ESG Integration and Technical Efficiency: A Comparative Frontier Analysis in Kuwait Financial Sector

Abstract

Sustainability initiatives have gained significant attention; however, limited research has examined whether ESG factors facilitate or hinder financial sector efficiency. This research investigates the differences between the traditional stochastic frontier analysis (SFA) and the ESG-integrated SFA model in explaining inefficiency. In addition, it examines how ESG factors influence inefficiencies in a truncated regression. The sample includes 9 banks (4 Islamic and 5 commercial) and 11 financial firms—all the ESG adopters in Kuwait financial sector. Quarterly data were employed from 2018 to 2023. The findings revealed that the ESG-integrated model improves the explanatory power of the cost function, partially reducing stochastic noise in financial operations. Moreover, ESG facilitates lending consistently and reduces the marginal cost of non-interest activities. Nonetheless, capital reliance in both models is associated with higher inefficiencies. Additionally, we found that financial institutions on average operate at 33% below the best-practice technology frontier, indicating moderate gaps across the sector. Overall, strong ESG alignment is associated with improved cost-efficiency when supported by strong institutional quality.

1. Introduction

Environmental, Social and Governance (ESG) factors are increasingly recognized worldwide as key drivers of sustainable growth [1]. Since the launch of the New Kuwait Vision 2035 (NKV), Kuwait has pursued reforms to reduce the reliance on the oil and gas sectors by promoting economic diversification, enhancing human capital, and pledging 15% of its local energy from renewable sources by 2030 [2]. More recently, the Boursa Kuwait ESG Reporting Guide (2021) has encouraged listed firms to disclose their ESG activities in accordance with international best practices. A key development is the mandatory requirement for all KSE-listed firms to produce ESG reports, starting in 2026. Certainly, these reforms have a direct impact on the relationship between ESG factors and efficient operational activities in the financial sector.

Despite recent positive shifts, ESG practices have been criticized due to slow progress, weak standards, and inefficient adoption of renewable technologies [2,3]. These challenges are particularly relevant in the financial sector, which plays a pivotal role in funding sustainable projects [4]. Banks and financial firms may be reluctant to adopt ESG practices due to the costs associated with their implementation [5]. However, ESG initiatives could enhance efficiency and long-term value. Global evidence supports the adoption of integrated ESG practices as a cost-efficiency tool against random shocks and managerial inefficiencies [5,6]. Yet previous findings do not consensually declare that ESG improves operational efficiency, as its impact on the cost function is context-dependent and sector-specific [7,8].

Adopting ESG solely for compliance, without embedding sustainability into core business operations, may limit technical efficiency and exacerbate climate-related risks [1]. Such integration is essential for enhancing transparency and strengthening stakeholder confidence. Sustainability integration should be prioritized to enhance service quality and promote greater social alignment [9]. Nonetheless, integration often requires significant investment in compliance infrastructure and staff training [10]. In Kuwait, the financial sector faces structural constraints, including the prioritization of intermediation activities rather than diversification, potentially limiting ESG considerations and affecting national economic sustainability [9,11]. The high cost of ESG implementation stems from a combination of environmental pressures, regulatory uncertainty, stakeholder awareness, and underdeveloped ESG frameworks, which together intensify the challenges in aligning with sustainability standards [7]. Hence, prudent resource management and identifying the optimal stage of ESG integration are critical.

This study aims to compare the traditional and ESG-integrated cost functions in Kuwait’s financial sector, including a sample of 20 financial firms and banks (ESG adopters), using quarterly data from 2018 to 2023. Importantly, monitoring the cost vs. benefit prudently safeguards depositors’ funds and promotes ESG advancement. Our work sheds the light on the ESG framework, which aligns directly with the United Nations Sustainable Development Goals (SDGs), mainly those related to climate action (SDG 13), responsible consumption and production (SDG 12), and sustainable economic growth (SDG 8). Integrating ESG into the financial system not only supports global sustainability goals but also enhances Kuwait’s long-term stability and stakeholders’ confidence. The remainder of this article is organized as follows: Section 1 presents the introduction, Section 2 presents the background and hypothesis, Section 3 presents the methodology. Section 4 and Section 5 include the estimation analysis, discussion, and conclusion and policy recommendations.

2. Background and Hypothesis Development

Under the stakeholder theory, ESG initiatives are a potential mechanism against inefficiency, as it considers engaging in risk to satisfy stakeholders’ needs and improve financial stability [12]. Conversely, the trade-off theory suggests that ESG practices may incur higher costs that reduce efficiency and detract from firms’ value by diverting scarce resources out of investment [13]. The resource-based view moreover argue that ESG activities are a strategic tool rather than costly, which enables firms to gain competitive advantages [14]. Yet ESG is still seen as a cost rather than value-adding, pointing to a need for policies that align development with sustainability goals [15]. Studies in the GCC extensively examine efficiency across Islamic and commercial banks using both parametric and non-parametric approaches. Ref. [16] decomposed the productive efficiency of Islamic banks into technical and cost components, revealing that inefficiencies increases costs by 14% to 29%. Refs. [17,18] found that Islamic banks are more capitalized but less cost-efficient, whereas commercial banks manage costs more effectively. Similarly, Ref. [19] showed that Islamic banks have stronger short-term solvency but lower cost-efficiency. Meanwhile, Ref. [20] observed increased efficiency during the Global Financial Crisis (GFC), suggesting that capital reliance is adversely associated with bank lending, and positively with operational efficiency.

Refs. [3,21] both studied the country-level differences within the GCC; the former found that GCC peers outperformed banks in Kuwait in terms of cost-efficiency. Conversely, the latter revealed that ESG factors show a positive and conditional impact on firm performance when supported by a robust institutional quality. Ref. [22] also highlighted that ESG factors are positively associated with financial stability, while institutional factors are linked to increased risk-taking. Studies focusing on commercial banks and financial firms in the GCC—including Kuwait—also confirm moderate efficiency levels [23,24]. Using accounting measures, Ref. [8] found that banks in the GCC region are more profitability-driven rather than cost-efficient. Ref. [25] highlighted the role of bank size in determining cost-efficiency; as bank grow larger, the average cost per unit of output increase due to complexity. Collectively, these studies indicate that most attempts in Kuwait and the GCC have focused on a binary comparison, with limited focus on how operational efficiency relates to ESG practice.

Globally, Refs. [1,5] found that integrating ESG into operational processes improves cost-efficiency and enhances stakeholder confidence. Firms with stronger ESG commitments achieve higher efficiency scores, although short-term inefficiencies might arise after the initial stages of integration. Similarly, Ref. [26] reported that social and environmental activities can improve cost-efficiency, whereas governance factors are insignificant, with institutional context moderating the effects. In Asia, Ref. [27] revealed that developed countries improve their cost-efficiency through environmental activities, while social and governance practices are strong drivers of efficiency in developing regions. In contrast, larger banks can attain higher levels of efficiency due to economies of scale. Ref. [28] observed an inverse U-shaped effect between disaggregate social and governance and cost-efficiency in Europe, indicating that moderate engagement promotes operational efficiency.

H1. 

There is a negative impact of aggregate ESG factors on cost-inefficiency.

H2. 

There is a negative impact of environmental factors on cost-inefficiency.

H3. 

There is a negative impact of social factors on cost-inefficiency.

H4. 

There is a negative impact of governance factors on cost-inefficiency.

This study contributes to the literature in several ways. First, it compares two SFA models and finds that the ESG-integrated model is more effective in reducing operational inefficiencies. Second, it derives and analyzes the cost-efficiency term based on the inefficiency component. Prior studies report mixed evidence on the ESG–efficiency nexus; some highlight potential disadvantages, such as high opportunity costs and complexity [25]. Global research has also explored ESG as a determinant of bank cost-efficiency across regions—including Asia [27], Japan [6], Europe [5,28], China [29], and Italy [26]. In contrast, within the GCC region, although studies such as [18,20,30] have addressed cost-efficiency, none have examined the role of ESG in explaining cost-inefficiency. Prior research also examined institutional quality and ESG separately [3,22,31]. Hence, this research addresses these gaps by examining the interactive effect of ESG and institutional quality factors on cost-inefficiency to capture how institutional settings shape ESG factors.

3. Methodology

The sample consists of 20 financial institutions operating in Kuwait, including 9 banks (4 Islamic and 5 commercial) and 11 financial firms. Based on quarterly data extracted from 2018 to 2023, all firms and banks are classified as ESG adopters according to Refinitiv (Eikon). Firm-level frontier variables were obtained from annual reports and Eikon. Non-ESG firms were excluded to focus accurately on how ESG engagement influences the cost structure of the financial sector in Kuwait.

A firm’s efficiency is determined by its ability to convert input resources into income-generating financial assets at minimal cost [5]. Efficiency measures can be either parametric or non-parametric. Non-parametric approaches such as data envelopment analysis (DEA), although widely employed in the GCC context, are sensitive to outliers and measurement errors [1]. Moreover, non-parametric methods attribute any deviation from the frontier to inefficiency, leading to an unreliable measure that lacks random shocks [5]. To address these limitations, our analysis adopts the one-step stochastic frontier analysis (SFA) proposed by [32]. SFA is a parametric technique that decomposes deviations from the frontier into two parts—a symmetric random error term and a one-sided inefficiency term [27]. Following the intermediation approach of [5,33], the research models a translog cost function considering three input and output prices. The total cost function is expressed in the translog form (Equation (1)) with the assumption of linear homogeneity in input prices, which is imposed by normalizing all prices by the labor price P2, following [33]. Cost Function (Equation (1)):

$$\begin{array}{cc}\hfill \mathit{ln}\left({\displaystyle \frac{{TC}_{it}}{P_{2it}}}\right)& = {\alpha }_0+{\alpha }_1{\mathit{ln}Y1}_{it}+{\alpha }_2{\mathit{ln}Y2}_{it}+{\alpha }_3{\mathit{ln}Y3}_{it}+{\beta }_1\mathit{ln}\left({\displaystyle \frac{P_{1it}}{P_{2it}}}\right)+{\beta }_2\mathit{ln}\left({\displaystyle \frac{P_{3it}}{P_{2it}}}\right)+{\displaystyle \frac{1}{2}}{\alpha }_{11}{\left({\mathit{ln}Y1}_{it}\right)}^2\hfill \\ & +{\displaystyle \frac{1}{2}}{\alpha }_{22}{\left(ln{Y2}_{it}\right)}^2+{\displaystyle \frac{1}{2}}{\alpha }_{33}{\left({lnY3}_{it}\right)}^2+{\alpha }_{12}{lnY1}_{it}{lnY2}_{it}+{\alpha }_{13}{lnY1}_{it}{lnY3}_{it}+{\alpha }_{23}{lnY2}_{it}{lnY3}_{it}\hfill \\ & +{\displaystyle \frac{1}{2}} {\beta }_{11}{\left[ln\left({\displaystyle \frac{{P1}_{it}}{{P2}_{it}}}\right)\right]}^2+{\displaystyle \frac{1}{2}} {\beta }_{22}{\left[ln\left({\displaystyle \frac{{P3}_{it}}{{P2}_{it}}}\right)\right]}^2+{\beta }_{12}ln\left({\displaystyle \frac{{P1}_{it}}{{P2}_{it}}}\right)ln\left({\displaystyle \frac{{P3}_{it}}{{P2}_{it}}}\right)+{\varphi }_{11} {lnY1}_{it}\mathit{ln}\left({\displaystyle \frac{P_{1it}}{P_{2it}}}\right)\hfill \\ & +{\varphi }_{12}{lnY1}_{it}ln\left({\displaystyle \frac{{P3}_{it}}{{P2}_{it}}}\right)+{\varphi }_{21}{lnY2}_{it} ln\left({\displaystyle \frac{{P1}_{it}}{{P2}_{it}}}\right)+{\varphi }_{22}{lnY2}_{it} ln\left({\displaystyle \frac{{P3}_{it}}{{P2}_{it}}}\right)+{\varphi }_{31}{lnY3}_{it} ln\left({\displaystyle \frac{P_{1it}}{P_{2it}}}\right)\hfill \\ & +{\varphi }_{32}{lnY3}_{it} ln\left({\displaystyle \frac{{P3}_{it}}{{P2}_{it}}}\right)+{\rho }_t+v_{it}+u_{it}\hfill \end{array}$$

(1)

where ${TC}_{it}$ is the total cost of firm i at time t, P1 is the price of funding, P2 is the price for labor used for normalization, and P3 is the capital price. Y1 presents the total loan output, and Y2 and Y3 are investments and non-interest income outputs. $v_{it}$ ~ N (0,${\sigma }_v^2$) is a symmetric noise term that captures the random shock, and $u_{it}$ $\ge 0$ is one-sided inefficiency term that captures cost-inefficiency. Further, to avoid economic distortion, we have included the aggregated ESG factors in the cost function as a cost-shifting variable, not as an input—since the ESG factors are not physical inputs like labor. In parallel, we are consistent with [1,5] in the input substitution. The ESG integration model is presented as follows. Cost Function (Equation (2)):

$$\mathit{ln}\left({\displaystyle \frac{{TC}_{it}}{P_{2it}}}\right)= f\left({\mathit{ln}Y}_{it}, {\mathit{ln}P}_{it} \right)+ \gamma {ESG}_{it}+{\displaystyle \frac{1}{2}}{\gamma }_{11}{\left(ESG\right)}^2+\sum _{m=1}^3{\delta }_m{ESG}_{it}{\mathit{ln}Y}_{mit}+\sum _{k=1}^2{\theta }_k{ESG}_{it} \mathit{ln}\left({\displaystyle \frac{P_{kit}}{P_{2it}}}\right)+{\rho }_t+v_{it}+u_{it}$$

(2)

Building on the first cost function in EQ1, the integration of ESG factors appears in EQ2, where the $f\left({\mathit{ln}Y}_{it}, {\mathit{ln}P}_{it} \right)$ presents the traditional cost function, and $\gamma {ESG}_{it}$ and ${\displaystyle \frac{1}{2}}{\gamma }_{11}{\left(ESG\right)}^2$ capture the linear and non-linear effects of ESG practice on costs. The interaction terms of ESG with $\mathit{ln}Y$ and $P_{kit}$ enable this integration to modify how costs respond to the changes in Y and P. Like EQ1, we separated the random shocks $v_{it}$ from inefficiency $u_{it}$ to precisely measure how ESG factors influence the cost function. After estimating the value of $u_{it}$—the inefficiency of bank i at time t—the analysis exponentiates the following value to derive the cost-efficiency score for each firm/bank (Equation (3)):

$$C{E}_{it}=exp\left(-{û}_{it} \right)$$

(3)

SFA measures the inefficiency term in log form, and thus taking the exponential undoes the log, ensuring that the value is in natural cost-efficiency units. $C{E}_{it}$ = 1 means the firm is fully efficient and operates exactly on the cost frontier. $C{E}_{it}>1$ means the firm is inefficient and spends more than necessary to produce output. Moreover, to assess the technological disparity among financial firms and banks, the technological gap ratio (TGR) is computed following [34]. The TGR quantifies how far each firm is from the best available frontier (Equation (4)):

$${TGR}_{it}{\displaystyle \frac{{CE}_{it}^{class}}{{CE}_{it}^{group}}}$$

(4)

where ${CE}_{it}^{group}$ is the efficiency score from the overall group frontier. ${CE}_{it}^{class}$ is the cost-efficiency score from the SFA. ${TGR}_{it}$= 1 means the firm uses the best technology available in the sector, and ${TGR}_{it}$< 1 means the firm lags behind the technological frontier, indicating that there is room for technology adoption. The traditional SFA—in EQ 1 and 2—assumes that firm-specific differences (e.g., business transformation function and risk-taking) are captured inside the inefficiency term $û$. However, some of these differences—in inputs and outputs—do not change over time [35]. Hence, for robustness checks, the true fixed effects (TFE) proposed by [36] separate the fixed firm characteristics from the inefficiency term, to check if unobserved time-invariant factors change the direction of this research finding, as follows (Equation (5)):

$$\mathit{ln}\left({\displaystyle \frac{{TC}_{it}}{P_{2it}}}\right)={\alpha }_i+f\left({\mathit{ln}Y}_{it}, {\mathit{ln}P}_{it}, {L1. ESG}_{it}\right)+{\rho }_t+v_{it}-u_{it}$$

(5)

The TFE model includes firm-specific fixed effects ${\alpha }_i$ to account for time-invariant characteristics. In addition, lagged ESG scores are included to capture time-delayed effects. Since the estimation of $û$ requires a second step of regression on other covariates [36], we implemented a truncated regression following the approach of [27,37] (Equation (6)):

$$\begin{gathered} {û}_{it}={\beta }_0+{\beta }_1({ESG}_{it}\times {IQF}_{it})+{\beta }_2({ENV}_{it}\times {IQF}_{it})+{\beta }_3({SOC}_{it}\times {IQF}_{it})+{\beta }_4({GOV}_{it}\times {IQF}_{it}) \\ +{\beta }_5({SIZE}_{it}\times {IQF}_{it})+{\beta }_6({CAP}_{it}\times {IQF}_{it})+{\beta }_7{GDP}_{it}+{\beta }_8{INF}_{it}+{\epsilon }_{it} \end{gathered}$$

(6)

where ${û}_{it}$ denotes the derived inefficiency score for firm i at time t. The explanatory variables include the aggregate ESG factors and their individual components: Environmental (ENV), Social (SOC), and Governance (GOV). To identify the impact of each pillar on inefficiency, separate truncated regressions were estimated for both aggregate and disaggregated ESG scores, including interactions with institutional quality factors. Additionally, firm-level characteristics such as size and capital were interacted with institutional quality to capture their joint effects. Macroeconomic controls, including GDP growth and inflation, were also controlled to account for macroeconomic conditions. The error term ${\epsilon }_{it}$ follows a truncated normal distribution at zero, reflecting the non-negativity of ${û}_{it}$.

4. Descriptive Analysis

Table 1. Descriptive statistics.

Variable	Obs.	Mean	Std. Dev.	Min.	Max.
Frontier Variables:
Total Assets	480	5558.13	8289.04	981.0	38,010.00
Deposits	480	4157.24	6755.32	0.00	30,461.29
Equity	480	733.91	1007.48	9.37	6286.11
Investments	480	1197.58	2156.46	3.78	10,949.00
Total Costs	480	52.44	74.94	0.034	529.21
Input Funds (P1)	480	42.18	62.09	0.00	456.12
Input Labor (P2)	480	10.34	17.09	0.00	80.82
Input Capital and Others (P3)	480	7.77	17.60	0.00	151.43
Output Loans (Y1)	480	3049.02	4671.94	200.02	22,484.76
Output Investments (Y2)	480	1197.58	2156.46	378.4	10,949.00
Output Non-Interest Income (Y3)	480	12.22	20.06	0.01	143.60
Firm-Level Variables:
ESG	480	26.27	17.48	0.00	71
Environmental Factors	480	10.21	12.63	0.00	60
Social Factors	480	24.65	19.35	0.00	78
Governance Factors	480	41.85	25.60	0.00	91
Capital Ratio	480	0.35	0.29	0.07	0.95
Size	480	7.20	2.08	2.28	10.45
Institutional Quality Factors:
Accountability	480	29.89	0.85	28.50	31.37
Political Stability	480	54.68	2.04	51.89	58.77
Government Effectiveness	480	50.70	3.62	45.71	57.55
Regulatory Quality	480	59.41	2.84	54.76	63.68
Rule of Law	480	58.87	1.29	57.55	61.43
Corruption Control	480	52.90	5.55	43.81	60.38
Average IQF	480	51.07	1.98	48.03	54.09
Macroeconomic Factors
GDP Growth	480	1.77	4.22	−8.00	7.00
Inflation	480	1.23	1.23	−13.0	9.10

Note: All monetary values are expressed in Kuwaiti Dinar (KWD) thousands.

presents the descriptive statistics of the SFA, firm-level, and institutional indicator key variables. The average total cost—sum of interest plus non-interest expenses—is 52.8 (SD = 74.94). Output variable Y1 presents the total loans and has a mean of 3049.02 (SD = 4671.94); Y1 is the largest output in the cost function, signifying the banks’ strong focus on intermediation activities. Y2 investments total 11,947.58 (SD = 2156.46), while Y3 non-interest income averages 12.22 (SD = 20.06). Furthermore, the input price of funding P1 takes an average value of 42.18 (SD = 62.09), largely dependent on the level of risk-taking in the firm [21]. The staff and labor expenses (P2) average 10.34 (SD = 17.09), while P3 captures the price of capital and other administrative expenses, averaging 7.77 (SD = 17.60). Notably, the largest cost component is funding related to interest expenses, and the largest output is loans, indicating the excessive focus on intermediation activities. While the second costly input is the staff, firms are investing in their human capital to enhance workforce capabilities. Alternatively, it could reflect inefficiencies or misallocation of resources within operational activities [8].

Key determinants of firm-level control, performance, and efficiency include total assets, deposits, and equity, which have average values of 5558.13, 4157.24, and 733.91 (SD = 8289.04; 6755.32; and 1007.48). ESG performance is generally below a moderate level, with an average score of 26.27 (SD = 17.48), reflecting a limited emphasis on sustainability practices [3]. The disaggregated score of the environmental pillar is relatively low at 10.21 (SD = 12.63), the social pillar averages 24.65 (SD =19.35), while the governance pillar is the highest at 41.85 (SD = 25.60). Social activities can be costly, exceeding their benefits [4]. These variations in ESG factors suggest differing levels of emphasis in the region. The average capital ratio is relatively high at 0.35 (SD = 0.29), reflecting a combination of factors, including a reliance on capital buffers from imprudent lending, or potential government-backed interventions supporting liquidity programs—evident during the recent pandemic, as argued by [11,38]. Lastly, following [22,31], the average institutional quality factors are included as exogenous governance environmental variables. The adoption of ESG activities relies heavily on the abilities of firms and the institutional quality factors—such as accountability, government effectiveness, regulatory quality, rule of law, and control for corruption—which shape how ESG strategies are implemented [22].

Table 2. Correlation matrix.

	${\mathit{u}}_{\mathit{i}\mathit{t}}$	E	S	G	SIZE	IQF	CAP	GDP	INF
$u_{it}$	1.00
E	0.08 *	1.00
S	0.21 ***	0.71 ***	1.00
G	0.28 ***	0.59 ***	0.70 ***	1.00
SIZE	0.09	0.55 ***	0.46 ***	0.41 ***	1.00
IQF	0.11 ***	0.21 ***	0.24 ***	0.26 ***	0.04	1.00
CAP	−0.28 *	0.39 ***	0.45 ***	0.27 *	−0.51 ***	0.10 **	1.00
GDP	−0.04	0.04	0.03	0.00	0.02	0.06 *	0.02	1.00
INF	−0.05	−0.02	−0.03	−0.03	0.01	−0.24	0.01	0.57 ***	1.00

Standard errors are in parenthesis; * p < 0.10, ** p < 0.05, and *** p < 0.01.

presents the correlation matrix for key variables. Notably, firm inefficiency—measured in

Table 3. SFA half-normal.

Output y	SFA (1)	SFA (2)
Cons.	0.215 *** (0.200)	0.501 (0.424)
${\alpha }_1{\mathit{ln}Y1}_{it}$	−0.134 *** (0.0.62)	−0.297 ** (0.096)
${\alpha }_2{\mathit{ln}Y2}_{it}$	0.274 *** (0.066)	0.345 *** (0.109)
${\alpha }_3{\mathit{ln}Y3}_{it}$	0.137 ** (0.059)	0.870 *** (0.104)
ESG		−0.146 *** (0.068)
${\beta }_1\mathit{ln}\left({\displaystyle \frac{P_{1it}}{P_{2it}}}\right)$	0.713 *** (0.085)	0.009 (0.138)
${\beta }_2\mathit{ln}\left({\displaystyle \frac{P_{3it}}{P_{2it}}}\right)$	−0.190 *** (0.069)	0.334 *** (0.105)
${\displaystyle \frac{1}{2}}{\alpha }_{11}{\left({\mathit{ln}Y1}_{it}\right)}^2$	0.022 ** (0.010)	0.025 * (0.014)
${\displaystyle \frac{1}{2}}{\alpha }_{22}{\left(ln{Y2}_{it}\right)}^2$	−0.096 *** (0.013)	−0.087 *** (0.021)
${\displaystyle \frac{1}{2}}{\alpha }_{33}{\left({lnY3}_{it}\right)}^2$	0.035 ** (0.014)	0.072 *** (0.024)
ESG_lnY1		0.002 * (0.001)
ESG_lnY2		−0.001 (0.001)
ESG_lnY3		−0.005 *** (0.001)
ESG_lnP1		0.002 (0.001)
ESG_lnP3		−0.002 (0.002)
${\alpha }_{12}{lnY1}_{it}{lnY2}_{it}$	0.025 *** (0.007)	0.088 *** (0.011)
${\alpha }_{13}{lnY1}_{it}{lnY3}_{it}$	−0.036 *** (0.008)	0.026 * (0.014)
${\alpha }_{23}{lnY2}_{it}{lnY3}_{it}$	−0.010 (0.009)	−0.147 *** (0.015)
${\displaystyle \frac{1}{2}} {\beta }_{11}{\left[ln\left({\displaystyle \frac{{P1}_{it}}{{P2}_{it}}}\right)\right]}^2$	−0.204 *** (0.023)	−0.030 (0.039)
${\displaystyle \frac{1}{2}} {\beta }_{22}{\left[ln\left({\displaystyle \frac{{P3}_{it}}{{P2}_{it}}}\right)\right]}^2$	0.141 *** (0.016)	0.095 *** (0.029)
${\beta }_{12}ln\left({\displaystyle \frac{{P1}_{it}}{{P2}_{it}}}\right)ln\left({\displaystyle \frac{{P3}_{it}}{{P2}_{it}}}\right)$	0.112 *** (0.014)	−0.027 (0.025)
${\varphi }_{11} {lnY1}_{it}\mathit{ln}\left({\displaystyle \frac{P_{1it}}{P_{2it}}}\right)$	−0.011 (0.007)	−0.005 (0.001)
${\varphi }_{12}{lnY1}_{it}ln\left({\displaystyle \frac{{P3}_{it}}{{P2}_{it}}}\right)$	0.053 *** (0.009)	−0.005 (0.013)
${\varphi }_{21}{lnY2}_{it} ln\left({\displaystyle \frac{{P1}_{it}}{{P2}_{it}}}\right)$	0.046 (0.012)	−0.012 (0.020)
${\varphi }_{22}{lnY2}_{it} ln\left({\displaystyle \frac{{P3}_{it}}{{P2}_{it}}}\right)$	−0.012 (0.0011)	−0.011 (0.020)
${\varphi }_{31}{lnY3}_{it} ln\left({\displaystyle \frac{P_{1it}}{P_{2it}}}\right)$	0.086 *** (0.009)	0.035 ** (0.016)
${\varphi }_{32}{lnY3}_{it} ln\left({\displaystyle \frac{{P3}_{it}}{{P2}_{it}}}\right)$	−0.110 *** (0.011)	0.021 (0.018)
Obs.	480	480
${\sigma }_v^2$	0.290 *** (0.016)	0.188 *** (0.049)
${\sigma }_u^2$	0.179 *** (0.0.28)	0.495 *** (0.023)
λ	0.615 *** (0.040)	0.380 *** (0.066)
Log Likelihood	−163.1662	−377.26
LR chi2 (df = 6)	11.77
Prob > chi2	0.0473

Standard errors are in parenthesis; * p < 0.10, ** p < 0.05, and *** p < 0.01.

—is positively correlated with ESG components, particularly social (0.21) and governance (0.28), with a minimal increase in environmental scores (0.08). This indicates that higher ESG scores are associated with higher inefficiencies. Size exhibits moderate positive correlations with all ESG pillars (0.55, 0.46, and 0.41), while the minimal positive correlation between size and inefficiency (0.09) reflects the level of complexity that demands high monitoring costs for resource allocation and market dynamics [26]. Institutional quality factors (IQF) are positively correlated with both inefficiency and ESG components (0.11, 0.21, 0.24, and 0.26). A strong IQF encourages firms to align with sustainability practices. While the positive relationship with inefficiency may reflect short-term costs, as initial ESG adoption often involves low-complexity initiatives [15]. However, as firms progress to medium stages, short-term costs increase [5]. Capital is associated with negative and significant inefficiency (−0.08), and well-capitalized firms tend to have a better ESG performance [14]. Macroeconomic factors show weak relationships; overall, GDP and IQF exhibit a minimal positive correlation (0.06), and they have slight positive associations with environmental and social activities (0.04, 0.03). GDP growth often comes with increased regulatory attention and stakeholder expectations, prompting banks to improve ESG practices [26]. However, the correlation matrix does not provide consistent evidence, as many relationships are statistically insignificant, suggesting the need for further investigation into the drivers of firm inefficiency and ESG relationships.

4.1. Estimation Analysis and Discussion

The stochastic frontier analysis assumes that total costs are influenced by inefficiency and random shocks [28]. As presented in Table 3, the likelihood ratio (LR chi2 = 11.77, df = 6, p = 0.047) suggests that including ESG factors and their interactions significantly improves the model explanatory power. From this perspective, ${\sigma }_u^2$ measures the dispersion of inefficiency, while λ reflects the importance of inefficiency compared to random noise [32]. In the traditional cost function SFA (1), the variance of stochastic noise is ${\sigma }_v^2=0.290$, while the inefficiency variance is ${\sigma }_u^2=0.179$, yielding λ = 0.615—indicating that inefficiency explains a larger portion of total cost variation. The ESG-integrated model SFA (2) shows a decline in noise variance to ${\sigma }_v^2\left(0.188\right)$, an increase in inefficiency variance to ${\sigma }_u^2 =$0.495, and a reduction in λ (0.380). The higher ${\sigma }_u^2$ denotes greater dispersion in firm-level efficiency once ESG scores are considered, reflecting bank/firm differences in managerial effectiveness or ESG practices. Meanwhile, the reduction in λ indicates that inefficiency becomes less dominant in explaining total cost variation—as ESG factors capture part of the stochastic noise previously treated as inefficiency. Hence, ESG integration help firms operate more cost-efficiently and reduce unexplained cost variations.

The output coefficients highlight different cost implications. In SFA (1), loans $(\mathit{ln}Y1)$ show a negative coefficient (β = −0.134) and a higher loan volume reduces firms’ total costs. Investments and non-interest income outputs, $\mathit{ln}Y2 \mathrm{and} \mathit{ln}Y3$, increase total costs (β = 0.274, 0.137). In SFA (2), loans $\mathit{ln}Y1$ become more negative (β = −0.297), while both investments $\mathit{ln}Y2$ and non-interest income $\mathit{ln}Y3$ increase costs (β = 0.345, 0.870). Notably, the ESG factors effect is negative and significant (β = −0.146), indicating that integrating ESG factors contributes to lower costs and mitigates the idiosyncratic risk, consistent with [6]. The interaction terms further illustrate the effect: ESG increases the marginal cost of loans, ESG_lnY1-(β = 0.002), while it significantly reduces the cost of NOI (β = −0.005), ESG_lnY3. Adopting ESG practices helps firms avoid operational disruptions, enhance brand image, and reduce costs associated with non-interest activities [4]. Further, input prices and non-linear effects, SFA (1), mean funding prices strongly increase total costs (β = 0.713), while capital prices reduce them (β = −0.190). In a traditional cost framework, this outcome is expected, as banks (or firms) rely heavily on funding activities, whereas a strong capital base facilitates risk absorption—in line with [30]. Moreover, the positive quadratic coefficients for loans ${\alpha }_{11}$ and non-interest income ${\alpha }_{33}$ (β = 0.022, 0.035) indicate that producing high volumes of these outputs can increase marginal costs. In contrast, the negative quadratic coefficient for Y₂ investments (β = −0.096) implies no excessive cost pressure. Funding and capital interactions show compounded effects on total costs; in both models, a high capital reliance exacerbates costs, ${\beta }_{22}$ (β = 0.141, 0.095). However, the effect is less pronounced in the ESG model. In summary, SFA 1 demonstrates that simultaneous high funding and capital reliance can hamper efficiency. It is clear from ${\varphi }_{12}$ that producing loans becomes slightly more expensive when capital is costly (β = 0.053), and non-interest income ${\varphi }_{31}$ is more costly when funding costs arise (β = 0.086), while it is less costly when capital prices increase ${\varphi }_{32}$ (β = −0.110). The ESG model is the preferred model economically, as the SFA (2) contributes to cost reduction and explanatory power; SFA (1) is statistically effective in the short term, but associated with vulnerabilities—external shocks [39]—due to the absence of sustainability practice.

The heterogeneity of technology among financial firms and banks is an important issue [35]. In our case, we examine banks and financial firms operating within the same country—under one institutional umbrella. However, financial firms and banks—Islamic and commercial—differ in terms of financial performance, innovation, and cost-efficiency [38]. A common factor among all institutions in the sample is that they offer traditional financial services and share the same business transformation function—inputs and outputs. However, differences in size, and credit risk vulnerabilities exist due to the fact that commercial banks enjoy a lower beta, on average, and higher ability to provide liquidity, while Islamic banks have lower credit risk levels because they operate under PLS (“Profit Loss Sharing”, a concept that prohibits interest rates) [38]. Consequently, technological gap considerations are essential when assessing cost-efficiency, to determine how each bank/firm utilizes technology more efficiently.

To reach our goal, we follow [34,35] in measuring the TGR in our fourth step in the methodology. First, technical inefficiencies are evaluated through the SFA. Second, data on the efficient units from each frontier group—Islamic banks, commercial banks, and financial services firms—are included to convert CE into the TGR. This method enables our research to identify which groups operate cost-efficiently and possess more advanced technologies with respect to their ESG performance.

Table 4. ESG-integrated model, cost-efficiency, and technological gap.

Firm	Class	TGR	$\mathit{C}{\mathit{E}}_{\mathit{i}\mathit{t}}$	Aggregate ESG
Kuwait Investment Co. Kscp	Financial firm	0.71	0.60	14.94
Ekktitab Holding Co.	Financial firm	0.79	0.57	4.67
Egypt Kuwait Holding Co.	Financial firm	0.65	0.38	15.17
Gulf Financial House Co.	Financial firm	0.63	0.57	34.30
Inovest b.s.c	Financial firm	0.76	0.74	11.5
Kamco Investment	Financial firm	0.71	0.49	31.22
Kuwait Finance House	Islamic bank	0.57	0.53	37.00
Kuwait International Bank	Islamic bank	0.55	0.60	22.72
Ahli Bank of Kuwait	Commercial bank	0.63	0.65	46.7
Burgan Bank of Kuwait	Commercial bank	0.61	0.56	34.25
Boubyan Bank of Kuwait	Islamic bank	0.63	0.64	49.38
Commercial Bank of Kuwait	Commercial bank	0.66	0.43	31.83
Gulf Bank of Kuwait	Commercial bank	0.64	0.62	39.95
Kuwait Project Company	Financial firm	0.60	0.61	17.00
Kuwait Financial Center	Financial firm	0.72	0.57	41.00
National Bank of Kuwait	Commercial bank	0.82	0.71	56.22
Noor Investment Co.	Financial firm	0.75	0.39	6.51
Tamdeen Investment Co.	Financial firm	0.73	0.45	11.70
Um al Qaiwain General Investment	Financial firm	0.74	0.71	14.79
Warba Bank of Kuwait	Islamic bank	0.66	0.52	5.11
Total		0.67	0.56	26.29

presents the cost-efficiency (CE), technological gap ratio (TGR), and aggregate ESG scores for each firm and bank included in the sample. The average TGR across the sample is 0.67, indicating that firms operate at about 33% below the best-practice frontier. The average cost-efficiency score is 0.56—on average, firms are about 44% away from the optimal frontier. Compared to other studies that employed the traditional cost function, Ref. [40] found that the average CE level in the emerging financial sector is 0.25. Ref. [20] observed a level of 0.37, while Ref. [21] reported CE levels of 0.46 for Islamic banks and 0.49 for commercial banks.

Individually, NBK stands out with the highest TGR (0.82) and CE (0.71) and strongest ESG engagement (56.22). Similarly, Inovest and Um AlQaiwain Investment show high technological adoption (TGR = 0.76 and 0.74; CE = 0.74 and 0.71). In contrast, firms with a weak ESG performance—such as Egypt Kuwait Holding (TGR = 0.65, CE = 0.38, ESG = 15.17) and NOOR Investment (TGR = 0.75, CE = 0.39, ESG = 6.51)—exhibit higher inefficiencies. Full ESG integration often requires a developed compliance infrastructure and continuous staff training [10]. Some firms or banks might struggle to adapt due to their unadjusted priorities—short-term profit motives [9]. Therefore, they may perceive ESG practice as a cost rather than value-added, given that the benefits of ESG implementation appear in the long term [5]. Furthermore, among Islamic banks (IBs), Boubyan Bank demonstrates a balanced performance (TGR = 0.63, CE =0.64, ESG = 49.38). Similarly, KFH and Warba demonstrate moderate performances (TGR = 0.57 and 0.66; CE = 0.53 and 0.52), with ESG = 37.00 and 5.11. Overall, the heterogeneity observed across banks/firms underscores the importance of ESG factors in shaping efficiency and technological advances. Notably, higher ESG scores tend to show sustainable cost-efficiency and a stronger TGR—such as NBK, Ahli Bank, and Boubyan.

4.2. Robustness Checks

To validate the main findings based on the half-normal model, we first analyzed inefficiency estimates from SFA (1) using the truncated regression to assess whether ESG factors when interacted with the IQF reduce inefficiency. We also compared inefficiency from SFA (2) with the results obtained under the TFE analysis—the results are shown in

Table 5. Truncated regression.

	${\mathit{u}}_{\mathit{i}\mathit{t}}$ (1)	${\mathit{u}}_{\mathit{i}\mathit{t}}$ (2)	${\mathit{u}}_{\mathit{i}\mathit{t}}$ (3)	${\mathit{u}}_{\mathit{i}\mathit{t}}$ (4)
ESG × IQF	−0.514 *** (0.074)
ENV × IQF		−0.016 ** (0.002)
SOC × IQF			−0.003 * (0.001)
GOV × IQF				−0.004 (0.001)
SIZ × IQF	−0.003 *** (0.001)	−0.007 *** (0.001)	−0.002 *** (0.000)	−0.001 *** (0.000)
CAP × IQF	0.001 *** (0.000)	0.005 ** (0.001)	0.005 *** (0.001)	0.005 *** (0.001)
GDP	−0.006 * (0.004)	−0.002 (0.002)	−0.001 (0.002)	−0.001 (0.002)
INF	0.002 * (0.001)	0.000 (0.000)	0.000 (0.000)	0.000 (0.000)
Cons.	0.445 *** (0.069)	0.930 *** (0.070)	0.956 *** (0.064)	0.961 *** (0.061)
Obs.	480	480	480	480
Pseudo	911.00	822.49	824.61	817.32
$\sigma$	0.044 *** (0.007)	0.043 *** (0.005)	0.043 *** (0.005)	0.044 *** (0.005)
Wald chi2(5)	121.95	694.99	632.24	517.05
Prob > chi2	0.000	0.000	0.000	0.000

Standard errors are in parenthesis; * p < 0.10, ** p < 0.05, and *** p < 0.01.

. While the main results offer insightful findings on integrating ESG factors, they do not account for unobserved time-invariant firm-specific effects, which could bias the estimated coefficients. The TFE addresses this limitation by separating firm-specific heterogeneity from $û$, allowing for a more precise exclusion of deviations arising from the cost frontier [36,41,42]. Multicollinearity tests (VIF = 2.2) indicate no serious collinearity concerns. Further, to address potential endogeneity and reverse causality, we included an instrumental variable for ESG (L.ESG) to examine whether past ESG practices have a persistent effect on the current inefficiency.

4.2.1. Truncated Regression

Following the approach of [37], the truncated regression is commonly adopted as a second-stage procedure to assess the determinants of efficiency scores. Unlike OLS, which is inappropriate in this context due to the non-negative nature of the inefficiency score, truncated regression is suitable as it accounts for the censored nature of the dependent variable $u_{it}$, which cannot be negative. The SFA captures inefficiency exogenously [5]; it does not fully account for endogenous firm-level decisions. To overcome this, the analysis highlighted this issue in examining the inefficiency with ESG and institutional quality factor interactions. The interaction between aggregate ESG and IQF shows a negative and significant coefficient (β = −0.514), confirming H1. This suggests that higher ESG engagement, supported by robust institutional quality, reduces cost-inefficiency and financial constraints, consistent with [3,29]. Environmental and IQF interactions exhibit a negative and significant impact on inefficiency (β = −0.016), supporting H2. Environmental activities such as energy savings measures can yield cost benefits—meanwhile, they are associated with high risk in the medium stages of adoption [28]. However, in the context of Kuwait, where environmental practice is still emerging, the benefits may be limited by land scarcity and costly environmental innovation activities such as green building. Despite these barriers, when committing firms to be in line with the KSE and carbon reduction strategies (NKV), outcomes may appear eventually.

The Social× IQF coefficient is negative and significant (β = −0.003), confirming H3. Social responsibility activities contribute to lowering inefficiency, particularly when supported by institutional quality. As stakeholders value social initiatives that reflect the firms’ responsibility toward employee welfare and community engagement, they ultimately enhance customer and employee trust [43]. Moreover, the governance × IQF interaction is negative but insignificant (β = −0.016), providing partial support for H4. Governance practices do not have a statistically robust effect in the current sample. These results are not in line with the conclusions of [22,44], where the former argued that IQF impact on financial performance is negative, and the latter revealed that social activities are costly, exceeding the benefit and thus eroding efficiency.

Firm and bank size exhibit a highly significant impact across all models, indicating that larger institutions operating under strongly regulated environment are more efficient—due to economies of scale that enhance their capacity to integrate ESG factors [3,5]. In contrast, higher capital reliance is associated positively with operational disruption—especially when firms tend to receive subsidies that support their intermediation activities, leading them to neglect sustainability initiatives [11]. Moreover, the macroeconomic factors, GDP and inflation, show significant impacts only in the aggregate model. Specifically, higher inflation is associated with greater inefficiency, while GDP growth reduces average inefficiency by −0.6%. However, their significance disappears in the disaggregated models, suggesting a limited effect of macroeconomic effects once IQF interactions are considered. Overall, the second-stage findings confirm that both the aggregated and disaggregated ESG performance help in reducing firm inefficiency.

4.2.2. True Fixed Effects (TFE)

Table 6. True fixed-effects SFA.

Output y	TFE
Cons.	−3.041 *** (0.155)
${\alpha }_1{\mathit{ln}Y1}_{it}$	−0.128 * (0.071)
${\alpha }_2{\mathit{ln}Y2}_{it}$	0.241 (0.176)
${\alpha }_3{\mathit{ln}Y3}_{it}$	−0.020 * (0.007)
L1. ESG	−0.003 *** (0.001)
${\beta }_1\mathit{ln}\left({\displaystyle \frac{P_{1it}}{P_{2it}}}\right)$	0.920 *** (0.085)
${\beta }_2\mathit{ln}\left({\displaystyle \frac{P_{3it}}{P_{2it}}}\right)$	0.061 * (0.044)
${\displaystyle \frac{1}{2}}{\alpha }_{11}{\left({\mathit{ln}Y1}_{it}\right)}^2$	−0.023 * (0.016)
${\displaystyle \frac{1}{2}}{\alpha }_{22}{\left(ln{Y2}_{it}\right)}^2$	0.056 * (0.038)
${\displaystyle \frac{1}{2}}{\alpha }_{33}{\left({lnY3}_{it}\right)}^2$	0.037 *** (0.015)
ESG_lnY1	0.002 *** (0.000)
ESG_lnY2	−0.001 * (0.000)
ESG_lnY3	−0.005 * (0.001)
ESG_lnP1	0.003 *** (0.001)
ESG_lnP3	−0.002 *** (0.001)
${\alpha }_{12}{lnY1}_{it}{lnY2}_{it}$	−0.038 *** (0.012)
${\alpha }_{13}{lnY1}_{it}{lnY3}_{it}$	−0.043 *** (0.015)
${\alpha }_{23}{lnY2}_{it}{lnY3}_{it}$	0.037 *** (0.014)
${\displaystyle \frac{1}{2}} {\beta }_{11}{\left[ln\left({\displaystyle \frac{{P1}_{it}}{{P2}_{it}}}\right)\right]}^2$	−0.172 *** (0.020)
${\displaystyle \frac{1}{2}} {\beta }_{22}{\left[ln\left({\displaystyle \frac{{P3}_{it}}{{P2}_{it}}}\right)\right]}^2$	0.134 *** (0.015)
${\beta }_{12}ln\left({\displaystyle \frac{{P1}_{it}}{{P2}_{it}}}\right)ln\left({\displaystyle \frac{{P3}_{it}}{{P2}_{it}}}\right)$	0.041 *** (0.010)
${\varphi }_{11} {lnY1}_{it}\mathit{ln}\left({\displaystyle \frac{P_{1it}}{P_{2it}}}\right)$	−0.045 *** (0.007)
${\varphi }_{12}{lnY1}_{it}ln\left({\displaystyle \frac{{P3}_{it}}{{P2}_{it}}}\right)$	0.061 *** (0.008)
${\varphi }_{21}{lnY2}_{it} ln\left({\displaystyle \frac{{P1}_{it}}{{P2}_{it}}}\right)$	−0.021 * (0.011)
${\varphi }_{22}{lnY2}_{it} ln\left({\displaystyle \frac{{P3}_{it}}{{P2}_{it}}}\right)$	0.019 * (0.010)
${\varphi }_{31}{lnY3}_{it} ln\left({\displaystyle \frac{P_{1it}}{P_{2it}}}\right)$	0.113 *** (0.011)
${\varphi }_{32}{lnY3}_{it} ln\left({\displaystyle \frac{{P3}_{it}}{{P2}_{it}}}\right)$	−0.112 *** (0.009)
Obs.	480
${\sigma }_v^2$	0.218 *** (0.016)
${\sigma }_u^2$	0.140 *** (0.010)
λ	1.550 *** (0.024)

Standard errors are in parenthesis; * p < 0.10, and *** p < 0.01.

shows that both stochastic noise and inefficiency variances (${\sigma }_v^2$ = 0.218, ${\sigma }_u^2$ = 0.140) are smaller compared to SFA (2), while the inefficiency ratio (λ) increased to 1.55. After controlling for firm fixed effects, inefficiency dominates random shocks more clearly, signifying that part of the variations previously attributed to inefficiency may reflect unobserved firm-specific characteristics. In terms of outputs, the TFE estimates show that a higher scale of lending supports the main findings; however, efficiency gains are lower (TFE = −0.128, SFA2 = −0.297). Investment activities (TFE = 0.241; SFA2 = 0.345) exhibit a slightly smaller magnitude and become insignificant once accounting for heterogeneity. The reversal in NOI effects (TFE = −0.020, SFA2 = 0.870) indicates that while NOI increases inefficiency in the short term (SFA (2)), controlling unobserved time-invariant factors reveals its persistent long-term inefficiency reduction. Complex ESG activities can temporarily increase inefficiency, consistent with [26]. The lagged ESG value complements our main findings; past ESG implementation partially reduced current inefficiency (TFE = −0.003, SFA2 = −0.146), the past ESG effect is smaller than the contemporaneous effect in SFA (2). Moreover, the non-linear effect of investment is positive and significant (TFE = 0.056, SFA2 = −0.087), expanding investment output under ESG integration demands a well-developed framework [10]. This reversion is due to the fact that TFE corrects for omitted time-invariant factors [45].

The ESG–loans interaction, ESG_lnY1 (TFE = 0.002), shows a positive and significant effect—weak ESG performance reduces lending more sharply [43]. In Kuwait’s financial sector, early stages of ESG implementation are associated with weaker lending and an inefficient allocation of funds—intermediation activities are often prioritized over ESG projects [9], potentially undermining prudent risk management and efficiency [4]. Conversely, negative coefficients for ESG interactions with investment and NOI (TFE = −0.001, −0.005) suggest that ESG adoption promotes efficient investment allocation and reduces the costs associated with fee-based activities [46]. Interactions with funding and capital prices confirm that controlling for heterogeneity reduces inefficiencies and lowers the capital-intensive reliance.

The price elasticity in TFE indicates that funding costs are more influential than capital: if funding prices increase by 1%, total costs rise by 0.92% (0.92), while if capital prices (P3) increase by 1%, total costs rise by only 0.06% (B2 = 0.06)—consistent with SFA (2). Notably, it is clear that capital and funding costs are one of the main drivers of cost-inefficiency, a potential sign of moral hazard behavior and opaque lending. These differences indicate that controlling for unobserved heterogeneity refines the interpretation of how outputs and input prices drive costs. The interaction terms of ESG with outputs and inputs also demonstrate smaller magnitudes but follow the same directional effects. Overall, TFE serves as a robustness check that complements SFA (2), and in both models, the findings reveal that ESG factors reduce cost-inefficiency even after accounting for FE.

5. Conclusions

This research compared two stochastic frontier models to assess how ESG influences cost structures in Kuwait’s financial sector using a sample consisting of 9 banks and 11 ESG-adopting firms listed on the KSE with quarterly data from 2018 to 2023. The first model, SFA (1), is the traditional cost function, while the second, SFA (2), is an ESG-integrated model. This comparison reveals significant differences that will enrich the Kuwaiti sustainable finance literature.

The findings indicate that ESG practice improves the explanatory power of the cost function, partially reducing stochastic noise in financial operations—in line with H1. At the output level, ESG integration reduces the reliance on capital and lending, reflecting its role in enhancing operational resilience and risk mitigation. The interaction findings further suggest that ESG reduces the marginal cost of non-interest activities—ESG factors support non-intermediation activities and diversify risk concentration. Nonetheless, the persistence of relatively high inefficiency variance signals structural and managerial challenges in fully embedding ESG practices into financial operations. The truncated regression findings show that aggregated ESG factors, when combined with a strong institutional quality, enhance firm efficiency. Environmental and social factors contribute to cost reduction, although their implementation may face barriers due to Kuwait’s developing environmental infrastructure and the long-term cost of social activities. Governance factors receive partial support—although statistically insignificant, strong governance mechanisms appear to enhance inefficiency mitigation. These results primarily reflect the early to mid-stages of ESG integration in Kuwait, which are characterized by low implementation complexity. Additionally, we find that financial institutions on average operate at 33% below the best-practice technology frontier, indicating moderate gaps across the sector. This research certainly has limitations overall the study offers a niche focus on Kuwait financial sector; it doesn’t consider the GCC countries where they share the same financial sector characteristics. Expanding the sample would be highly enriching for the GCC and MENA studies. Additionally, considering technological gaps through meta-frontier analysis and isolating the effect for each ownership profile in a separate SFA would enhance the results; along with integrating ESG factors into profit-efficiency is also an advantage for future research.

Policy Recommendations

Given Kuwait’s institutional and financial landscape, the financial sector is highly consolidated, with a large market share held by a few banks [46]. Many of these banks or firms benefit from implicit government backing under a “too-big-to-fail” framework, which increases the moral hazard. In parallel, underdeveloped regulatory frameworks exacerbate informational asymmetry, while low stakeholder awareness limits effective ESG progress [3]. Consequently, ESG factors may be pursued for compliance rather than as a value-adding commitment. Across all empirical models, a high reliance on capital and traditional activities is observed, and both contribute to operational inefficiency. Sound sustainable practice involves offering incentives for ESG adopters. Since both aggregated and disaggregated ESG components were found to reduce costs, it is important to promote public–private partnerships (e.g., green IPO) to strengthen environmental projects, enhance transparency and accountability [27], and alleviate financial constraints [29]. Since bail-in mechanisms are uncommon in the region, effective ESG integration could serve as a long-term stability tool rather than costly capital. When carefully implemented, ESG initiatives have the potential not only to improve financial-sector efficiency but also to promote national well-being and support the objectives of Kuwait’s Vision 2035.