Disclosure of Sustainability Practices in Annual Reports and the Funding Cost of Cooperative Financial Organizations

Abstract

This study aimed to analyze the level of disclosure of information representing sustainability practices from the stakeholders’ perspective and its relationship with the funding cost of cooperative financial organizations. The level of disclosure was measured using 46 information indicators representing sustainability practices from the stakeholders’ perspective, identified in the annual reports of cooperative financial organizations (CFOs) listed in the World Cooperative Monitor 2023, totaling 155 observations. The relationship between disclosure and the cost of financing was analyzed using a random effects estimator with cluster-robust standard errors. The results demonstrate a negative relationship between the disclosure of sustainability practices and the funding cost. When disaggregated by sustainability pillar, the results show that disclosure in the social, environmental, and cultural pillars is negatively associated with funding cost, while the economic pillar shows no statistically significant effect. This suggests that disclosing sustainability-related information from the stakeholders’ perspective reduces the cost of funding and enhances the legitimacy of CFO managers, setting them apart from traditional banks. This study examines the relationship between sustainability disclosure and funding cost in CFOs by adapting validated indicators and applying a robust econometric approach. Unlike existing literature focused on traditional banks, it empirically investigates how sustainability disclosure affects information asymmetry, funding costs, and managerial legitimacy within the cooperative financial sector.

1. Introduction

This study analyzed the level of disclosure of information representative of sustainability practices, from the perspective of stakeholders, and its relationship with the cost of financing of cooperative financial organizations (CFOs). CFOs, like corporations, are characterized by problems arising from the separation of control and ownership (Jensen & Meckling, 1976). CFOs are managed by a small group of members within a dispersed ownership structure. This configuration grants decision-making power to managers, allowing a misalignment of interests with other members and generating information asymmetry (Mercer et al., 2018; Njuguna & Mathuva, 2024).

Based on the assumption of information asymmetry between managers and associates, the problems of adverse selection and moral hazard emerge (Arrow, 1963; Akerlof, 1970). Regarding adverse selection, Akerlof (1970) highlights that it occurs whenever an individual or a group of individuals has the freedom to decide whether to enter, continue, or terminate a contract, as well as choose its associated conditions, values, or characteristics. Thus, when considering only pecuniary gains, the associate or potential associate may opt for financial products or services from large banks if they do not have information that differentiates CFOs in terms of sustainability (Macagnan & Seibert, 2021). On the other hand, moral hazard occurs when CFOs pursue their interests, such as distributing personal benefits, to the detriment of the collective objectives of the members. Such practices can lead to fraud and other irregularities, putting the integrity of the organization at risk, reducing efficiency, and, in more extreme cases, leading to its operational discontinuity (Macagnan & Seibert, 2021; Kumkit et al., 2023). These actions result in reducing residual cash flows that should benefit members.

Information disclosure is the mechanism for reducing information asymmetry. In the context of CFOs, the disclosure of sustainability practices takes on a distinct meaning, as it goes beyond mere transparency regarding financial information by incorporating elements that reflect a genuine commitment to sustainable development. Historically, sustainability has evolved from a profit-centered approach to an integrated model that incorporates environmental and social dimensions. In this context, the management of resources, emissions, and waste has become a priority, while aspects such as relationship quality, safety, and human rights have taken on a central role. This evolution has been consolidated in the triple bottom line paradigm, which seeks to balance the interests of people, the planet, and profit (Elkington, 1998; Coelho et al., 2023). Recently, the concept of sustainability has been expanded to include the cultural dimension, recognizing its dynamic and multidimensional nature. Initially linked to the social pillar, cultural sustainability has come to be considered the fourth pillar of sustainable development, highlighting the importance of values, practices, and knowledge in preserving social identity and ensuring its continuity across generations. This expansion strengthens the connection among the economic, environmental, social, and cultural dimensions, emphasizing the influence of beliefs and traditions in decision-making processes. Moreover, it underscores the need for transformative learning—an essential element in driving ecological and social change within the capitalist system—aiming to promote a sustainable balance between environmental conservation, social justice, and economic growth (Macagnan & Seibert, 2021).

Although traditional financial organizations (TFOs), such as banks, also adopt sustainability practices, they are driven by profit maximization and are not primarily focused on the communities in which they operate. CFOs reflect the principles that guide the cooperative model, which is oriented towards collective goals and the generation of socioeconomic value for the society in which they are inserted. The disclosure of sustainability practices reinforces the cooperative identity and meets the expectations of stakeholders who value initiatives with a positive impact on the economic, social, environmental, and cultural domains (Aris et al., 2018; Yakar Pritchard & Çalıyurt, 2021).

Furthermore, by demonstrating alignment with the interests of their stakeholders, CFO managers reinforce their legitimacy, as they contribute to maintaining the social contract that underpins the existence of these organizations (Suchman, 1995; O’Donovan, 2002). In other words, when managers disclose, for example, that the cooperative provides financial training for its members, funds projects aimed at reducing pollutant and greenhouse gas emissions, or promotes the appreciation of local and regional culture, they stimulate and strengthen sustainability practices (Macagnan & Seibert, 2021). This understanding aligns with Deegan’s (2002) perspective, which asserts that legitimacy does not depend solely on the actual conduct of organizations but also on how society perceives such conduct. Thus, the consistent and transparent disclosure of these practices fosters the formation of positive perceptions regarding the reasonableness of managerial behavior, legitimizing their actions before stakeholders (Suchman, 1995; O’Donovan, 2002).

As a result, CFOs strengthen their image in response to community demands, which is expected to lead to an increase in deposit values and, consequently, a reduction in their funding cost (Duguma & Han, 2018; Henock, 2019; Duguma & Han, 2021; Yitayaw, 2021). Although CFOs are not profit-driven, strong financial performance, enhanced by a lower funding cost, would contribute to the long-term continuity of their operations by the social and economic benefits they provide to the communities in which they operate (Anakpo et al., 2023).

The level of disclosure was measured based on the sustainability 46 indicators validated by members and experts. The population consisted of the 43 CFOs in the World Cooperative Monitor 2023, which ranks the 300 largest cooperatives worldwide. Of this total, 12 CFOs were excluded from the sample for not making available, on their websites, either any or all the annual reports from 31 December 2018 to 31 December 2022. Thus, the final sample comprised 31 CFOs, totaling 155 observations. A random effects estimator with cluster-robust standard errors was used.

The results indicate that, on average, the disclosure of information representing sustainability practices reaches 72%, with the social pillar standing out at 79% and the economic pillar at 70% of the indicators published. The random effects estimator with cluster-robust standard errors revealed a negative relationship between the level of sustainability disclosure and the funding cost. When the analysis is disaggregated by sustainability pillar, the results indicate that disclosure in the social, environmental, and cultural pillars is negatively associated with the funding cost, whereas the economic pillar does not exhibit a statistically significant effect. These findings suggest that higher levels of sustainability-related disclosure are associated with lower funding costs for CFOs. This contributes to the literature by providing evidence from the cooperative financial sector, highlighting how different dimensions of sustainability disclosure influence financial outcomes—an area still underexplored compared to studies on traditional financial institutions (Anakpo et al., 2023; Andrieș & Sprincean, 2023).

We concluded that disclosing sustainability practices can be a legitimacy of CFO managers. The disclosure of information across the economic, social, environmental, and cultural pillars demonstrates alignment with stakeholder expectations, reducing information asymmetry and funding costs. The negative relationship between disclosure and funding cost indicates that transparency not only validates management but also facilitates access to capital under more favorable conditions. Furthermore, adequate disclosure strengthens the financial and reputational stability of cooperatives, reinforcing their social contract and distinguishing them from traditional banks, which are driven by profit maximization. The study follows the following structure. Section 2 presents a literature review and hypothesis development. Section 3 describes the methodology employed in the study, while Section 4 outlines the respective results. Finally, Section 5 presents the concluding remarks.

2. Literature Review and Hypothesis Development

Like other cooperatives, CFOs are autonomous associations formed by individuals who voluntarily come together to meet common economic, social, and cultural needs and aspirations through a collectively owned and democratically managed structure (Jamaluddin et al., 2023). Their social role is based on the mobilization and allocation of financial resources, aiming to provide accessible credit to individuals and organizations facing barriers to the conventional financial system (Pasara et al., 2021; Arestis & Phelps, 2023). Since their emergence in the 19th century with the Schulze-Delitzsch and Raiffeisen credit societies, CFOs have developed through diverse structures and business models, adapting to the needs of the communities in which they operate (McKillop et al., 2020).

For example, in Europe, CFOs serve both members and the public, adopting two main organizational structures: networked and integrated. The networked structure consists of small autonomous cooperatives that form part of a larger network, allowing for shared liquidity and solvency protection, with decentralized and independent operations, as seen in Poland and Hungary. In contrast, the integrated structure subordinates cooperatives to a central entity, which is responsible for coordinating and standardizing strategic, operational, and control functions, as well as managing capital, liquidity, and regulatory compliance. Although local cooperatives retain some operational autonomy, most decisions are centralized. This model is adopted in countries such as Germany, Austria, Spain, Finland, France, and the Netherlands, with variations in the level of integration depending on each country’s regulatory and operational specificities (Beccalli et al., 2023).

In North America, CFOs adopt distinct models, reflecting the economic and social particularities of each country. In Canada, a highly centralized model prevails, with federations such as Desjardins serving workers, small business owners, and local communities through an integrated federative structure. In the United States, credit unions operate in a decentralized manner, providing financial services to employees of specific companies, trade union members, and local communities. Despite this decentralization, these cooperatives collaborate through Credit Union Service Organizations (CUSOs), which provide support in technology and risk management, enhancing operational efficiency and strengthening the sustainability of the cooperative model (McKillop et al., 2020).

However, considering that they are legal fictions for contractual nexus purposes, the problems arising from the separation of control and ownership are also present in CFOs, regardless of the adopted structure or implemented business model (Berle & Means, 1932; Coase, 1937; Jensen & Meckling, 1976). This situation arises because asset control is exercised by a small group of members who make up the board of directors and the executive management. On the other hand, most members, although they are the owners of the assets (principals), do not have direct control over them (Mercer et al., 2018; Njuguna & Mathuva, 2024). They bear the risks associated with managerial decisions within a dispersed ownership structure, which grants managers greater autonomy (Kumkit et al., 2023).

This context favors the emergence of information asymmetry problems when associates do not have access to the information that CFOs receive, making it difficult to monitor them. (Kyazze et al., 2020). As a result, there may be negative effects on member expansion and retention, as they are unable to assess whether managerial decisions genuinely reflect the social well-being of the local community (Macagnan & Seibert, 2021). From the perspective of legitimacy theory, this lack of transparency can undermine stakeholders’ perceptions of whether managerial actions align with economic, social, environmental, and cultural expectations, jeopardizing their legitimacy and, consequently, the trust and support of members (Suchman, 1995; Deegan, 2002; O’Donovan, 2002; Macagnan & Seibert, 2021).

Legitimacy theory posits that the disclosure of managerial decisions, particularly regarding CFOs’ sustainability practices, serves as a mechanism to mitigate information asymmetry problems (Mathuva et al., 2017). In this sense, disclosure enables members and potential members to identify the cooperative values and principles inherent in sustainability practices, which are reflected in managerial decisions. Consequently, the pecuniary and non-pecuniary benefits provided by these organizations become more evident, distinguishing them from TFOs, such as large banks (Yakar Pritchard & Çalıyurt, 2021). This recognition strengthens the legitimacy of cooperatives and their managers in society, fostering trust and loyalty among their stakeholders (Macagnan & Seibert, 2021). In turn, this facilitates resource mobilization through the collection of deposits from member savers, ultimately leading to a reduction in the funding cost of these organizations (Duguma & Han, 2018; Henock, 2019; Duguma & Han, 2021; Yitayaw, 2021).

The empirical literature highlights that CFOs direct the disclosure of their sustainability practices to their members, which demonstrates that managers’ decisions are in line with the interests of stakeholders, preserving the social contract that underpins their legitimacy (Suchman, 1995; Deegan, 2002; O’Donovan, 2002; Mathuva et al., 2017; Macagnan & Seibert, 2021). This should be reflected in financial performance by reducing CFOs’ financing costs. However, there is little research on this topic (Galletta et al., 2021; Agnese & Giacomini, 2023; Andrieș & Sprincean, 2023). Even though limited, empirical evidence suggests that greater adoption and disclosure of sustainability practices are associated with a reduction in the funding cost of traditional financial organizations (Andrieș & Sprincean, 2023). Considering this and the values and principles of cooperation, the following hypothesis was established for this study:

Hypothesis 1:

The level of disclosure of information representing sustainability practices has a negative relationship with the funding cost of cooperative financial organizations.

The next section details the methodology employed to test the hypothesis developed in this study.

3. Research Design

3.1. Econometric Model and Variables

The hypothesis formulated in this study was tested using a random effects estimator with cluster-robust standard errors, selected due to the presence of heteroskedasticity and autocorrelation, as identified by the diagnostic tests described in Section 4.3. Below, the corresponding estimator equation is presented, along with a description of the variables used.

FC_it = β₀ + β₁LDS_it + β₂DEP_it + β₃AE_it + β₄LLP_it + β₅OR_it + β₆ROA_it + β₇SPREAD_it + β₈OECD_it + β₉INF_it + u_i + ϵ_it,

(1)

where

FC_it = is the dependent variable representing the funding cost of CFOs i in period t.

LDS_it = is the variable of interest representing the level of disclosure of information related to sustainability practices of CFOs i in period t.

DEP_it = is the control variable representing the proportion of the deposit balance relative to the total assets of CFOs i in period t.

AE_it = is the control variable representing the proportion of total administrative expenses relative to the total assets of CFOs i in period t.

LLP_it = is the control variable representing the total loan loss provisions recognized by CFO i in period t.

OR_it = is the control variable representing the total operating revenue of CFO i in period t.

ROA_it = is the control variable representing the return on assets of CFOs i in period t.

SPREAD_it = is the control variable representing the spread of CFOs i in period t.

OECD_it = is the control variable indicating whether CFOs i in period t are headquartered in an OECD member country.

INF_it = is the control variable representing the inflation rate of CFO i in period t.

i = represents the CFOs.

t = represents the panel time in years.

β₀ = is the intercept of the estimator.

β₁, β₂, …, β₉ = are the coefficients that estimate the impact of the explanatory variables.

u_i = unit-specific random effect component (unobserved, time-invariant).

ϵ_it = idiosyncratic error term.

The next section provides further details regarding the variable of interest, followed by a description of the dependent and control variables.

3.1.1. Variable of Interest

The variable of interest, LDSit, represents the level of disclosure of information related to CFOs’ sustainability practices from the perspective of stakeholders. To define it, the indicators developed by Macagnan and Seibert (2021) were adopted, and, due to the generic nature of these indicators, it was necessary to adapt them to the context of CFOs. This adaptation was developed, firstly, through an analysis of annual reports from 25 CFOs.

The second stage consisted of validating the adapted indicators through consultation with 17 CFOs and 6 specialists, PhDs, and researchers in cooperatives, and both groups received the study by Macagnan and Seibert (2021). All suggested adjustments were incorporated into the indicators. In this version, the economic pillar was composed of 16 indicators, the social pillar of 15, the environmental pillar of 8, and the cultural pillar of 7, totaling 46 indicators in Appendix A.

A scale of 0, 0.5, and 1 was used to measure the level of disclosure of information related to CFOs’ sustainability practices. When the annual reports did not present the information associated with the indicator, a score of 0 was assigned. If the information was presented incompletely, with only a brief mention, a score of 0.5 was assigned. For example, if the report mentioned only that the CFO provided training to its employees during the fiscal year, a score of 0.5 was assigned. When the information was presented completely, with details and quantitative data, a score of 1 was assigned. If the report included both a detailed description of the training sessions and quantitative information, such as the number of hours of training or the number of employees trained, a score of 1 was assigned.

Therefore, the level of disclosure of information related to sustainability practices for each CFO in the sample was quantified by dividing the sum of the identified sustainability practices by the total number of indicators, as per the following formula:

$${LDS}_j={\displaystyle \frac{{\sum }_{i=1}^n{X}_{ij}}{n_j}}$$

(2)

This method of calculation is commonly used in the disclosure and governance literature, resulting in a percentage level (LDS_j), where

n_j = represents the total number of expected indicators for each cooperative financial organization.

j = represents the CFOs.

i = expresses the number of indicators.

As the indicator x_ij was disclosed, a value of 0.5 or 1 was assigned; otherwise, it received a value of 0. Finally, it is noteworthy that no distinct weights were assigned to the pillars or indicators.

In the following section, a detailed description of the dependent and control variables is provided.

3.1.2. Dependent Variable

The funding cost (FC_it) is the dependent variable that represents the financial charges on deposits and other costly funding sources of CFOs. It was calculated by the ratio of interest expenses and financial charges to the total deposits and costly liabilities. Previous studies have also used funding cost as the dependent variable (Ghosh & Ansari, 2018; Agnese & Giacomini, 2023; Andrieș & Sprincean, 2023).

3.1.3. Control Variables

The control variables employed in this study are grounded in the literature, which recognizes them as relevant determinants of the cost of funding in traditional financial organizations (TFOs). These variables include size (total operating revenue), credit risk (loan loss provisions), profitability (ROA and spread), operational efficiency (administrative expenses divided by total assets), liquidity (total deposits divided by total assets), regulatory aspects (headquartered in an OECD member country), and inflation (inflation rate). The selection of these variables is consistent with empirical evidence from recent studies on TFOs, such as those by Arnould et al. (2020), Azmi et al. (2021), Agnese and Giacomini (2023), Andrieș and Sprincean (2023), Tran et al. (2024), and Kamau and Simo-Kengne (2025). The following section provides a detailed description of each control variable used in this study.

Size

Regarding the size of TFOs, the literature presents inconclusive findings about the direction of its relationship with the cost of funding (Kamau & Simo-Kengne, 2025). From the ‘too big to fail’ perspective, larger TFOs tend to benefit from lower funding costs. This occurs because the market assumes that, in times of crisis, these organizations would receive government support to prevent systemic effects, thereby reducing the perceived default risk among creditors and, consequently, the risk premium required to fund them. On the other hand, large TFOs require significant resources to sustain their operations, which may lead to higher funding costs. Therefore, the relationship may be either positive or negative (Kusi & Opoku-Mensah, 2017). In this study, the total operating revenue of CFOs, transformed using the natural logarithm, was used as the control variable for size.

Credit Risk

Credit risk reflects the asset quality of TFOs. Accordingly, the deterioration of financial fundamentals resulting from lower asset quality may lead to an increase in the cost of funding. The funding cost of TFOs largely depends on market perceptions of counterparty credit risk, which is influenced by a range of fundamentals, including the quality of the organization’s assets (Arnould et al., 2020). In line with this view, Agnese and Giacomini (2023) found that the worse the asset quality—as measured by non-performing loans—the higher the funding cost of TFOs. In the econometric model, credit risk was controlled for using the loan loss provisions recognized by CFOs. This variable was standardized to have a mean of zero and a standard deviation of one.

Profitability

From a profitability standpoint, more profitable TFOs tend to be perceived as less risky by creditors and depositors. Consistent earnings generation signals stronger financial health, greater capacity to absorb adverse shocks, and lower dependence on external funding. This perception facilitates access to financial resources at a lower cost of funding (Andrieș & Sprincean, 2023). The findings of Tran et al. (2024) and Kamau and Simo-Kengne (2025) indicate that higher levels of profitability are associated with lower funding costs in TFOs. In the econometric model of this study, ROA and spread—both expressed as ratios—were used as control variables for profitability.

Operational Efficiency

Another relevant factor in determining the funding cost of TFOs is operational efficiency, which is typically measured by operating costs. High operating costs indicate inefficiencies in cost management, which result in higher funding costs for TFOs (Kusi & Opoku-Mensah, 2017). Azmi et al. (2021) found a positive relationship between lower operational efficiency and the funding cost of TFOs. This implies that TFOs with higher operational efficiency tend to face lower funding costs. In the case of CFOs, operational efficiency was controlled for in the econometric model using the ratio of total administrative expenses to total assets. This variable is expressed as a ratio.

Liquidity

The literature suggests that a higher proportion of deposits relative to total assets results in a lower funding cost for TFOs (Azmi et al., 2021). This indicator reflects a more stable and generally less expensive source of funding for financial organizations. Moreover, greater reliance on deposits can be interpreted as a sign of lower liquidity risk, which strengthens creditors’ and investors’ confidence in the organization’s financial stability (Arnould et al., 2020). The control variable for liquidity was defined as the ratio of total deposits to total assets of CFOs, expressed as a ratio.

Regulatory Aspects

The study by Kamau and Simo-Kengne (2025) indicates that financial regulations—such as interest rate caps and the adoption of Basel Accords—indirectly influence the funding cost of TFOs by promoting revenue and asset diversification. This diversification, in turn, helps reduce the risk perceived by the market and, consequently, the cost of funding. Based on this finding, it is reasonable to infer that CFOs headquartered in OECD member countries—whose regulatory systems are generally more robust and aligned with international standards—tend to benefit more significantly from this relationship. A more mature regulatory environment supports prudential practices and higher levels of diversification, which contributes to a lower funding cost compared to CFOs based in countries with less developed regulatory frameworks. For control purposes, a dummy variable was used, taking the value 1 when the CFO is headquartered in an OECD member country and 0 otherwise.

Inflation

Inflation is considered one of the key macroeconomic determinants of the funding cost of TFOs (Kusi & Opoku-Mensah, 2017). In inflationary environments, financial agents tend to demand higher returns as compensation for the loss of purchasing power, which translates into an increase in the funding cost of these organizations. The empirical findings of Kamau and Simo-Kengne (2025) show that the inflation variable has a positive and significant coefficient in relation to the funding costs of TFOs. In this study, which focuses on a sample of CFOs, the inflation rate was included in the econometric model in standardized form, with a mean of zero and a standard deviation of one. The control variables employed in the study are summarized in

Table 1. Control variables.

Bibliography	Representation	Description	Variable	Expected Sign
Agnese and Giacomini (2023), Andrieș and Sprincean (2023), and Kamau and Simo-Kengne (2025)	Size	Total operating revenue in natural logarithm.	OR	+/−
Arnould et al. (2020) and Agnese and Giacomini (2023)	Credit risk	Standardized total loan loss provisions.	LLP	+
Andrieș and Sprincean (2023), Tran et al. (2024), and Kamau and Simo-Kengne (2025)	Profitability	ROA and spread expressed as ratios.	ROA and SPREAD	−
Azmi et al. (2021)	Operational efficiency	Proportion of total administrative expenses relative to total assets.	AE	+
Arnould et al. (2020) and Azmi et al. (2021)	Liquidity	Proportion of the deposit balance relative to total assets.	DEP	−
Kamau and Simo-Kengne (2025)	Regulatory Aspects	Headquarters of CFOs located in an OECD member country.	OECD	−
Kamau and Simo-Kengne (2025)	Inflation	Standardized inflation rate.	INF	+

Source: Prepared by the authors.

.

The dependent (FC_it) and control variables (ROA_it, SPREAD_it, AE_it, and DEP_it) were transformed using natural logarithms to improve estimator performance by adjusting the scale of the variables, reducing variability, and linearizing potential non-linear relationships among them. In the case of the dummy variable, a value of 1 was assigned if the organization’s headquarters were located in an OECD member country; otherwise, a value of 0 was assigned.

The next section presents the population and the sample analyzed in this study.

3.2. Population and Sample

The population of this study is composed of CFOs classified in the 2023 World Cooperative Monitor, prepared by the International Cooperative Alliance (ICA) with support from the European Research Institute on Cooperative and Social Enterprises (EURICSE). The monitor ranks the 300 largest cooperatives in the world based on turnover and turnover over GDP per capita, both in US dollars. The analysis period was defined based on annual financial and non-financial reports, with a reference date between 31 December 2018 and 31 December 2022. This period was selected as it coincides with the announcement of the Davos Manifesto by the World Economic Forum, which marked a renewed global emphasis on stakeholder capitalism and sustainability—an agenda that may particularly benefit CFOs given their cooperative structure and stakeholder-oriented governance model (Novkovic, 2022). The reference date of 31 December 2023 was excluded, as not all organizations had published their annual reports by the end of the data collection.

For sampling purposes, 12 CFOs were excluded for not providing any or all the annual reports for the period in question. Thus, the final sample of the study consisted of 31 CFOs, covering 5 years, resulting in 155 observations. The next section describes the sources of evidence and the data collection and analysis techniques.

3.3. Sources of Evidence and Data Collection and Analysis Techniques

Data collection occurred in two stages. First, the annual financial and non-financial reports, with reference dates between 31 December 2018 and 31 December 2022, were obtained from the CFOs’ websites. Then, content analysis was applied, assigning scores of 0, 0.5, or 1 to each indicator assessed. The results were recorded in a spreadsheet organized by reference date, and the level of disclosure (variable of interest) was calculated by dividing the sum of the scores by the total number of indicators for each organization in each analyzed year.

In the second stage, data for the dependent and control variables were collected. Financial data were extracted from the financial statements and converted to euros, as 42% of the organizations in the sample are headquartered in Europe. The functional currencies of the remaining organizations were converted based on exchange rates from the Central Bank of Brazil. Data for the dummy variable were sourced from the OECD website. Inflation data were obtained from the International Monetary Fund (IMF) website. These data were recorded following the same procedure as the variable of interest.

With the data organized, descriptive and correlation analyses were conducted, in addition to the VIF tests for multicollinearity, Breusch–Pagan-Godfrey and White tests for heteroskedasticity, and Wooldridge test for autocorrelation in panel data. The robust Hausman test was used to choose between fixed and random effects models, with the random effects estimator with cluster-robust standard errors being selected due to the presence of heteroskedasticity and autocorrelation.

The following section presents the results obtained in this study, along with their respective analyses and discussions.

4. Results

4.1. Descriptive Analysis of the Variable of Interest

Table 2. Descriptive analysis of the variable of interest.

Max	Min	Std. Dev.	Mean	Obs	Year
0.92	0.36	0.18	0.70	31	2018
0.96	0.33	0.19	0.72	31	2019
0.92	0.38	0.18	0.71	31	2020
0.93	0.38	0.18	0.73	31	2021
0.91	0.46	0.11	0.76	31	2022
0.96	0.33	0.17	0.72	155	Total

Source: Prepared by the authors.

shows the analysis description of the variable of interest that, on average, the level of disclosure of sustainability-representative practices in CFOs is 72%. The year in which this average reached its highest value was 2022, with a level of 76%. Additionally, except for 2020, there has been a consistent improvement in this average. Aris et al. (2018) identified that CFOs in Malaysia had scores above 60% in sustainability indicators, outperforming cooperatives from other sectors.

Table 3. Descriptive analysis of the variable of interest by pillar.

Max	Min	Std. Dev.	Mean	Obs	Pillar
0.97	0.34	0.14	0.70	155	Economic
1.00	0.20	0.20	0.79	155	Social
1.00	0.00	0.36	0.69	155	Environmental
1.00	0.14	0.21	0.68	155	Cultural
0.96	0.33	0.17	0.72	155	Total

Source: Prepared by the authors.

presents the descriptive analysis of the variable regarding the level of disclosure of sustainability-representative practices in CFOs, segmented by pillar.

Table 3 reveals that, on average, CFOs tend to disclose more sustainability-representative practices associated with the social and economic pillars, with percentages of 79% and 70%, respectively. Yakar Pritchard and Çalıyurt (2021) identified that cooperatives in the financial sector disclose more indicators of economic and social performance compared to other sectors.

Table 4. Indicators with the highest and lowest levels of disclosure in the economic and social pillars.

Economic
Mean	Sum	Indicators	Ranking
0.9903	153.50	Investments	1
0.9903	153.50	Default (credit extension)	1
0.9516	147.50	Financial statements	2
0.9452	146.50	Loans and borrowings (funding)	3
0.3774	58.50	Board members’ compensation	14
0.2935	45.50	Executive compensation	15
0.2548	39.50	Member turnover	16
Social
Mean	Sum	Indicators	Ranking
0.9613	149.00	Cooperative governance structure	1
0.9258	143.50	Number of employees	2
0.9065	140.50	Social actions and campaigns	3
0.6387	99.00	Sustainable Development Goals (SDGs)	13
0.6194	96.00	Member benefits plan	14
0.5065	78.50	Financial education for members	15

Source: Prepared by the authors.

presents the 3 indicators with the highest and lowest levels of disclosure in the economic and social pillars.

As shown in Table 4, the economic indicators with the highest level of disclosure are ‘investments’ and ‘default (credit extension)’, both at 99.03%, followed by ‘financial statements’ (95.16%) and ‘loans and borrowings (funding)’ (94.52%). These results indicate that CFO managers prioritize the disclosure of information that demonstrates their commitment to fiduciary duty. The lack of perception of transparent management and controlled risks can lead members to withdraw their deposits, increasing the funding cost for cooperatives (Gómez-Biscarri et al., 2022).

On the other hand, CFOs less frequently disclose information about ‘executive compensation’, ‘board members’ compensation’, and ‘member turnover’. The lack of transparency in the remuneration of managers can lead to speculation, hinder monitoring by members, and reduce their participation in assemblies and managerial decisions (Kumkit et al., 2023). This scenario can favor the perpetuation of autocratic management and the increase in direct and indirect financial benefits for the managers (Poli, 2022; Van Rijn et al., 2023).

In the social pillar, the indicator ‘cooperative governance structure’ was the most disclosed by CFOs, with a disclosure level of 96.13%, followed by ‘number of employees’ (92.58%) and ‘social actions and campaigns’ (90.65%). These results are consistent with the empirical literature on cooperatives. Macagnan and Seibert (2021), when analyzing cooperatives from various sectors, found that the most disclosed sustainability practices on their websites include ‘social actions and campaigns’ (76.92%), ‘number of members’ (76.92%), ‘social projects’ (74.73%), ‘continuous education program’ (69.23%), ‘cooperative principles’ (65.93%), and ‘number of employees’ (64.84%).

In contrast, the indicators ‘Sustainable Development Goals (SDGs)’, ‘member benefits plan’, and ‘financial education for members’ are the least disclosed by CFOs, with disclosure levels of 63.87%, 61.94%, and 50.65%, respectively. In the latter two cases, the reports analyzed generally presented benefits and training for the community at large, without specifying whether the members were included. Despite this, even the least disclosed social indicators show high disclosure levels, indicating the managers’ effort to promote information aligned with the social mission of the cooperatives. Allen et al. (2023) highlight that the managers of these organizations actively seek to contribute to the community’s well-being, reflecting the dual mission of cooperatives to generate both economic and social impact. This commitment, rooted in the history of cooperatives, helps to legitimize the managers before the stakeholders and maintain the social contract (Suchman, 1995; Deegan, 2002; O’Donovan, 2002).

Table 5. Indicators with the highest and lowest levels of disclosure in the environmental and cultural pillars.

Environmental
Mean	Sum	Indicators	Ranking
0.7419	115.00	Environmental sustainability policy	1
0.7387	114.50	Sustainable technologies (electric vehicles, renewable energy, etc.)	2
0.7355	114.00	Reduction in pollutants and greenhouse gases	3
0.6645	103.00	Environmental investments	6
0.6484	100.50	Environmental preservation project	7
0.5581	86.50	Consumption of natural resources and waste management	8
Cultural
Mean	Sum	Indicators	Ranking
0.8774	136.00	Mission, vision, principles, and values of the cooperative	1
0.8484	131.50	Encouragement of local and regional culture	2
0.8387	130.00	Cooperative history	3
0.8129	126.00	Sponsorship of actions/activities in local and regional culture	5
0.4065	63.00	Cooperative education for members	6
0.1806	28.00	Cooperative library (physical or virtual)	7

Source: Prepared by the authors.

presents the 3 indicators with the highest and lowest levels of disclosure in the environmental and cultural pillars.

Table 5 shows that the most widely publicized indicator was ‘environmental sustainability policy’ (74.19%), followed by ‘sustainable technologies’ (73.87%) and ‘reduction in pollutants and greenhouse gases’ (73.55%). The least publicized indicators were ‘environmental investments’ (66.45%), ‘environmental preservation projects’ (64.84%), and ‘consumption of natural resources and waste management’ (55.81%). In the cultural pillar, the most publicized indicators were ‘mission, vision, principles, and values of the cooperative’ (87.74%), ‘encouragement of local and regional culture’ (84.84%), and ‘cooperative history’ (83.87%).

As with the social pillar, the reports of CFOs that did not adequately address this indicator highlighted cooperative education actions for the community, without specifying whether members were included. From the perspective of legitimacy theory, this lack of clarity can undermine the social contract between managers and stakeholders, suggesting that the cooperative does not prioritize the education and training of its members, which is inconsistent with the fifth cooperative principle (Mathuva et al., 2017). The next section presents the descriptive statistics of the dependent variable and the control variables used in the econometric model of this study.

4.2. Descriptive Analysis of the Dependent and Control Variables

Table 6. Descriptive analysis of the dependent and control variables.

Max	Min	Std. Dev.	Mean	Obs	Variable
0.3256961	0.0002627	0.0383408	0.0272043	155	FC (%)
0.9498093	0.6777741	0.0454742	0.8256758	155	FC (log)
2.99 × 1010	4.15 × 1007	7.50 × 1009	5.20 × 1009	155	OR (EUR)
24.12002	17.54014	1.582328	21.31325	155	OR (log)
3.00 × 1009	−1.36 × 1009	6.11 × 1008	3.61 × 1008	155	LLP (EUR)
4.316143	−2.822623	1	6.61 × 10−11	155	LLP (Std)
0.0625149	0.0001794	0.0085785	0.0082424	155	ROA (%)
0.8738951	0.6472962	0.0339723	0.7893331	155	ROA (log)
0.9777932	0.0320616	0.2132314	0.597525	155	SPREAD (%)
0.9989351	0.8509259	0.0233428	0.9712554	155	SPREAD (log)
0.1150692	0.000474	0.0203502	0.0220664	155	AE (%)
0.9016478	0.7018653	0.0375762	0.8293636	155	AE (log)
0.8942324	0.0085831	0.1776204	0.637649	155	DEP (%)
0.9956525	0.8118335	0.0287695	0.9773681	155	DEP (log)
72.4	−1.1	9.048594	4.795484	155	INF (%)
7.471273	−0.6515359	1	5.68 × 10−10	155	INF (Std)
1	0	0.468979	0.6774194	155	OECD

Source: Prepared by the authors.

shows the mean, standard deviation, minimum, and maximum values of the dependent and control variables.

As presented in Table 6, the average funding cost of CFOs is 2.72%, with 63.76% of this funding sourced from deposits. On average, total operating revenue amounts to EUR 5.2 billion, accompanied by a mean spread of approximately 60%, which contributes to an average return on assets (ROA) of 0.8%. Loan loss provisions average EUR 361 million, representing 0.12% of total assets. The average ratio of administrative expenses to total assets is 2.21%. The mean inflation rate across the countries included in the sample is 4.8%. Regarding the regulatory environment, 21 CFOs in the sample are headquartered in OECD member countries, as indicated by the dummy variable. The random effects estimator with cluster-robust standard errors and its corresponding results are presented next.

4.3. Random Effects Estimator Analysis and Discussion of the Hypothesis Result

Prior to the analysis using the random effects estimator with cluster-robust standard errors, the VIF test was conducted to check for the existence of multicollinearity among the explanatory variables. When analyzing the result (mean VIF = 2.31), there is no multicollinearity among the explanatory variables. Therefore, subsequently, the Breusch–Pagan–Godfrey and White heteroskedasticity tests were conducted, as well as the Wooldridge test for autocorrelation in panel data.

Table 7. Heteroskedasticity and autocorrelation tests.

Results	Null hypothesis (H0)	Tests
Prob > chi2 = 0.1054	There is no heteroskedasticity	Breusch–Pagan-Godfrey
Prob > chi2 = 0.0002	There is no heteroskedasticity	White
Prob > F = 0.0000	There is no first-order autocorrelation	Wooldridge

Source: Prepared by the authors.

presents the results of these tests.

The results indicate that the null hypothesis of the White heteroskedasticity test should be rejected at the 5% significance level. Additionally, in the Wooldridge test for autocorrelation, the null hypothesis of no first-order autocorrelation in the variables is also rejected. Therefore, the basic assumptions for applying the POLS estimator in panel data are not satisfied. Accordingly, a robust Hausman test was performed to assess the suitability of the random effects model relative to the fixed effects model. Considering that the robust Hausman test yielded a result of Prob > chi2 = 0.8958, the random effects model was selected. In this case, the analysis employed the random effects estimator with cluster-robust standard errors due to the presence of heteroskedasticity and autocorrelation, and the results are presented in

Table 8. Results from the random effects model (cluster-robust standard errors).

Random-Effects GLS Regression
Group Variable: Id
Number of obs = 155
Number of groups = 31
Obs per group:					R-squared:
min = 5					Within = 0.4871
avg = 5.0					Between = 0.8029
max = 5					Overall = 0.7715
Wald chi2(9) = 376.25
Prob > chi2 = 0.0000				corr(u_i, X) = 0 (assumed)
(Std. err. adjusted for 31 clusters in Id)
[95% Conf. Interval]	p > |z|	z	Std. Err.	Coef.	FC
−0.0425333 −0.0044004	0.016 **	−2.41	0.0097279	−0.0234668	LDS
−0.5091239 −0.0929122	0.005 *	−2.84	0.1061784	−0.3010181	DEP
−0.0004447 0.0034116	0.132	1.51	0.0009838	0.0014835	LLP
0.0034098 0.0072008	0.000 *	5.49	0.0009671	0.0053053	INF
0.2209324 0.7559847	0.000 *	3.58	0.1364954	0.4884586	AE
−10.682731 −0.2066497	0.012 **	−2.51	0.3765584	−0.9446905	SPREAD
0.0065145 0.0165495	0.000 *	4.50	0.00256	0.011532	OR
−0.1519177 0.1156135	0.790	−0.27	0.068249	−0.0181521	ROA
−0.0826619 −0.0350282	0.000 *	−4.84	0.0121517	−0.058845	OECD
0.9989691 1.916418	0.000	6.23	0.2340474	1.457694	_cons
				0.01932097	sigma_u
				0.01159677	sigma_e
	(fraction of variance due to u_i)			0.73515352	rho

* 1% significance level. ** 5% significance level. Source: Prepared by the authors.

.

Table 8 shows that the level of disclosure of sustainability practices by CFOs (LDS) is negatively associated with the funding cost (FC) at the 5% significance level. This finding aligns with the results of Agnese and Giacomini (2023) and Andrieș and Sprincean (2023), who reported that increased transparency in the disclosure of environmental, social, and governance (ESG) practices reduces the funding cost of TFOs. In addition, the variables liquidity (DEP) and regulatory aspects (OECD) also demonstrate a negative relationship with the funding cost at the 1% significance level, consistent with the findings of Azmi et al. (2021) and Kamau and Simo-Kengne (2025), respectively, in studies conducted on TFOs.

With regard to profitability, ROA does not exhibit statistical significance, diverging from the results reported by Kamau and Simo-Kengne (2025). By contrast, the SPREAD variable shows a negative association with the funding cost at the 5% significance level. These results suggest that more profitable CFOs tend to benefit from lower funding costs, consistent with the pattern observed in TFOs (Tran et al., 2024). Meanwhile, the variables inflation (INF), operational efficiency (AE), and size (OR) exhibit a positive relationship with the funding cost (FC) at the 1% significance level, corroborating the findings of previous studies on TFOs (Azmi et al., 2021; Agnese & Giacomini, 2023; Kamau & Simo-Kengne, 2025). Lastly, the credit risk (LLP) variable was not statistically significant in explaining variations in the funding cost (FC).

Subsequently, sustainability disclosure levels were estimated by pillar—economic, social, environmental, and cultural—using the random effects estimator with cluster-robust standard errors. Summary results are presented in

Table 9. Summary of results from the random effects model with cluster-robust standard errors, by sustainability pillar.

[95% Conf. Interval]	p > |z|	z	Std. Err.	Coef.	FC	Appendix
−0.0250769 0.0129425	0.532	−0.63	0.009699	−0.0060672	LDS (Economic)	Appendix B
−0.0336183 −0.0006718	0.041 **	−2.0	0.0084049	−0.0171451	LDS (Social)	Appendix C
−0.0150367 0.0012288	0.096 ***	−1.66	0.0041494	−0.006904	LDS (Environmental)	Appendix D
−0.0250348 0.0015274	0.083 ***	−1.73	0.0067762	−0.0117537	LDS (Cultural)	Appendix E

** 5% significance level. *** 10% significance level. Source: Prepared by the authors.

, with detailed outputs available in Appendix A, Appendix B, Appendix C, Appendix D and Appendix E.

Table 9 shows that the level of sustainability disclosure in the social, environmental, and cultural pillars is negatively associated with the funding cost of CFOs, with significance levels of 5%, 10%, and 10%, respectively. By contrast, the economic pillar did not exhibit statistical significance. An opposite pattern is observed in TFOs, where the governance factor has a predominant impact on the funding cost of these organizations (Agnese & Giacomini, 2023; Andrieș & Sprincean, 2023). This result signals to CFO managers that disclosing the cooperative’s social, environmental, and cultural initiatives can facilitate access to lower-cost funding through member deposits. In the context of CFOs, this means that when members perceive that the cooperative trains its employees and members, supports social projects, engages in environmental preservation initiatives, finances the acquisition of sustainable technologies, and promotes local cultural activities, among other actions, they are more willing to mobilize deposits that contribute to lowering the cooperative’s funding cost. This behavior distinguishes CFOs from TFOs, where actions aimed at profit maximization tend to carry greater importance (Galletta et al., 2021).

Considering the results, we can state that the reduction in information asymmetry, which represents sustainability practices, is statistically related to the cost of financing. Therefore, we cannot disregard the hypothesis that the disclosure of information would reduce the financing cost of CFOs, which also legitimizes the decisions adopted by their managers (Suchman, 1995; Deegan, 2002; O’Donovan, 2002). CFOs depend on deposits from their members to fulfill their role in financial and social inclusion. To mobilize these resources, members must perceive the management as aligned with the interests of the stakeholders. The disclosure, in annual reports, of practices such as social actions, training of employees and members, cultural sponsorship, and funding of environmental projects signals the cooperative’s commitment to social well-being (Macagnan & Seibert, 2021).

From the perspective of legitimacy theory, disclosure practices validate the role of management in the social contract, promoting the acceptance and trust of stakeholders, which are essential for the financial and reputational sustainability of CFOs. These practices differentiate CFOs from TFOs by balancing economic interests with social well-being and the interests of their stakeholders in a more comprehensive manner (Pedron, 2023). Furthermore, adopting sustainability disclosure practices can reduce information asymmetry, attract capital from conscious stakeholders, and enable access to green credit lines or tax incentives. By strengthening their reputation and mitigating regulatory risks, CFOs can secure better credit terms, reduce funding costs, and ensure their financial stability. The results of this study confirm the negative relationship between the level of disclosure of sustainability practices and the funding cost of these organizations.

5. Final Considerations

This study analyzed the level of disclosure of sustainability practices and its relationship with the funding cost of CFOs from the stakeholders’ perspective. To this end, the sustainability indicators of Macagnan and Seibert (2021) were adapted and validated with members and experts, covering the economic, social, environmental, and cultural pillars. On average, the level of disclosure was 72%, with higher values in the social (79%) and economic (70%) pillars, reflecting the dual mission of cooperatives to meet both economic and social objectives. The analysis used the random effects estimator with cluster-robust standard errors, identifying a negative relationship between the level of disclosure and the funding cost. Furthermore, when examining the level of sustainability disclosure by pillar, it was found that the social, environmental, and cultural pillars are negatively associated with the funding cost of CFOs, whereas the economic pillar did not exhibit statistical significance.

The results show that the disclosure of sustainability practices is crucial for strengthening the legitimacy of CFOs, considering the perspective of legitimacy theory. Disclosure in the economic, social, environmental, and cultural pillars demonstrates alignment with stakeholder expectations and reinforces their dual mission of generating both economic and social impact. Greater transparency reduces information asymmetry, contributing to the reduction in funding costs. Furthermore, proper disclosure strengthens the financial and reputational stability of cooperatives, consolidating their social contract and distinguishing them from traditional profit-driven banks.

It is important to note that this study used only sustainability practices indicators specific to CFOs, without considering other possible proxies. The analysis focused on annual financial and non-financial reports within a specific period, excluding other sources or periods. The regulatory compliance of the disclosed practices and the accuracy of the information presented were not investigated.

This study has certain limitations. First, the use of sustainability indicators adapted to the perspective of Brazilian members may restrict the generalizability of the findings to other contexts. Second, the absence of a consolidated database limited the analysis to fewer periods and organizations, thereby reducing the breadth of the results. Third, the omission of alternative proxies for the variables of interest and control may have influenced the outcomes. Additionally, potential changes in the funding dynamics of the financial system resulting from the COVID-19 pandemic were not examined.

Future studies could expand the analysis of sustainability disclosure in CFOs by adapting indicators to different cultural and regulatory contexts. It is recommended to extend the analysis period and include more cooperatives, including those from other economic sectors, for a broader comparative view. The incorporation of additional proxies, such as governance variables, operational performance, and market characteristics, could deepen the understanding of the economic and social impacts of sustainability disclosure.