al. (2021) in G7 countries. However, the relationship between GDP and CO2 emission is not always direct. Liu, Guoc, & Xiao (2019) state that a high GDP growth, coupled with technological innovations and efficient use of energy, reduces the level of atmospheric CO2, whereas a low GDP growth rate, along with reliance on mining and hydrocarbon industries, increases the level of atmospheric CO2.

Like GDP, industrialization has been found to adversely influence environmental quality in many parts of the world. Industrial processes, particularly in cement and iron/steel industries, significantly contribute to GHG emissions. However, countermeasures such as using substitutes and recycling effectively reduce these emissions (Liu, Dong, Geng, Lu, & Ren, 2014). Studies by Samargandi (2017), Mahmood (2022), and Azam, Rehman, & Ibrahim (2022) in Saudi Arabia, GCC, and OPEC and all found that industrialization causes a substantial increase in atmospheric CO2. Moreover, industrialization had short-term as well as long-term effects on CO2 emissions in Saudi Arabia (Mahmooda, Alkhateeb, & Furqan, 2020). In contrast, Raheem & Ogebe (2017) studied 20 African countries and observed that industrialization improves environmental quality through its indirect effect on income per capita, which overshadows the direct effect of industrialization on CO2 emissions. Similar results were reported by eight ASEAN and three other nations by Elfaki et al. (2022).

Researchers have also studied the components of trade openness—exports and imports—both individually and collectively, for their impact on the environmental quality. Dou, et al. (2021) report a heterogeneous negative combined effect of trade openness on the environment in China, Japan, and Korea. Azam, Rehman, and Ibrahim (2022) observed similar results in OPEC countries. Contrarily, Mahmood (2022) found a positive effect of trade openness on environmental quality by reducing CO2 emissions. At the individual level, exports tend to improve the quality of the environment, whereas imports deteriorate it (Wahab et al., 2021; Dou et al., 2021). Trade openness can be used as a strategic tool for managing environmental quality by exporting low-emission products and importing high products (Islam, Kanemoto, & Managi, 2016; Majeed et al., 2022).

Other determinants of environmental quality include oil and energy sources. Oil and non-renewable energy are detrimental to the environment, whereas renewable energy is beneficial (Sreenu, 2022; Omri & Saidi, 2022; Malik et al., 2020; Zou, 2018),

To summarize, the impact of economic development on CO2 emissions, the commonly used metric for environmental quality, has primarily been studied through the Environmental Kuznets Curve (EKC) model. Banks, as key liquidity providers, play a crucial role in the economic growth of a country by channeling resources within its economic system. Existing studies focus on the relationship between CO2 emissions and banking sector dynamics, including banking structure, banking competition, bank concentration (Kim, Wu, & Lin, 2022; Kim, Wu, & Lin, 2020), and banking development (Zafar et al., 2019). On the lending front, studies have related CO2 emissions to bank lending to households and businesses (Kim, Wu, & Lin, 2020), commercial banking, and domestic credit to the private sector (Obiora et al., 2020; Ntarmah et al., 2022). However, given the larger role played by banks in an economy, there remains a gap in comprehensively examining how lending to private and government sectors interacts with environmental quality, in terms of CO2 as well as GHG emission. This gap is particularly relevant to the GCC countries that are heavily reliant on fossil fuels. Therefore, we hypothesize the impact of lending to the government and private sectors on not only CO2 emissions but also other GHG emissions, which constitute almost one-fourth of GHG emissions but tend to have a more severe impact than CO2.

Theoretical model

Building on our review of the literature, we propose the following theoretical model, which is illustrated in Fig. 1.

Arrows indicate the direction of effect, while signs represent the anticipated impact.

Full size image

Dependent variables

We use two dependent variables: carbon dioxide (CO2) emissions and greenhouse gas (GHG) emissions, as measures of the environmental quality. The increase in their atmospheric levels due to anthropogenic activities since the Industrial Revolution has been responsible for environmental degradation. CO2 emissions account for three-fourths of GHG emissions, and many studies have used it as a proxy for environmental quality (World Bank, 2007; Andreoni & Galmarini, 2012). The remaining one-fourth of GHG emissions consist of gases that have a more severe impact on the environment (Fei, et al., 2011), and excluding them is tantamount to leaving out a substantial amount of information related to environmental quality. Hence, in agreement with Liu, et al. (2014), Kim, Wu, & Lin (2020), and Kim, Wu, & Lin (2022), we take GHG emissions along with CO2 emissions as comprehensive metrics for environmental quality.

Independent variables

Bank lending is the primary explanatory variable in this study. Banks act as catalysts of economic growth through liquidity management and credit flow in the economy (Shaista, 2021). A well-developed banking sector promotes financial innovation and efficiency, fosters economic growth by minimizing financial constraints, channeling funds to value-generating investments, and enforcing prudence in business decisions (Jayakumar et al., 2018; Tonguraia & Vithessonthi, 2018; Obiora et al., 2020).

We use two lending variables: bank lending to the government (GOV) and bank lending to the private sector (PS). The impact of lending to the government on the quality of the environment depends on the robustness of intuitional and macroeconomic fundamentals, as well as the government’s priorities in the allocation of funds (Abbas & Christensen, 2009; Ntarmah et al., 2022). On the other hand, commercial banks’ domestic credit to the private sector tends to increase CO2 emissions, irrespective of the country’s level of development (Obiora et al., 2020). Although all the countries in the GCC are committed to controlling emissions, their levels of commitment differ. Therefore, we expect positive/negative impact of lending to the government and a negative impact of lending to the private sector on the environmental quality.

Gross domestic product (GDP) is the total gross value added by all producers within an economy. One of the primary ingredients of GDP growth is the use of energy, which releases GHG gases into the atmosphere (Nordhaus, 1977). Nevertheless, past studies have reported mixed effects of GDP on emissions. Mitić, Munitlak, & Zdravković (2017), Farhani & Ozturk (2015), Zou (2018), Malik, et al. (2020), and Wahab, et al., (2021) report a positive impact of GDP on CO2 and GHG emissions, while Holtz-Eakin & Selden (1995), Bertinelli & Strobl (2005), Dinda (2004), and Aeknarajindawat et al., (2020) found a negative impact of GDP on emissions. Given that GCC countries are at different levels of economic development, we anticipate a mixed effect of GDP on emissions.

Manufacturing value added (MVA) represents the net output after subtracting intermediate inputs from the total outputs in an economy, without deducting depreciation of fabricated assets or the depletion and degradation of natural resources. Previous studies from around the globe have found a detrimental effect of MVA on environmental quality (Mahmood, 2022; Azam, Rehman, & Ibrahim, 2022; Samargandi, 2017). The GCC economies are predominantly dependent on fossil fuels, and MVA is expected to have a worsening effect on environmental quality.

Trade openness (TRD) is the sum of exports and imports of goods and services as a proportion of a country’s GDP. Researchers assert that trade openness scales up the level of production, resulting in a substantial rise in energy consumption and an increase in emissions (Ahmed, Rehman, & Ozturk, 2016; Islam, Kanemoto, & Managi, 2016; Dou, Zhao, Malik, & Dong, 2021). However, studies have shown that exports have a positive impact on the environment, while imports have a negative impact (Wahab et al., 2021; Dou et al., 2021). The GCC countries are net exporters of emissions as they contribute nearly a quarter to the world’s fossil fuel exports. Therefore, trade openness is expected to have a positive effect on environmental quality in the GCC.

Econometric model

The theoretical model is transformed into the following two econometric models related to CO2 emissions and GHG emissions:

where CO2 and GHG represent emissions of carbon dioxide and emissions of greenhouse gases in the atmosphere, respectively. PS is the claims on the private sector by banks and GOV is the claims on the government sector by banks. GDP stands for gross domestic product per capita, MVA for manufacturing value-added, and TRD for trade openness. i stands for country, t for time, and ({epsilon }_{{it}}) for the error term.

Data

We used twenty-year data for CO2 emissions, spanning 2001 to 2020, and nineteen-year data for GHG emissions, from 2001 to 2019, in six GCC countries: the UAE, Bahrain, Kuwait, Oman, Qatar, and Saudi Arabia. We combined annual data from these countries over their respective periods to create balanced panels, resulting in 120 country-years for CO2 emissions and 114 country-years for GHG emissions.

Variable measurement and data sources

CO2 emissions and GHG emissions are measured by the volume released into the atmosphere. CO2 is the major constituent of GHG, which also includes methane, nitrous oxide, and industrial fluorinated gases. The data on CO2 emissions, measured in tons, was obtained from Our World in Data and the GHG emissions data, in million tons of carbon dioxide equivalent units, was sourced from Climate Watch. To ensure uniformity in measurement scales, we converted the CO2 emissions data into million tons.

Claims on the central government by banks (GOV) comprise of loans extended by banks to the central government institutions, net of deposits. The claims on the private sector by banks (PS) include gross credit from banks to individuals, enterprises, and nonfinancial public entities not included under net domestic credit. The data for both variables was obtained from World Development Indicators, with compiled data sourced from central banks.

Gross domestic product is the total gross value added by all producers within an economy. It includes product taxes and excludes subsidies from the value of products. Gross domestic product per capita (GDP) is computed by dividing gross domestic product by midyear population. GDP data was sourced from the World Development Indicators.

Manufacturing value added (MVA) refers to the net output of a country’s manufacturing sector after subtracting intermediate inputs from total outputs, without deducting depreciation of fabricated assets or depletion and degradation of natural resources. The MVA data was acquired from the World Development Indicators.

Trade openness (TRD) represents the sum of a country’s exports and imports of goods and services as a proportion of its gross domestic product. The TRD data was obtained from the United Nations Industrial Development Organization (UNIDO).

Methodology

We employ the panel autoregressive distributed lag (ARDL) method of Pesaran and Shin (1998) to estimate our models. The rationale for using panel ARDL is that it allows simultaneous study of both long-run and short-run relationships among modeled variables. Furthermore, we use pooled mean group (PMG) ARDL estimators because they assume homogeneity in the long run while allowing for heterogeneity in the short run (Pesaran, Shin, & Smith, 1999). The generalized ARDL versions of models (1) and (2) are follows:

where, LCO2 and LGHG represent the log of CO2 emissions and the log of GHG emissions, respectively. LPS is the log of the claims of the private sector by banks, and LGOV is the log of the claims of the government sector by banks. LGDP stands for the log of gross domestic product per capita, LMVA for the log of manufacturing value-added, and LTRD for the log of trade openness. ({delta }_{i}) represents the coefficient of the lagged dependent variable; p and q are the optimal lags for regressand and regressors, respectively; i and t countries and time, respectively; ({varphi }_{i}) country-specific fixed effect; and ({varepsilon }_{{it}}) the error term.

The generalized ARDL models are reparametrized into error correction models (ECM) to obtain short-run dynamics and long-run relationships. The cointegration factor integrates short-run dynamics with long-run equilibrium while preserving the long-run relationship. The reparametrized ECM models for (3) and (4) are of the following form:

where γ and ({alpha }_{1,2,ldots .,5}) represent short-run coefficients; ({{boldsymbol{theta }}}_{i}=-left(1-{delta }_{i}right)), country-specific speed of adjustment coefficient; (left[{{LCO}2}_{i,t-j}-{{boldsymbol{beta }}}_{i}^{{prime} }{{boldsymbol{X}}}_{i,t}right]) is the error correction term; ({{boldsymbol{beta }}}_{i}^{{prime} }) is the vector of long-run relationship, and ({{boldsymbol{X}}}_{i,t}) is the vector of regressors (LPS, LGOV, LGDP, LMA, and LTRD).

One of the requirements of ARDL is that none of the variables in the model should be integrated of order two. We performed unit root tests to determine the stationarity of each variable before estimating the model. The choice of unit root tests depends on the presence or absence of the cross-sectional dependence in data variables, for which we used Breusch-Pagan LM, Pesaran scaled LM, Bias-corrected scaled LM, and Pesaran CD tests. In addition, to establish model fitness, we report root mean square error (RMSE) and residual graphs.

Descriptive statistics

The average annual CO2 emissions in the GCC over the period of study were 148 million tons, with a standard deviation of 172 million tons. In comparison, the average GHG emissions were 175.719 million tons, with a standard deviation of 190.82 million tons. The mean bank lending variable to the private sector, PS, stood at 337.35 billion, with a standard deviation of 453.94 billion, whereas the average bank lending to the government, GOV, was 103.632 billion, with a standard deviation of 154.77 billion. The average GDP in the region was $31,760, with a standard deviation of $18,002. The average of MVA and TRD, as proportions of GDP, were 10.81 and 108.59, respectively, with standard deviations of 3.09 and 31.43. The t-statistics for all mean values are greater than 2, confirming that the mean values are significantly different from zero (for details, see Table 1).

Table 2 shows the summary statistics at the country level. Saudi Arabia is the largest CO2 emitter, with an average emission of 500 million tons per year, which is more than three times the 157 million tons emissions of the second largest emitter, the UAE. With annual average emissions of 26.7 million tons, Bahrain contributes the least to atmospheric CO2 in the region. A similar pattern is observed for GHG emissions, where Saudi Arabia leads with an average annual emission of 563.769 million tons, followed by the UAE with mean annual emissions of 194.7 million tons. Here also, Bahrain contributes the least to the emissions, with an average of 40.127 million tons per year. CO2 is the major component of GHG emissions, but its percentage contribution to total GHG emissions varies widely across the member countries. The share of CO2 in overall GHG emissions ranges from 663% in Bahrain to 95.4% in Qatar, with an average share of 84.3% at the regional level.

Saudi Arabia also leads the region in bank lending to the private sector, PS, with an average outstanding amount of 895.31 billion, while Bahrain had the least PS, with a mean annual amount of 6.24 billion. On the other hand, the average lending to the government, GOV, ranged from 2.455 billion in Oman to 294.759 billion in the UAE.

The GDP across the GCC countries is quite dispersed. The average GDP per capita ranges from $16,934 in Oman to $60,627 in Qatar, with the UAE holding a distant second place with a GDP of $38,640. The range of MVA is from 7.79 in Kuwait to 16.78 in Bahrain, while TRD falls between 75.69 in Saudi Arabia and 147.18 in Bahrain.

Stationarity of the variables

Prior to conducting the stationarity test, we assessed the cross-sectional dependence of the variables in the panel using Breusch-Pagan LM, Pesaran scaled LM, Bias-corrected scaled LM, and Pesaran CD tests (see Table 3)

The test statistics of all four tests for all variables are significant at a 1% level, leading to the rejection of the null hypothesis of no cross-sectional dependence in the variables. The only exception is the Pesaran CD test, which does not reject the null hypothesis for LMVA. The observed cross-sectional dependence is not surprising as the GCC countries have geographic proximity, share social and cultural values, maintain close political and economic ties, and have witnessed simultaneous economic growth.

The presence of cross-sectional dependence of variables necessitates the use of second-generation panel unit root tests for assessing their stationarity. We applied two second-generation unit root tests: the Panel Analysis of Nonstationarity in Idiosyncratic and Common Components (PANIC) test of Bai and Ng (2004) and the Cross-sectionally Augmented IPS (CIPS) test of Pesaran (2007). The PANIC tests include cardinality of retained factors and idiosyncratic pooled tests with respective null hypotheses of ‘retain common factors’ and ‘no cointegration among all cross-sections’. The null hypothesis of CIPS tests is ‘the distribution has unit root’. PANIC and CIPS tests allow testing of stationarity without constant, with constant, and with constant and trend (Hlouskova & Wagner, 2005). The stationarity test results are shown in Table 4.

The cardinality test reveals that all factors are retained in all variables. The PANIC idiosyncratic pooled test of Bai and Ng indicates stationarity with constant in all variables, except LGOV, by rejecting the null hypothesis of no cointegration among all cross sections. However, with constant and trend, the null is rejected in all variables, including GOV, except for GHG, the distribution of which does not show any trend. Likewise, Pesaran CIPS rejects the null of the unit root in all variables at constant, except LGOV. The results of CIPS with constant and trend are almost similar to those obtained with constant only.

ARDL estimation

This section presents the empirical results of the CO2 emissions model followed by those of the GHG emissions model.

Carbon dioxide emissions

Table 5 shows that both bank lending variables, PS and GOV, have a long-run detrimental effect on the environmental quality in the GCC region. Both variables add to the level of atmospheric CO2, with the marginal effect of PS being nearly five times greater than that of GOV. In addition, GDP, which is significant at 10%, also contributes to the atmospheric CO2, further deteriorating environmental quality.

On the contrary, business activity variables—MVA with a positive coefficient and TRD with a negative coefficient—do not have a statistically significant impact on CO2 emissions in the long run. Furthermore, none of the variables in the model have a short-run effect on CO2 emissions.

The error correction term, represented by the cointegration coefficient, has a significant negative coefficient of 0.65, which affirms the existence of a long-run relationship between explanatory variables and CO2 emissions in the GCC region.

The root mean square error (RMSE) of the model, which represents difference between the actual values and the model’s average prediction, is 6 percent, and the sum of squared residual (SSR) or unexplained variance in the data is 42.6 percent.

Cross-country carbon dioxide emissions

Table 6 presents cross-country short-run ARDL estimates for countries in the GCC. The impact of explanatory variables on CO2 emissions varies significantly in both nature and magnitude across these countries. The cointegration coefficients range from −0.21 (Qatar) to −1.11 in Kuwait, which is the only country with a coefficient below −1. The coefficients confirm the long-term effect of the explanatory variables on CO2 emission in all countries, with Kuwait showing oscillatory convergence to the long-term equilibrium.

The influence of the banking variables varies significantly within the countries under study. In Bahrain, Oman, and Qatar, both variables have a harmful effect on environmental quality, whereas in the UAE, they have no impact. In the first group, the marginal effect of PS is greater than that of GOV. In Kuwait and Saudi Arabia, PS does not impact emissions, while GOV reduces CO2 emissions, improving environmental quality. GDP appears to be the weakest driver of emissions, with a negative impact on the emissions in Kuwait and a positive impact in Qatar. MVA displays contrasting effect on the emissions: it improves the quality of the environment by reducing atmospheric CO2 in the UAE, Bahrain, and Qatar, but worsens it by increasing atmospheric CO2 in Kuwait and Saudi Arabia. Moreover, the marginal effect of MVA is strongest among all variables in four countries. TRD has a positive impact on environmental quality in Bahrain, while it has a negative impact in the UAE and Oman.

To summarize, the behavior of the CO2 model varies significantly among the countries under study. In the UAE, only MVA and TRD are significant, with positive and negative effects on CO2 emissions, respectively. Bahrain reveals a positive impact of the two lending variables, along with a negative impact of MVA and TRD. In Kuwait, GOV and GDP negatively affect CO2 emission, but MVA has a positive impact. In Oman, two lending variables, along with TRD, are significant contributors to CO2 emission. In Qatar, all variables, except TRD, are significant, with MVA having a negative impact on emissions. Lastly, Saudi Arabian environmental quality is affected positively by GOV and negatively by MVA, with the latter exerting a stronger influence.

Greenhouse gas emissions

The test results of the GHG emissions model reveal that all variables in the model have a significant long-term impact on GHG emissions in the GCC region (Table 7, Panel A). With the exception of TRD, all variables contribute to an increase in GHG emissions, leading to a deterioration in environmental quality. However, the marginal effect of TRD, the environment-friendly variable, is the strongest among all variables. The coefficients of variables in the model have the same signs as those in the CO2 emissions model. The cointegration coefficient of the model is −0.34, indicating the presence of long-run relationship among variables as modeled.

Among all variables, only GOV and MVA have a significant but contrary, short-run impact on GHG emissions. GOV reduces emissions and has a positive effect on environmental quality, whereas MVA increases emissions and has a negative impact. However, the absolute marginal effect of MVA is substantially higher than that of GOV (for details, see Table 7, Panel B).

The RMSE and SSR of GHG model are lower than those of the CO2 model, indicating that the GHG model is a better fit model. The RMSE of the GHG model is less than 2% compared to 6% for the CO2 model, while the SSR is 4.2% for the GHG, compared to 42.6% for the CO2 model.

Cross-country greenhouse gas emissions

The cross-country results of the GHG model in Table 8 reveal differences in its behavior across the GCC member countries. For all countries, the cointegration coefficients range from −0.88 to −0.003, and they are significant in all cases except Saudi Arabia. The significance of cointegration coefficients indicates the existence of a long-run relationship between explanatory variables and GHG emissions. The variation in the coefficients reflects the difference in the pace of adjustment to long-term equilibrium in response to short-term shocks, similar to the behavior observed in the CO2 model.

The influence of bank lending variables on GHG emissions is quite diverse across countries. The first variable, PS, benefits the environment in the UAE and Kuwait, where it lowers emissions, but harms the environment in Oman, Qatar, and Saudi Arabia, where it increases emissions. Notably, in the latter group, the other lending variable, GOV, has a counteracting effect by improving the environmental quality through reduced GHG emissions. Similarly, GOV positively impacts the environmental quality in the UAE and Bahrain. The only exception is Kuwait, where GOV adversely affects environmental quality by contributing to increased GHG emissions.

Like bank-lending variables, non-banking variables also exert varying effects on GHG emissions. GDP is significant in all countries; however, it has a positive impact on emissions in Bahrain, Kuwait, and Saudi Arabia, while having a negative impact in the UAE, Oman, and Qatar. MVA, on the other hand, contributes positively to GHG emissions in all countries except Kuwait. Finally, TRD positively impacts emissions in the UAE, Oman, and Qatar, but has a negative effect in Bahrain and Saudi Arabia.

Within the GCC countries, the behavior of the drivers of environmental quality varies. In the UAE, MVA and TRD have a negative impact on environmental quality, while the remaining variables have a positive impact. The marginal effect of the environment-damaging variables is more severe. In Bahrain, MVA and GDP harm the environment, while TRD and GOV improve it. The environment in Kuwait is damaged by GOV and GDP, whilst MVA and TRD benefit it. As in the UAE, all variables are significant in Qatar and Saudi Arabia. In both countries, MVA and PS degrade the environmental quality, whereas GOV enhances it. GDP and TRD have opposing effects on the environment; GPD improves the environment in Qatar but destroys it in Saudi Arabia, while TRD improves the environment in Saudi Arabia but destroys it in Qatar.

We examine the long-term and short-term effects of bank lending to the government and private sectors on environmental quality in the GCC countries, using CO2 and GHG emissions as proxies for environmental quality. Alongside bank lending, we use GDP, MVA, and TRD as control variables. The test results from the CO2 and GGH models reveal considerable differences in their behaviors.

Starting with the long-term effect at regional level, we find that lending variables, PS and GOV, have significant impact on the emissions of both CO2 and GHG. These findings align with those reported by Obiora et al. (2020) and Ntarmah et al. (2022) on CO2 emissions.

GDP also has a positive long-term influence on both types of emissions. Similar results on CO2 emissions were observed by Mitić, Munitlak, and Zdravković (2017) and Wahab, et al. (2021). Furthermore, MVA and TRD have long-term effects only on GHG emissions: MVA contributes positively to the emissions, while TRD contributes negatively. The absence of the effect of MVA on CO2 emissions contradicts the findings of Azam, Rehman, & Ibrahim (2022), who reported a positive effect in GCC and OPEC, as well as those of Elfaki et al. (2022), who found a negative influence in ASEAN countries.

Among all variables, TRD is the only one with a negative influence on GHG emissions, which is in consonance with the results reported by Antweiler et al. (2001) and Frankel & Rose (2005). Additionally, it has the highest absolute marginal effect on GHG emissions. The insignificance of TRD in CO2 emissions disagrees with the findings of Dou, et al., (2021) and Azam, Rehman, & Ibrahim (2022).

In the short run, none of the explanatory variables affect CO2 emissions. However, two variables, GOV and MVA, have a negative and a positive effect on GHG emissions, respectively.

At the country level, the findings of the two models are quite revealing. The nature of the impact of explanatory variables is not consistent across countries in either model; similar observations have been reported by Liu, Guoc, & Xiao (2019). One striking difference between the models is the absence of a composite effect of all explanatory variables on CO2 emissions in any country, which contrasts with the presence of such an effect on GHG emissions in four countries. In two other countries, excluding one variable each—PS in Bahrain and TRD in Kuwait—all other variables have a significant impact on GHG emissions.

Another noteworthy aspect of our findings is the opposing effect of some explanatory variables in two models within the same country. For instance, in the UAE and Qatar, MVA has a negative effect on CO2 emissions but a positive impact on GHG emissions. Conversely, in Kuwait and Saudi Arabia, it has a positive impact on CO2 emissions and a negative impact on GHG emissions. Similarly, GDP decreases CO2 emissions while increasing GHG emissions in Bahrain, whereas in Qatar, it increases CO2 emissions while decreasing GHG emissions. In addition, GOV adds to CO2 emissions in Bahrain and Oman, while it reduces GHG emissions. Table 9 shows the signs of significant variables in both models. The variations in the behavior of CO2 and GHG emissions models in the short run reflect the dissimilarities in economic structures and policy approaches across GCC countries.

Our model results raise two important questions. First, what explains the difference in the short- and long-run behaviors of the explanatory variables? Second, what causes cross-country differences in the drivers of CO2 and GHG emissions? We examine the policy documents of the GCC nations for plausible explanations.

The observed differences in the short- and long-run behaviors of the models can be attributed to the disparity between the duration of our study and the timings of the initiation of actions to curb emissions. Our study spans over 20 years for CO2 and 19 years for GHG, while any significant environmental initiative by a bank in the GCC occurred in 2016 when the First Bank of Abu Dhabi committed $10 billion to lend and invest in green projects (Ministry of Climate Change & Environment, 2021).

Before delving into the cross-country differences, it is important to highlight the negative impact of GOV on GHG emissions at the regional level. This reflects the commitment of all GCC nations to control GHG emissions and represents an initial outcome of their concerted steps in this direction. These efforts include the establishment of carbon capture, utilization, and storage (CCUS) facilities by the leading companies in oil and petrochemical production, such as Aramco and SABIC in Saudi Arabia, ADNOC in the UAE, Qatar Petroleum in Qatar, and BAPCO in Bahrain (International Monetary Fund, 2021). Notably, Saudi Arabia, the UAE, and Qatar collectively account for 10 percent of the CO2 captured globally (Al-Sarihi, 2023).

Cross-country variations in the explanation of emissions can be ascribed to the differences in their levels of commitments and targets set for GHG emissions control. According to the Middle East Institute (2023) report on achieving net zero in GCC countries, the UAE and Oman aim to achieve net zero emissions status in 2050, while Bahrain, Kuwait, and Saudi Arabia target this goal in 2060. Furthermore, emission reduction targets for 2030 are set at 7% by Oman, 25% by Qatar, and 31% by the UAE. Saudi Arabia’s goal is to reduce annual GHG emissions by 278 million tons of carbon dioxide equivalent, which equals 38% of its 2019 GHG emissions. Meanwhile, Qatar has not established a timeline for achieving net zero, and Bahrain and Kuwait have not declared their emission reduction goals. Nonetheless, all GCC countries have set up climate governance bodies.

We emphasize that the explanations for the differences in the model results, both in the short- and long-run and across countries, are based on the published reports and need further empirical investigation.

Fitness of the models

Table 10 shows the diagnostic statistics for the CO2 and GHG models. The residual metrics of root mean square error, the sum of squared residuals, and regression error are significantly lower for the GHG model, indicating that the predictive values of the GHG model are closer to the actual values. The information criteria tests—AIC, SC, and HQ—further confirm that the GHG model provides better estimates of the drivers of emissions in the GCC.

In addition to the model diagnostics, residual graphs confirm the superiority of the GHG model. Within the 95 percent confidence band, the residuals for CO2 and GHG fall within ∓.10 and ∓.02, respectively. Also, the variability of the standardized residual around 0 for the GHG model is lower than that for the CO2 model (see Fig. 2).

A Displays the actual, fitted, and residual values of the CO2 emission model, along with the 95% band of the residuals and the standardized residual distribution. B Presents the same for the GHG emission model, facilitating comparison of model fit and residual behavior.

Full size image

We examine the relationship between bank lending and environmental quality in the GCC at both regional and country levels. Our analysis considers two types of bank lending: lending to the private sector and lending to the government sector. For measuring environmental quality, we use carbon dioxide emissions and greenhouse gas emissions as proxies.

At the regional level, we find that both bank lending variables, along with all other non-lending variables, except trade openness, contribute to greenhouse gas emissions in the long run, thereby damaging environmental quality. In contrast, trade openness reduces greenhouse gas emissions. In the case of carbon dioxide emissions, two bank lending variables and GDP only have a long-term positive impact on the emissions. In the short run, none of the variables affects carbon dioxide emission, while lending to the government has a negative effect, and manufacturing value-added has a positive effect on greenhouse gas emissions.

At the country level, we observe a long-term effect of explanatory variables on both carbon dioxide and greenhouse gas emissions in all countries, except for greenhouse gases in Saudi Arabia. However, there are disparities in how explanatory variables affect two types of emissions both within and across the countries. In some cases, the explanatory variables have opposing effects on the carbon dioxide and greenhouse gas emissions within the same country. Furthermore, the variables that are neutral in their impact on carbon dioxide emissions become active contributors or inhibitors when overall greenhouse gases are modeled. This leads to the conclusion that omitting greenhouse gases other than carbon dioxide from the study results in a loss of critical information related to environmental quality.

Policy implications

The role of the banking sector in combating the threats of environmental degradation is well recognized by GCC countries. An indicator of the effectiveness of the initiatives undertaken by both governments and banks to enhance environmental quality in the region is the short-term negative effect of bank lending to the government sector on greenhouse gas emissions. This observed effect, along with the behavior of other explanatory variables in the two models, underscores the importance of holistic planning for improving environmental quality. Policies that ignore total greenhouse gas emissions and focus solely on carbon dioxide emissions are likely to be ineffective, as they risk omitting critical information about the sources of environmental degradation in the GCC region.

Regulating bank lending practices, together with incentives for green investments and clean manufacturing practices, is expected to play a pivotal role in achieving a balance between economic growth and environmental sustainability. Given the varying impact of bank lending variables and economic factors on emissions, devising country-specific strategies is imperative for effective outcomes. With respect to lending to the private sector, although initiatives have been taken, there is a need for more concerted plans and policies for green financing. Such efforts are critical because the private sector is expected to play a leading role in achieving the goal of net-zero economies in the region.

Limitations and future research

While this study offers valuable insights into the relationship between bank lending and environmental quality in the GCC region, its findings cannot be generalized globally. Although GCC countries have made headway in adopting technologies to control emissions, such as carbon capture, we did not include it as an explicit factor due to a lack of access to the relevant data. Future research could extend our study at the global level and examine the impact of technological innovation on emissions in a holistic manner, considering all components of greenhouse gases.