Persistence of non-GAAP measures and the quality of their disclosure: Evidence from Brazilian banks

ABSTRACT

The purpose of this article is to examine the relationship between the disclosure of qualitative information and the persistence of non-GAAP (Generally Accepted Accounting Principles) quantitative measures of recurring income in Brazilian banks. Previous research has primarily focused on the effectiveness of non-GAAP quantitative measures and the composition of exclusions. However, there is a scarcity of studies addressing the quality of non-GAAP qualitative disclosures and their influence on reported financial metrics. As the debate over regulating non-GAAP disclosures grows, this study stands out by providing evidence from the Brazilian banking context, where non-GAAP quantitative measures are regulated but qualitative ones are not. The results also support arguments for regulating qualitative disclosures. This study directly benefits users of non-GAAP information and banking market regulators by showing that improving qualitative disclosures can make quantitative metrics more useful and reliable in decision-making. The earnings persistence model was used for 31 Brazilian banks (publicly traded and privately held) that disclosed quarterly non-GAAP information from 2010 to 2023. The qualitative information was transformed into an index for the empirical analysis.

Keywords:
non-GAAP quality; persistence; recurring income; banks

RESUMO

Este artigo teve como objetivo verificar a relação entre a divulgação de informações qualitativas e a persistência das medidas quantitativas não-GAAP (generally accepted accounting principles - princípios contábeis geralmente aceitos) de resultado recorrente em bancos brasileiros. Pesquisas anteriores concentram-se, majoritariamente, na efetividade das medidas quantitativas não-GAAP, como a composição das exclusões. Há escassez de estudos que abordam a qualidade das informações qualitativas não-GAAP e sua influência sobre as métricas financeiras reportadas. Com a ampliação dos debates sobre a regulação das informações não-GAAP, esta pesquisa destaca-se ao trazer evidências do contexto bancário brasileiro, no qual as medidas quantitativas não-GAAP são reguladas, mas não as qualitativas. Os resultados fortalecem os argumentos a favor da regulação também das divulgações qualitativas. A pesquisa contribui diretamente para os usuários das informações não-GAAP e para reguladores do mercado bancário, ao indicar que melhorias nas divulgações qualitativas podem elevar a utilidade e a confiabilidade das métricas quantitativas no processo decisório. Foi utilizado o modelo de persistência dos lucros de 31 bancos brasileiros (abertos e fechados) que divulgaram informações não-GAAP trimestrais no período de 2010 a 2023. As informações qualitativas foram transformadas em um índice para a análise empírica.

Palavras-chave:
qualidade não-GAAP; persistência; resultado recorrente; bancos

1. INTRODUCTION

This study examines the relationship between the disclosure of qualitative information and the persistence of non-GAAP (Generally Accepted Accounting Principles) quantitative measures in the Brazilian banking market. The underlying assumption is that if banks disclose transparent non-GAAP qualitative information, this quality will also be observed in the metric accompanying the respective disclosure. In this study, the metric is recurring income, a non-GAAP measure used by Brazilian banks derived from GAAP income that excludes transitory items (non-recurring revenues and expenses), i.e., net income adjusted for non-recurring items. To this end, the model from the literature on accounting information quality related to earnings persistence was used.

Chen et al. (2021) tested non-GAAP qualitative information in U.S. capital markets using 12 qualitative indicators set forth in Regulations G and S-K and the Compliance and Disclosure Interpretations (C&DIs). This set of guidelines, issued by the U.S. Securities and Exchange Commission (SEC, 2003) via Release No. 33-8176/2003, aims to ensure that users of financial statements have access to high-quality information for evaluating company performance. Non-GAAP quantitative measures fall within the scope of these guidelines. The authors found that these indicators convey useful information to investors and that higher-quality disclosure of non-GAAP qualitative information is associated with excluding more transitory items (i.e., higher-quality non-GAAP quantitative measures).

In this sense, the usefulness of non-GAAP disclosures - which encompass both qualitative and quantitative information - may be related to the information gap left by GAAP measures for consumers of this information. A study by Hribar et al. (2022) involving U.S. companies found that accounting standards encourage the voluntary disclosure of non-GAAP quantitative measures. When GAAP limits managers' discretion through certain requirements, managers provide more disclosures of non-GAAP quantitative measures. Conversely, when standards allow for greater discretion, fewer non-GAAP disclosures were observed.

Similarly, the growing use of non-GAAP quantitative measures in market practices highlights their informational relevance and ability to address perceived limitations in financial statements. For example, rating agencies make systematic adjustments to reported accounting profit by eliminating the effects of non-recurring items to measure the ability to generate sustainable income (Gu & Chen, 2004). Recurring income metrics also receive significant attention from analysts and managers during investor conference calls because they are considered more reliable indicators of operating performance than GAAP earnings measures (Bradshaw et al., 2018).

Ribeiro et al. (2019) compared several non-GAAP quantitative performance measures with their closest GAAP earnings measures to determine the higher-quality metrics among companies in the Australian capital market. They used earnings quality attributes already established in the literature, such as persistence, value relevance, predictive power, and smoothing, to make this determination. They found that non-GAAP performance measures are more persistent, have greater value relevance, and have greater predictive power and smoothing than the closest equivalent GAAP metric.

In this context, earnings persistence is a standard attribute of earnings quality because it reflects sustainability over time. Therefore, it is a desirable attribute of accounting information quality since persistent earnings are considered more reliable (Francis et al., 2004). Furthermore, persistent earnings provide greater predictive power regarding firms' future performance (Dechow et al., 2010).

As in the literature on earnings quality, persistence is also used as a proxy for the quality of non-GAAP quantitative measure disclosures (Bentley et al., 2018; Doyle et al., 2003; Landsman et al., 2007). However, this quality is usually evaluated solely from a quantitative perspective, with a few exceptions. For example, Chen et al. (2021) analyze qualitative aspects. Additionally, studies assessing the quality of non-GAAP disclosures are conducted in the context of non-financial entities. The gap is even wider when examining studies that investigate financial institutions due to the unique way these entities generate earnings, which differs from the real economy.

In Brazil, Sousa et al. (2021) studied banking institutions and found that their reported non-GAAP quantitative measures of recurring income are more persistent than their accounting profits presented in financial statements prepared according to local and international GAAP. Although the authors addressed persistence as a proxy for the quality of non-GAAP measures using a methodology similar to Ribeiro et al. (2019) , they did not address the qualitative aspects of non-GAAP quantitative measure disclosures, as is common in the literature.

According to the Basis for Conclusions (BC) of the International Financial Reporting Standards Foundation (2019), the disclosure of this non-GAAP qualitative information is important to users of financial statements for explaining non-GAAP quantitative measures. IFRS 18, this standard, sets forth certain non-GAAP disclosure requirements for management performance measures. Investors requested these requirements from the International Accounting Standards Board for entities that disclose non-GAAP quantitative measures. This is a voluntary practice in most jurisdictions.

The Brazilian banking sector is considered strictly regulated; however, the regulation is solely directed at non-GAAP quantitative disclosures. The Central Bank of Brazil (BCB, 2020) governs these specific requirements through BCB Resolution No. 2/2020. The resolution requires financial institutions to disclose the non-GAAP quantitative measure of recurring income but does not establish requirements for disclosing non-GAAP qualitative information, which is considered voluntary.

Assuming users deem qualitative information important for understanding non-GAAP quantitative measures, the main findings of this study demonstrate the average influence of this type of qualitative information on the quality of disclosed non-GAAP quantitative measures, from the perspective of persistence in non-GAAP quantitative measures of recurring income. Furthermore, at the extremes of the sample distribution - higher (lower) levels of non-GAAP qualitative information disclosure - there is evidence of greater (less) persistence in the non-GAAP quantitative measure of recurring income. Thus, managers potentially have different incentives to disclose more or less transparent information, which falls outside the scope of this study.

This study addresses a gap in Brazilian research on non-GAAP qualitative disclosures in banking institutions, which are usually excluded from studies. The study contributes to the literature on voluntary disclosure by testing a qualitative index of voluntary disclosures in Brazil and to the literature on non-GAAP disclosures by examining how this qualitative index impacts the quality of a non-GAAP quantitative metric.

Additionally, the study has the potential to contribute to local and global discussions about non-GAAP quantitative measures among users and regulatory agencies. It could encourage jurisdictions to adopt practices that improve the quality of quantitative measures and qualitative information. Since the study presents findings on the Brazilian banking market, regulatory agencies such as the BCB may be encouraged to adopt these principles or support the development of guidelines for disclosing non-GAAP qualitative information.

Regarding external users, notably analysts and investors, this study may have a broader impact because they are primary stakeholders in non-GAAP information. Evidence that non-GAAP qualitative information influences the persistence of non-GAAP quantitative measures contributes to increased transparency and reduced information asymmetry in the Brazilian banking market.

2. LITERATURE REVIEW AND HYPOTHESIS

According to the BC of IFRS 18, voluntary disclosures encompass a broad range of practices, including non-GAAP quantitative measures. These measures are frequently used by financial statement users to assess the sustainability of an entity's performance. These metrics tend to persist over time and provide more accurate predictions of cash flows (Dechow et al., 1999; Fairfield et al., 1996). Within the Brazilian banking sector, the BCB recognized the frequent use of the non-GAAP quantitative measure of recurring income by banks that voluntarily disclosed this measure in their management reports until 2020. As of BCB Resolution No. 2/2020, disclosure of this measure became mandatory in the notes to financial statements.

Publicly traded banks must comply with the requirements of the Brazilian Securities and Exchange Commission (CVM, 2022). However, Resolution No. 156/2022, which establishes compliance criteria for disclosing the non-GAAP quantitative measure of earnings before interest, taxes, depreciation, and amortization, does not apply to the banking industry because this measure is incompatible with the financial intermediation activities of these entities.

The increase in the number of entities reporting non-GAAP measures, coupled with the rise in the number of items disclosed and the lack of regulatory standards to improve the disclosure of these measures, has prompted researchers to conduct studies on the quality of non-GAAP quantitative measures. However, investors consider these measures to be of little use when they are disclosed without explanations. In accordance with the BC of IFRS 18, few studies have investigated the qualitative information that accompanies these measures. Discussions surrounding the disclosure of non-GAAP qualitative information are recent and limited to U.S. capital markets. For example, Chen et al. (2021) conducted a study on this topic.

They developed an index of qualitative characteristics of non-GAAP disclosures indicating the transparency of qualitative information inherent in non-GAAP quantitative measures. These disclosure characteristics are set forth in the SEC's regulatory framework (Regulations G, S-K, and C&DIs), which provides guidelines for voluntary non-GAAP disclosure should companies opt for this type of reporting. The study concludes that the index of qualitative disclosures provides informational value to users and that entities with higher-quality disclosures provide higher-quality non-GAAP earnings, more effectively excluding transitory items.

Previous empirical studies have examined the relationship between exclusions in non-GAAP quantitative measures - whether recurring or non-recurring income - and companies' future performance. This is one of the most common ways to evaluate the quality of non-GAAP reporting since managers and analysts claim that excluded items do not reflect entities' ordinary operations (Black et al., 2018). This research approach is typically referred to in previous studies as the persistence of exclusions.

Ribeiro et al. (2019) contribute to this debate by showing that the non-GAAP quantitative measures used by Australian firms are more persistent and have greater value relevance, predictive power, and income smoothing than their GAAP equivalents. This finding aligns with market practices, including those observed in the Brazilian market. For instance, rating agencies routinely adjust GAAP earnings to eliminate non-recurring items, aiming to assess more permanent earnings (Gu & Chen, 2004). Furthermore, analysts and investors place greater emphasis on non-GAAP quantitative measures than on GAAP earnings in their interactions with management during conference calls and institutional earnings presentations (Bradshaw et al., 2018).

In the Brazilian context, Sousa et al. (2021) corroborate previous studies by testing the persistence of non-GAAP measures of recurring income in banks. This is in line with persistence models from accounting information quality literature and the findings of Ribeiro et al. (2019) . They found that the non-GAAP quantitative measures of recurring income used by Brazilian banks are more persistent than the accounting profits reported in their financial statements, which are prepared according to local and international GAAP.

However, Doyle et al. (2003) assert that although non-GAAP earnings correspond to persistent profit measures, items excluded from these metrics are not necessarily transitory. Additionally, Choi et al. (2007) discuss differences in the persistence of non-GAAP earnings components disclosed by management and analysts. They argue that if non-GAAP earnings were adjusted only for truly transitory items, substantial disparities between management's and analysts' adjustments would not make sense.

The study by Choi et al. (2007) raises a key question regarding the quality of what is disclosed and the possible effects of such opacity on non-GAAP quantitative measures. The study points out that the disparities are concentrated in operating items and discontinued operations and that disclosures with informational opacity create confusion regarding the persistence of certain exclusions considered in the calculation of non-GAAP quantitative measures.

In this regard, voluntarily disclosing qualitative information about non-GAAP quantitative measures can reduce information asymmetry between managers and investors. By detailing the criteria used for adjustments and exclusions of specific items, companies provide greater clarity on the nature of these components. This allows market participants to distinguish more precisely between recurring and transitory items. This level of transparency enables a better assessment of the persistence of non-GAAP quantitative measures because it mitigates the uncertainties related to informational opacity identified by Choi et al. (2007) . Therefore, clearer qualitative disclosure of these metrics results in smaller discrepancies in interpretation between managers and analysts, thereby increasing the perceived quality of the disclosed non-GAAP quantitative measure.

Additionally, from a signaling theory perspective, improved qualitative disclosure of non-GAAP quantitative measures can be interpreted as a positive signal of a firm's informational quality. Firms can convey a message of trust and credibility regarding their financial figures to the market by opting for more transparent and detailed practices in communicating these results. This aligns expectations and strengthens the assessment of the persistence of these metrics. Studies such as those by Chen et al. (2021) support this idea, finding that high levels of qualitative disclosure are associated with higher-quality non-GAAP quantitative measures. Thus, voluntary qualitative disclosure can strengthen investor confidence, reduce asymmetries, and improve the interpretation of reported earnings sustainability, supporting the following research hypothesis:

H₁: The persistence of the non-GAAP quantitative measure of Brazilian banks is directly associated with the quality of the disclosed information, as measured by the non-GAAP qualitative index (NGQI).

3. METHODOLOGICAL PROCEDURES

3.1 Sample, Time Period, and Data

The research sample consists of Brazilian banks whose shares are traded on B3 S.A. - Brasil, Bolsa, Balcão (B3) and/or that have American Depositary Receipts (ADRs) traded on the New York Stock Exchange. The sample also includes banks classified in segments 1 (S1), 2 (S2), and 3 (S3) of National Monetary Council (CMN) Resolution No. 4,553 (2017). This resolution classifies institutions in the respective segments if their size exceeds 0.1% of Brazil's gross domestic product (GDP). Entities in segments 4 (S4) and 5 (S5) were included in the sample only if they were listed on B3 due to an inability to access the necessary data to conduct this study. Banks listed and/or belonging to segments S1, S2, and S3 are expected to report non-GAAP metrics more frequently than entities not included in these segments.

In addition, the banks must have disclosed non-GAAP quantitative measures during the study period, spanning data from the first quarter of 2010 to the fourth quarter of 2023. This period was selected based on the availability of data on the banks' institutional websites. Although some institutions had reports available for periods prior to the study, the number of banks that provided this information was negligible. Therefore, the first cutoff point was set as the first quarter of 2010. The fourth quarter of 2023 is the end of the study period because it was the latest quarter available at the time of data collection.

Table 1 shows the initial and final composition of the sample and the list of banks included in the study, totaling 519 observations (bank/quarter).

Table 1
Composition of the bank sample

Panel A: Sample definition
Description			n
Members of S1			6
Members of S2			6
Members of S3			56
Listed on B3 S.A. - Brasil, Bolsa, Balcão and not members of S1, S2, or S3			5
(=) Initial sample			73
(-) Data unavailable for model estimates			42
(=) Final sample			31
Observations			519
Panel B: Banks included in the sample
ABC-BRASIL	BMG	Citibank	Paraná Banco
Amazônia	BNDESPAR	CSF	Pine
Banco do Nordeste	Bocom	Daycoval	Safra
Banese	Bradesco	Itaú	Santander
Banpará	BRB	Mercantil	Sofisa
Banrisul	BTG	Nubank	Votorantim
BB	BS2	Original	XP
BDMG	Caixa Econômica Federal	Pan

• Source: Prepared by the authors.

The data necessary for the study were obtained from the Refinitiv Eikon database, the BCB website, and the banks' institutional websites. Information from the institutional websites was manually extracted from performance reports and explanatory notes. In the latter case, only notes to financial statements covering the period after BCB Resolution No. 2/2020 was published were consulted. Although the resolution took effect in 2021, some banks began disclosing non-GAAP quantitative measures of recurring income in their notes to financial statements as early as the third quarter of 2020. Therefore, these reports were only consulted in cases where the non-GAAP quantitative measure of recurring income was not disclosed in the banks' performance reports.

3.2 Econometric Models

To test the research hypothesis, the earnings persistence model (Equation 1) adapted from the earnings quality literature (Dechow et al., 2010) was estimated using ordinary least squares.

$$R{I}_{i,t+1}={\alpha }_0+{\alpha }_1{RI}_{i,t}+{\alpha }_2{NGQI}_{i,t}+{\alpha }_{3}R{I}_{i,t}*{NGQI}_{i,t}+{\alpha }_4{Size}_{i,t}+{\alpha }_5{Grow}_{i,t}+{\alpha }_6{List}_{i,t}+{\alpha }_7{∆GDP}_t+u_{i,t+1}$$ (1)

where RI_i,t+1 is the recurring income of bank i in quarter t+1, weighted by the average total assets between quarters t+1 and t; RI_i,t is the recurring income of bank i in quarter t - a proxy for persistence - weighted by the average total assets between quarters t and t-1; NGQI_i,t is the NGQI of bank i in quarter t; Size_i,t is the size of bank i in quarter t; Grow_i,t is the growth of bank i in quarter t; List_i,t is a dummy variable where 1 indicates whether bank i is listed on B3 or has ADRs in quarter t, and 0 otherwise; ΔGDP_t is the change in current GDP values in quarter t; and u_i,t+1 are the regression residuals for bank i in quarter t+1.

To test hypothesis H1, which states that the persistence of the non-GAAP quantitative measure is directly associated with the NGQI, the interaction term RI * NGQI is expected to be positive. This is because higher levels of transparency in non-GAAP qualitative information imply higher-quality disclosure, which may result in greater predictability of the non-GAAP quantitative measure of future recurring income (RR_i,t+1 ). Thus, α₃ is expected to be positive, significant, and greater than α₁ .

The NGQI variable was measured using a collection of qualitative non-GAAP data that included items from an index developed by Chen et al. (2021) based on Regulations G, S-K, and C&DIs, as well as items applicable to the Brazilian banking industry derived from BCB Resolution No. 2/2020, CVM Resolution No. 156/2022, and IFRS 18. Following Bardin (2016) , a content analysis was conducted on the reports used in the study to answer questions regarding 11 items of non-GAAP qualitative information. These items were grouped into eight categories and incorporated into the NGQI. Table 2 presents the 11 items, the responses, the justifications for each response, and the source of each question.

Table 2
Items Composing the NGQI

Category	Item	Adapted Question	Answers	Justification for the answers	Source of the question
Labeling	1	Is the non-GAAP quantitative measure labeled as such or as a management performance measure?	Yes	The measure is explicitly labeled as either non-GAAP or a management performance measure.	Regulations G, S-K, and C&DIs (Chen et al., 2021) and IFRS 18
			Partial	The bank does not use labels for non-GAAP or management performance measures. However, including these measures in the net income discussion section implicitly suggests that they are non-GAAP or management performance measures.
			No	The non-GAAP quantitative measure is not labeled as such.
	2	Is the GAAP measure to be adjusted labeled as such?	Yes	The measure is labeled as GAAP, or the narrative indicates that it is derived from the financial statements.	Regulations G, S-K, and C&DIs (Chen et al., 2021)
			Partial	Labeling the metric as GAAP is implicit through the presentation of GAAP results before disclosing the non-GAAP quantitative measure or through the precise use of official financial statement line items.
			No	The GAAP measure to be adjusted is not labeled as such.
Discussion of usefulness	3	Does the bank explain why the disclosed non-GAAP quantitative measure or the adjustments (excluded items) provide useful information about its financial performance?	Yes	The bank provides a clear and detailed discussion of why the non-GAAP quantitative measure and the adjustments (excluded items) are useful for understanding its financial performance.	IFRS 18
			Partial	While the bank does provide a discussion of the usefulness of the non-GAAP quantitative measure or the adjustments made, the explanation is generic and lacks specific details regarding its relationship to financial performance.
			No	The bank does not discuss the usefulness of the non-GAAP quantitative measure or the adjustments made (excluded items).
Explanations	4	Does the bank explain the composition of the non-GAAP quantitative measure or the adjustments made (excluded items)?	Yes	The bank clearly explains the composition of the non-GAAP quantitative measure or the adjustments made (excluded items).	CVM Resolution No. 156/2022 and IFRS 18
			Partial	There is some explanation, but details and clarity are lacking. Some components are addressed in a generic manner while others lack sufficient clarity.
			No	There are no explanations regarding the composition of the non-GAAP quantitative measure or the adjustments made (excluded items).
Prominence	5	When reporting a non-GAAP financial measure, does the bank also report the most directly comparable GAAP measure alongside it?	Yes	The most directly comparable GAAP measure is presented alongside the non-GAAP quantitative measure in a clear manner, in the same context or in close proximity, allowing for an easy and immediate comparison.	Regulations G, S-K, and C&DIs (Chen et al., 2021) and IFRS 18
			Partial	The directly comparable GAAP measure is not presented immediately before or after the non-GAAP quantitative measure. This requires more effort on the part of the user to compare the two measures.
			No	The most directly comparable GAAP measure is not presented alongside the non-GAAP quantitative measure.
	6	Does the bank provide the most directly comparable GAAP measure before the non-GAAP quantitative measure in the title, highlights, or first two paragraphs of the earnings section?	Yes	The most directly comparable GAAP measure is presented before the non-GAAP quantitative measure. It is clearly highlighted in the title, highlights, or first two paragraphs of the earnings section. The presentation is structured so that users see or consider the GAAP measure first before being exposed to the non-GAAP quantitative measure.	Regulations G, S-K, and C&DIs (Chen et al., 2021)
			Partial	The most directly comparable GAAP measure is mentioned in a less prominent position, either alongside or in the same sentence or paragraph as the non-GAAP quantitative measure, but without clear emphasis.
			No	The bank does not provide the most directly comparable GAAP measure in the title, highlights, or first two paragraphs of the earnings section.
Reconciliation	7	Is a reconciliation provided between the non-GAAP quantitative measure and the most directly comparable GAAP measure?	Yes	A clear reconciliation is provided between the non-GAAP and GAAP quantitative measures.	CVM Resolution No. 156/2022 and IFRS 18
			Partial	A reconciliation is provided, but it is insufficient or lacks clarity.
			No	No reconciliation is provided between the non-GAAP financial measure and the most directly comparable GAAP measure.
Consistency	8	When non-recurring items are present, is the same non-GAAP quantitative measure consistently disclosed in prior years?	Yes	The bank consistently discloses the same non-GAAP financial measure year over year.	Regulations G, S-K, and C&DIs (Chen et al., 2021), BCB Resolution No. 2/2020, and IFRS 18
			Partial	While there is some consistency in the disclosure of the non-GAAP quantitative measure, there have been minor variations in methodology or presentation over the years. Alternatively, the bank has changed the non-GAAP quantitative measure but has provided only partial explanations for these changes.
			No	When non-recurring items are present, the bank does not consistently disclose the same non-GAAP financial measure as in prior years.
Typical activities	9	In the breakdown of items classified as non-recurring, is it clear that these items are unrelated or only incidentally related to the institution's typical activities?	Yes	The items reported as non-recurring are clearly not related to, or only incidentally related to, the bank's typical activities.	BCB Resolution No. 2/2020
			Partial	While there is a relationship between the items classified as non-recurring and the bank's typical activities, there is some justification for their classification as non-recurring.
			No	It is unclear whether the items reported as non-recurring are unrelated or only incidentally related to the bank's typical activities.
Frequency	10	Is it clear in the breakdown of items classified as non-recurring that these items are not expected to occur frequently in future financial years?	Yes	It is clear that items classified as non-recurring are exceptional and/or not expected to occur frequently in future financial years.	BCB Resolution No. 2/2020
			Partial	The justification for the future non-recurrence of these items is vague, incomplete, or ambiguous.
			No	It is unclear whether these items are not expected to occur frequently in future financial years.
Tax effects	11	Does the bank present the effect of income taxes on non-GAAP exclusions in its textual or tabular reconciliation?	Yes	The effect of income taxes is clearly presented, either in narrative form or in a table.	Regulations G, S-K, and C&DIs (Chen et al., 2021) and IFRS 18
			Partial	The information is available, but it is fragmented, making it difficult to understand.
			No	The bank does not present the effect of income taxes on non-GAAP exclusions in its textual or tabular reconciliations.

• Note: The questions were adapted from Regulations G, S-K, and the Compliance and Disclosure Interpretations (C&DIs) of the U.S. Securities and Exchange Commission (Chen et al., 2021), Resolution No. 2/2020 of the Central Bank of Brazil (BCB), Resolution No. 156/2022 of the Brazilian Securities and Exchange Commission (CVM), and International Financial Reporting Standards Foundation (IFRS) 18.
• NGQI = non-GAAP (Generally Accepted Accounting Principles) qualitative index.
• Source: Prepared by the authors.

Responses to the 11 items were converted to the following values: 1 for "yes"; 0.5 for "partially"; and 0 for "no." Next, the response values were summed and divided by 11 to calculate the NGQI for each bank/quarter observation, as shown in Equation 2:

$${NGQI}_{i,t}=\sum _{n=1}^{11}{QIDis}_{n,i,t}/11$$ (2)

where QIDis_n,i,t is item n in Table 2, which takes the value 1 (for the response "yes"), 0.5 (for the response "partially"), or 0 (for the response "no"), with respect to bank i in quarter t.

As shown in Equation 2, the different items in the NGQI were not assigned weights. This decision is based on the rationale that since this study does not focus on a particular user, it would be arbitrary to specify which type of item would be most relevant for assigning weights, as Murcia (2009) argued. Thus, to mitigate potential subjectivity in the selection process, it was decided that weights would not be assigned to the NGQI categories and items.

As a requirement for content analysis, validation tests were conducted on the IQNG, although it is based heavily on items that were already validated in the study by Chen et al. (2021) . However, since some items were excluded and others were incorporated, a new validation of the NGQI was conducted using Cronbach's alpha following Murcia (2009) . Overall, the test coefficients reported consistency levels considered "good" or "very good," attesting to the reliability expected of the NGQI.

Table 3 presents the expected signs for each control variable, their respective rationales, how they were measured, and the references that support these relationships.

Table 3
Composition of control variables

Variables (expected signs)
Size (+)	Rationale	Since the dependent variable is a measure of performance, and since larger banks may exhibit better financial performance, the persistence models tested controlled for bank size.
	Measurement	Natural logarithm of Total Assetsit .
	References	Fabio (2019) , Petria et al. (2015) , and Sousa et al. (2021) .
Growth (+)	Motivation	Since growth in the total assets of banking institutions is one of the main sources of bank profitability and RI is a measure of bank profitability relative to their average total assets, a control variable for the growth of these assets relative to Brazil's GDP was included.
	Measurement	$\left(\frac{{TotalAsset{s}_{i,t}inrelationtoGDP}_t}{{TotalAsset{s}_{i,t-1}inrelationtoGDP}_{t-1}}\right)-1$
	Reference	Resolution No. 4,553/2017 of the National Monetary Council
List (+)	Rationale	Previous studies have shown that banks listed on stock exchanges tend to perform better than unlisted banks. Furthermore, these institutions may have greater incentives to deliver more predictable performance to attract investors.
	Measurement	A dummy variable was used, assigning a value of 1 if the bank is listed on B3 or has ADRs and a value of 0 otherwise.
	References	Fabio (2019) and Sousa et al. (2021).
ΔGDP (+)	Rationale	To control for the influence of economic growth on the dependent variable, the change in GDP was included in the econometric models. During periods of higher economic activity, it is likely that demand for credit increases, leading to better performance by banking institutions.
	Measurement	$\left(\frac{{GDP}_t}{{GDP}_{t-1}}\right)-1$
	Reference	Dantas et al. (2013) , Petria et al. (2015) , and Sousa et al. (2021)

• ADR = American Depositary Receipts; B3 = B3 S.A. - Brasil, Bolsa, Balcão; GDP = gross domestic product.
• Source: Prepared by the authors.

Additionally, to compare the persistence of recurring income and non-GAAP qualitative disclosures with that of net income reported in accordance with GAAP, a test of equality of coefficients was conducted between the alphas of the model in Equation 1 - with and without the effects of non-GAAP qualitative disclosures - and the model in Equation 3.

$${NI}_{i,t+1}={\alpha }_0+{\alpha }_1{NI}_{i,t}+{\alpha }_2{Size}_{i,t}+{\alpha }_3{Grow}_{i,t}+{\alpha }_4{List}_{i,t}+{\alpha }_5{∆GDP}_t+u_{i,t+1}$$ (3)

where NI_i,t+1 is bank i's net income in quarter t+1, weighted by the average total assets between quarters t+1 and t, and NI_i,t is bank i's net income in quarter t - a proxy for the persistence of GAAP earnings - weighted by the average total assets between quarters t and t-1.

The coefficient equality test aims to compare the coefficients α₁ RI_i,t and α₃ RI_i,t * NGQI_i,t of the model in Equation 1 - estimated without and with the effects of the NGQI, respectively - with α₁ NI_i,t of the model in Equation 3. It is expected that the recurring income will be more persistent than the net income reported in accordance with GAAP, and that the disclosure of non-GAAP qualitative information will make this non-GAAP quantitative measure even more persistent.

3.3 Additional Tests

To investigate the behavior of the NGQI at the extremes of the distribution, two additional analyses were conducted. The first aimed to identify whether higher NGQI values are associated with a non-GAAP quantitative measure of recurring income that is more persistent than the sample average. The second aimed to determine whether lower NGQI values are associated with a non-GAAP quantitative measure of recurring income that is less persistent than the sample average. Interaction terms were constructed to measure the effects of the degree of persistence of the non-GAAP quantitative measure of recurring income among observations where the NGQI variable falls in the last (0.75) and first (0.25) quartiles of the distribution. These correspond to cases with the highest and lowest non-GAAP qualitative indices, respectively, as shown in the models in equations 4 and 5.

$${RI}_{i,t+1}={\beta }_0+{\beta }_1{RI}_{i,t}+{\beta }_2{QI4}_{i,t}+{\beta }_3{RI}_{i,t}*{QI4}_{i,t}+{\beta }_4{Size}_{i,t}+{\beta }_5{Grow}_{i,t}+{\beta }_6{List}_{i,t}+{\beta }_7{∆GDP}_t+u_{i,t+1}$$ (4)

$${RI}_{i,t+1}={\delta }_0+{\delta }_1{RI}_{i,t}+{\delta }_2{QI1}_{i,t}+{\delta }_3{RI}_{i,t}*{QI1}_{i,t}+{\delta }_4{Size}_{i,t}+{\delta }_5{Grow}_{i,t}+{\delta }_6{List}_{i,t}+{\delta }_7{∆GDP}_t+u_{i,t+1}$$ (5)

where QI4_i,t is a dummy variable that takes the value 1 for bank i in the 0.75 quartile of the NGQI and 0 otherwise in quarter t, and QI1_i,t is a dummy variable that takes the value 1 for bank i in the 0.25 quartile of NGQI and 0 otherwise in quarter t.

Consistent with earlier arguments, it is expected that recurring income for quarter t will also predict future recurring income (RI_t+1 ) for the models in equations 4 and 5. However, given higher disclosure of non-GAAP qualitative information, recurring income is expected to be more persistent than in the overall sample. Thus, in the presence of higher non-GAAP qualitative indices (QI4), it is expected that the coefficient β₃ will be significant and positive and greater than β₁. Under conditions of lower non-GAAP qualitative indices (the 0.25 quartile - QI1), the recurring income of the sample (δ₁) is expected to be significant and positive and greater than the recurring income of entities with low non-GAAP qualitative indices (δ₃).

4. RESULTS

4.1 Descriptive Statistics

To better understand the data distribution, Table 4 provides descriptive statistics for the continuous (Panel A) and categorical (Panel B) variables used in the study.

Table 4
Descriptive statistics for continuous and categorical variables - Period: 2010.1 to 2023.4

Panel A: Continuous variables
Variables	n	Mean	Median	Standard deviation	Minimum	Maximum
RI	519	0.0056	0.0041	0.0087	-0.0207	0.0795
NI	519	0.0061	0.0039	0.0129	-0.0246	0.1045
NGQI	519	0.7817	0.8636	0.1667	0.0909	1.0000
Size	519	11.1714	11.1200	0.8534	9.4877	12.3995
Grow	519	0.0158	0.0060	0.0802	-0.3751	0.5935
ΔGDP	519	0.0213	0.0262	0.0383	-0.0618	0.1008
Panel B: Categorical variables
	Dummy	n	%
List	0	169	32.56
	1	350	67.44
	Total		519	100.00

• ΔGDP = change in gross domestic product (GDP); Grow = growth in institutions' assets relative to GDP; NGQI = non-GAAP (Generally Accepted Accounting Principles) qualitative index; NI = net income; List = indicates banks listed on B3 S.A. - Brasil, Bolsa, Balcão, or those with American Depositary Receipts; RI = recurring income; Size = bank size.
• Source: Prepared by the authors.

The descriptive statistics in Panel A initially suggest that the measures of central tendency for the RI variable reflect the banks' recurring returns on assets. However, upon examining the minimum and maximum values, a significant discrepancy is observed. Banco Original reported a negative return on assets, with a minimum value of -2% in the first quarter of 2023. In other words, even with non-GAAP adjustments, its earnings measure showed a loss. Meanwhile, BNDESPAR achieved an approximate 8% return (maximum value) on its assets in the first quarter of 2022.

Regarding the NI variable, the descriptive analysis shows that net income, on average, exhibits positive profitability, with a mean of 0.61% and a median of 0.39% of the banking institutions' average assets. However, the range observed between the minimum (-2.46%) and maximum (10.45%) values indicates considerable asymmetry and heterogeneity in the banks' profitability, as evidenced by the high standard deviation of 1.29%.

For the NGQI variable, the average compliance rate was 78.17% for the disclosure of non-GAAP qualitative information, in accordance with SEC Regulations G, S-K, and C&DIs, BCB Resolution No. 2/2020, CVM Resolution No. 156/2020, and IFRS 18. BRB had the lowest compliance rate in the second and third quarters of 2022. The highest rates were reported by Banese, BB, Itaú, Banrisul, ABC-Brasil, BNDESPAR, and Nubank in certain quarters of the review period. These institutions fully complied with the expected requirements.

Regarding bank size, Size has the highest standard deviation compared to the other variables. The heterogeneity in the size of the institutions stems from the objective of the study, which is to cover banks of different sizes. As for the growth of the banking institutions' assets relative to GDP (Grow), BNDESPAR's total assets decreased by about 38% in the fourth quarter of 2022. Conversely, BTG grew by 59% relative to Brazilian GDP, marking an increase in its total assets in the first quarter of 2019 - the highest growth rate in the data set.

Descriptive statistics on the level of Brazilian economic activity (ΔGDP) indicate economic growth during the period under review for both the mean and the median, although the lowest recorded value was an economic contraction in the second quarter of 2020, likely due to the initial effects of the pandemic.

Panel B shows that 67.44% of the sample consists of banks listed on B3 or with ADRs on the U.S. capital market (List variable). This suggests that most of the observed data pertains to banks with greater incentives to disclose non-GAAP quantitative measures of recurring income and accompanying qualitative information.

4.2 Regression Model Estimates

4.2.1 Persistence of recurring income

Table 5, Panel A presents the results of the estimates with robust standard errors, corrected for heteroscedasticity and autocorrelation. Column 1 shows the model from Equation 1 without considering the effects of the NGQI. This is solely to verify the persistence of recurring income prior to considering the effects of the NGQI. Column 2 shows the results of the complete model from Equation 1 to test the effects of the NGQI on the persistence of RI. Column 3 shows the estimation of the model from Equation 3, which was proposed by Dechow et al. (2010) and includes control variables. Panel B presents the results of the test for equality of coefficients to verify whether there is a difference between the NI, RI, and RI * NGQI coefficients.

Table 5
Persistence of RI with NGQI interaction and comparison with persistence of NI - Period 2010.1 to 2023.4

Models	Eqs. 1 and 2	${RI}_{i,t+1}={\alpha }_0+{\alpha }_1{RI}_{i,t}+{\alpha }_2{NGQI}_{i,t}+{\alpha }_{3}R{I}_{i,t}*{NGQI}_{i,t}+{\alpha }_4{Size}_{i,t}+{\alpha }_5{Grow}_{i,t}+{\alpha }_6{List}_{i,t}+{\alpha }_7{∆GNP}_t+u_{i,t+1}$
	Eq. 3	${NI}_{i,t+1}={\alpha }_0+{\alpha }_1{NI}_{i,t}+{\alpha }_2{Size}_{i,t}+{\alpha }_3{Grow}_{i,t}+{\alpha }_4{List}_{i,t}+{\alpha }_5{∆GNP}_t+u_{i,t+1}$
Panel A: Regression estimates
Variables		1	2	3
C		0.0127	0.0307**	0.0179
		(0.0082)	(0.0128)	(0.0151)
RI		0.6339***	-1.1322***
		(0.0741)	(0.3838)
NGQI			-0.0084***
			(0.0019)
RI * NGQI			1.9432***
			(0.4099)
NI				0.0214
				(0.0456)
Size		-0.0009	-0.0018	-0.0010
		(0.0007)	(0.0011)	(0.0013)
Grow		0.0032	0.0042	0.0129
		(0.0033)	(0.0028)	(0.0103)
List		-0.0007*	-0.0010**	-0.0001*
		(0.0004)	(0.0004)	(0.0005)
ΔGDP		0.0034	0.0063*	0.0143
		(0.0035)	(0.0036)	(0.0132)
N		519	519	519
R²_w		0.3691	0.4456	0.0221
R²_b		0.9302	0.8419	0.0011
R²_o		0.7806	0.7468	0.0197
F-test		56.51***	55.22***	2.18*
Panel B: Test for equality of coefficients
NI		0.0214	0.0214
RI		0.6339
RI * NGQI			1.9432
χ2		23.3200	6.4800
Prob > χ²		0.0000	0.0109

• Note: Robust standard errors, corrected for heteroscedasticity and autocorrelation, are shown in parentheses.
• NI_t+1 = net income for quarter t+1; ΔGDP = change in gross domestic product (GDP); C = constant; Grow = growth in institutions' assets relative to GDP; NGQI = non-GAAP (Generally Accepted Accounting Principles) qualitative index; NI = net income for quarter t; 𝐿𝑖𝑠𝑡 = indicates banks listed on B3 S.A. - Brasil, Bolsa, Balcão, or those with American Depositary Receipts; RI_t+1 = recurring income for quarter t+1; RI = recurring income for quarter t; Size = bank size.
• Statistical significance level: *** at 1%, ** at 5%, and * at 10%.
• Source: Prepared by the authors.

Panel A, Column 1 shows that the persistence coefficient, represented by the RI variable, is positive and significant at the 1% level. It is also close to 1, suggesting that future recurring income (RI_t+1 ) is, on average, explained by current recurring income (RI). This finding is consistent with previous literature (Doyle et al., 2003; Francis et al., 2004; Sousa et al., 2021).

Incorporating the effects of the NGQI through the interaction term RI * NGQI - testing the hypothesis that the persistence of banking institutions' non-GAAP quantitative measures is directly associated with the NGQI - supports hypothesis H₁ by showing that the coefficient 𝛼 3 (1.9432) is positive and greater than 𝛼 1 . This indicates that an increase in the NGQI intensifies the persistence of RI, highlighting the positive moderating effect of disclosure quality on RI persistence. Regarding the coefficient α₁ (RI) in Column 2, the negative sign suggests low baseline persistence, corresponding to the effect of RI when NGQI approaches 0 and capturing the persistence of the group with the lowest disclosure quality.

The findings suggest that voluntary disclosure of non-GAAP qualitative information enhances the persistence of recurring income and improves its predictability (RI_t+1 ). In summary, the higher the quality of the NGQI, the greater the persistence of recurring income. Similar to the findings of Chen et al. (2021) , who found improvements in the quality of non-GAAP exclusions when interacting with the NGQI in U.S. non-financial firms, this study indicates that, in the Brazilian banking market, the NGQI serves a similar function. The NGQI improves the informational quality of the non-GAAP quantitative measure reported by banks. By disclosing their recurring income measures alongside qualitative information explaining these metrics, banks make their reported measure better able to predict future earnings.

Regarding the control variables, Size and Grow showed no statistically significant relationship with future recurring income (RI_t+1 ). This suggests that the size and growth of banks are not explanatory factors for the non-GAAP quantitative measure of future recurring income. This result is counterintuitive since one would naturally expect higher (lower) levels of profitability to be influenced by the size or growth of the institutions (Fabio, 2019; Petria et al., 2015). It is worth noting that Sousa et al. (2021) found a similar result, suggesting that this may be a characteristic specific to the Brazilian banking market. Additional tests using market capitalization and shareholders' equity as measures of size and/or growth yielded equivalent results to those found when using total assets.

Banks listed on B3 or with ADRs on the U.S. capital market (List) have a negative relationship with future recurring income (RI_t+1 ). This behavior is observed in all estimates, with slight variations in the magnitude of the coefficients. The findings suggest that, on average, banks listed on B3 or with ADRs on the U.S. capital market exhibit future performance (measured by recurring income) that is 0.0010 units lower than unlisted banks. This is contrary to the expectation that listed banks have greater incentives to deliver predictable performance to attract investors (Fabio et al., 2019; Sousa et al., 2021).

When analyzing the level of economic growth (ΔGDP), a positive relationship with future recurring income (RI_t+1 ) can be observed in Column 2. This corroborates previous literature (Dantas et al., 2013; Petria et al., 2015; Sousa et al., 2021) based on the idea that during periods of economic growth, demand for credit increases, consequently affecting the performance of banking institutions.

Panel B shows the tests of equality between the net income persistence coefficients (α₁ NI_i,t ), derived from the estimation in Column 3, which explains future net income (NI_t+1 ), and the coefficients of the terms for (i) recurring income persistence (α₁RI), obtained in Column 1, and (ii) the interaction between recurring income persistence and the NGQI (α₃RI_i,t * NGQI), obtained in Column 2. The NI coefficient was small and not significant (α₁ = 0.0214; p = 0.868), indicating that the ability of net income to persist is low across the set of banking institutions analyzed. This suggests the presence of transitory components.

The test of equality between the coefficients α₁NI_i,t and α₁RI rejects the null hypothesis (χ² = 23.32; p = 0), showing that the persistence of RI is statistically greater than that of net income. Furthermore, the test of equality between the parameters α₁NI_i,t and α₃RI * NGQI shows a statistically significant difference (χ² = 6.48; p = 0.0109), indicating that the persistence of RI * NGQI is significantly different and more economically relevant than the persistence of GAAP net income.

4.2.2 Extremes of the NGQI

Table 6 presents the estimate results from the models in equations 4 and 5, which assess the effects of the extremes of the NGQI, complementing the findings in Table 5. The mean coefficients of the individual variables (RI and NGQI) may be influenced by extreme levels of qualitative information transparency. On average, non-GAAP qualitative information is positively associated with more persistent non-GAAP quantitative measures. Therefore, in scenarios with high (low) levels of qualitative information transparency, the persistence of the non-GAAP quantitative measure will likely be greater (lower).

Table 6
Estimates of the extremes of the non-GAAP qualitative index - Period: 2010.1 to 2023.4

Models	Eq. 4	${RI}_{i,t+1}={\beta }_0+{\beta }_1{RI}_{i,t}+{\beta }_2{QI4}_{i,t}+{\beta }_3{RI}_{i,t}*{QI4}_{i,t}+{\beta }_4{Size}_{i,t}+{\beta }_5{Grow}_{i,t}+{\beta }_6{List}_{i,t}+{\beta }_7{∆GDP}_t+u_{i,t+1}$
	Eq. 5	${RI}_{i,t+1}={\delta }_0+{\delta }_1{RI}_{i,t}+{\delta }_2{QI1}_{i,t}+{\delta }_3{RI}_{i,t}*{QI1}_{i,t}+{\delta }_4{Size}_{i,t}+{\delta }_5{Grow}_{i,t}+{\delta }_6{List}_{i,t}+{\delta }_7{∆GDP}_t+u_{i,t+1}$
Variables		Model 4	Model 5
C		0.0227*	0.0206
		(0.0132)	(0.0126)
RI		0.2548*	0.6835***
		(0.1327)	(0.0315)
QI4		-0.0021***
		(0.0006)
RI * QI4		0.4320***
		(0.1366)
QI1			0.0029***
			(0.0007)
RI * QI1			-0.5540***
			(0.1221)
Size		-0.0016	-0.0016
		(0.0011)	(0.0011)
Grow		0.0042	0.0047
		(0.0034)	(0.0034)
List		-0.0010*	-0.0010*
		(0.0005)	(0.0005)
ΔGDP		0.0048	0.0056
		(0.0037)	(0.0039)
N		519	519
R²_w		0.3897	0.3959
R²_b		0.9135	0.9005
R²_o		0.7589	0.7499
F-test		43.88***	45.04***

• Note: Robust standard errors, corrected for heteroscedasticity and autocorrelation, are shown in parentheses.
• ΔGDP = change in gross domestic product (GDP); C = constant; Grow = growth in institutions' assets relative to GDP; QI1 = observations below the 0.25 quartile of the NGQI [non-GAAP (Generally Accepted Accounting Principles) qualitative index]; QI4 = observations above the 0.75 quartile of the NGQI; List = indicates banks listed on B3 S.A. - Brasil, Bolsa, Balcão, or those with American Depositary Receipts; RI_t+1 = recurring income for quarter t+1; RI = recurring income for quarter t; Size = bank size.
• Statistical significance level: *** at 1%, ** at 5%, and * at 10%.
• Source: Prepared by the authors.

The findings reveal two key issues in the model in Equation 4, which reports the results for the upper end of the NGQI - those concentrated in the 0.75 quartile (QI4). First, the RI variable, which represents the persistence of recurring income not captured by the interaction variable in Equation 4, has a positive and significant coefficient at the 10% level. This indicates that although the coefficient is statistically significant, the relationship is not strong. Second, when isolating the NGQI of banks with the highest levels of non-GAAP qualitative disclosures (QI4), there is an increase in the persistence of recurring income (RI * QI4). Consequently, the resulting interaction coefficient is positive, significant, and greater than the coefficient corresponding to the sample mean not captured by the interaction (β₃ > β₁). The findings suggest that higher transparency regarding non-GAAP qualitative information leads to a higher informational quality of the non-GAAP quantitative measure of recurring income. This is because recurring income in the current period is more predictive of future recurring income (RI_t+1) in banks with higher non-GAAP qualitative indices.

Analyzing the lower end of the NGQI (the 0.25 quartile, or QI1), we observe in the model in Equation 5 that the persistence of recurring income (RI) coefficient is positive and significant at the 1% level. It is also close to 1. When the effect of lower indices (QI1) is interacted with recurring income (RI * QI1), the reported interaction term coefficient is negative, statistically significant, and lower than the coefficient for the remainder of the sample (δ₃ < δ₁). This indicates that, at low levels of transparency in non-GAAP qualitative disclosures, recurring income also exhibit low informational quality.

Combining the findings from the models in equations 4 and 5 shows that when analyzing the extremes of non-GAAP qualitative disclosures, the transparency level of this information is reflected in the quality of the reported non-GAAP quantitative measure. The findings suggest that at higher levels of non-GAAP qualitative information disclosure, the non-GAAP quantitative measure of recurring income is more persistent, and at lower levels, it is less persistent. Furthermore, examining the extremes corroborated the regression results of the model in Equation 1 once again, as it was observed that the extremes of the indices impact the quality of recurring income differently.

Overall, the findings from the main model and the additional tests indicate that non-GAAP qualitative disclosures are positively associated with greater persistence of non-GAAP quantitative measures in Brazilian banks, and these relationships may be more or less pronounced at higher or lower levels of qualitative disclosure.

5. CONCLUSION

This study examined the relationship between non-GAAP qualitative information and the persistence of non-GAAP quantitative measures in Brazilian banks. The argument is that disclosing transparent non-GAAP qualitative information (a proxy for non-GAAP quality) leads to higher informational quality in terms of persistence of non-GAAP quantitative measures.

To this end, the earnings persistence model from the accounting information quality literature was employed. Empirical tests revealed that, on average, the persistence of the non-GAAP quantitative measure of recurring income reported by Brazilian banks is directly associated with the NGQI practiced by these institutions. This association is statistically significant in relation to the net income reported in accordance with GAAP. Upon analyzing the extremes of the NGQI, the study revealed that the non-GAAP quantitative measure of recurring income exhibited greater persistence at higher levels of transparency of non-GAAP qualitative information and less persistence at lower levels. Thus, the findings of this study indicated that increases in non-GAAP qualitative indices translate into an increase in the informational quality of the non-GAAP quantitative measure.

The evidence found in this study makes important contributions to the Brazilian banking sector. First, it helps fill a gap in the literature exploring issues inherent to non-GAAP disclosures by banking institutions, which have historically reported this type of information but received little attention from researchers who typically excluded such entities from their research samples.

Second, the study contributes to Brazilian regulatory agencies, particularly the BCB and the CVM. The reported results may raise red flags for regulators because requiring guidelines related to non-GAAP qualitative information could improve the quality of the non-GAAP quantitative measure of recurring income reported by banks. Since this non-GAAP qualitative information is voluntary, requiring its disclosure could have an even more beneficial effect on the quality of reported recurring income, a measure already mandated by the BCB. Furthermore, investors may consequently observe improvements in the quality of the quantitative non-GAAP measure resulting from greater transparency in non-GAAP qualitative information, and they also benefit from the findings of this study.

A key limitation of the study is that it is not possible to draw inferences about banking institutions in other countries because the sample was restricted to Brazilian banks and specific elements of Brazilian regulation developed by the BCB and the CVM were incorporated into the NGQI validated in this study. Therefore, future research could extend the study to banking institutions on a global scale and analyze common factors to create a more generalized index. Additionally, such research could broaden the applicability of these results by using an object of analysis other than performance reports, such as market perception based on analyst consensus.

ACKNOWLEDGMENTS

The authors thank the anonymous reviewers of RC&F and the members of the doctoral dissertation committee, Prof. Dr. Roberto Carlos Klann, Prof. Dr. Edilson Paulo and Prof. Dr. Ducineli Régis Botelho, for their contributions to the improvement of this study.

REFERENCES

• Banco Central do Brasil (Central Bank of Brazil). (2020). BCB Resolution No. 2, dated August 12, 2020. Establishes general criteria for consortium administrators and payment institutions to prepare and disclose individual and consolidated financial statements, as well as procedures for financial institutions and other Central Bank of Brazil-authorized institutions to prepare, disclose, and submit financial statements. https://www.bcb.gov.br/estabilidadefinanceira/exibenormativo?tipo=Resolu%C3%A7%C3%A3o%20BCB№=2
  » https://www.bcb.gov.br/estabilidadefinanceira/exibenormativo?tipo=Resolu%C3%A7%C3%A3o%20BCB№=2
• Bardin, L. (2016). Análise de conteúdo (Revised and expanded edition). Edições 70 Brasil. (Original paper published in 1977).
• Bentley, J. W., Christensen, T. E., Gee, K. H., & Whipple, B. C. (2018). Disentangling managers’ and analysts’ non-GAAP reporting. Journal of Accounting Research, 56(4), 1039-1081. https://doi.org/10.1111/1475-679X.12206
  » https://doi.org/10.1111/1475-679X.12206
• Black, D. E., Christensen, T. E., Ciesielski, J. T., & Whipple, B. C. (2018). Non-GAAP reporting: Evidence from academia and current practice. Journal of Business Finance and Accounting, 45(3-4), 259-294. https://doi.org/10.1111/jbfa.12298
  » https://doi.org/10.1111/jbfa.12298
• Bradshaw, M. T., Christensen, T. E., Gee, K. H., & Whipple, B. C. (2018). Analysts’ GAAP earnings forecasts and their implications for accounting research. Journal of Accounting and Economics, 66(1), 1-26. https://doi.org/10.1016/j.jacceco.2018.01.003
  » https://doi.org/10.1016/j.jacceco.2018.01.003
• Chen, H. C., Lee, Y. J., Lo, S. Y., & Yu, Y. (2021). Qualitative characteristics of non-GAAP disclosures and non-GAAP earnings quality. Journal of Accounting and Economics , 72(1), 1-24. https://doi.org/10.1016/j.jacceco.2021.101402
  » https://doi.org/10.1016/j.jacceco.2021.101402
• Choi, Y. S., Lin, S., Walker, M., & Young, S. (2007). Disagreement over the persistence of earnings components: Evidence on the properties of management-specific adjustments to GAAP earnings. Review of Accounting Studies, 12, 595-622. https://doi.org/10.1007/s11142-007-9048-x
  » https://doi.org/10.1007/s11142-007-9048-x
• Comissão de Valores Mobiliários (Brazilian Securities and Exchange Commission) (2022). CVM Resolution No. 156, dated June 23, 2022. Provides for the voluntary disclosure of non-accounting information referred to as LAJIDA and LAJIR. https://conteudo.cvm.gov.br/legislacao/resolucoes/resol156.html
  » https://conteudo.cvm.gov.br/legislacao/resolucoes/resol156.html
• Conselho Monetário Nacional (National Monetary Council). (2017). CMN Resolution No. 4,553, dated January 30, 2017. Establishes the segmentation of all financial institutions and other institutions authorized to operate by the Central Bank of Brazil for the purposes of the proportional application of prudential regulations. https://www.bcb.gov.br/estabilidadefinanceira/exibenormativo?tipo=RESOLU%C3%87%C3%83O№=4553
  » https://www.bcb.gov.br/estabilidadefinanceira/exibenormativo?tipo=RESOLU%C3%87%C3%83O№=4553
• Dantas, J. A., Medeiros, O. R., Galdi, F. C., & Costa, F. M. (2013). Gerenciamento de resultados em bancos com uso de TVM: Validação de modelo de dois estágios. Revista Contabilidade & Finanças, 24(61), 37-54. https://doi.org/10.1590/S1519-70772013000100005
  » https://doi.org/10.1590/S1519-70772013000100005
• Dechow, P. M., Ge, W., & Schrand, C. (2010). Understanding earnings quality: A review of the proxies, their determinants, and their consequences. Journal of Accounting and Economics , 50(2-3), 344-401. https://doi.org/10.1016/j.jacceco.2010.09.001
  » https://doi.org/10.1016/j.jacceco.2010.09.001
• Dechow, P. M., Hutton, A. P., & Sloan, R. G. (1999). An empirical assessment of the residual income valuation model. Journal of Accounting and Economics , 26(1-3), 1-34. https://doi.org/10.1016/S0165-4101(98)00049-4
  » https://doi.org/10.1016/S0165-4101(98)00049-4
• Doyle, J. T., Lundholm, R. J., & Soliman, M. K. (2003). The predictive value of expenses excluded from pro forma earnings. Review of Accounting Studies , 8, 145-174. https://doi.org/10.1023/A:1024472210359
  » https://doi.org/10.1023/A:1024472210359
• Fabio, C. D. (2019). Earnings quality and external governance on banks: Empirical evidence in the European context. International Journal of Accounting, Auditing and Performance, 15(4), 410-431. https://doi.org/10.1504/IJAAPE.2019.106426
  » https://doi.org/10.1504/IJAAPE.2019.106426
• Fairfield, P. M., Sweeney, R. J., & Yohn, T. L. (1996). Accounting classification and the predictive content of earnings. The Accounting Review, 71(3), 337-355. https://www.jstor.org/stable/248292
  » https://www.jstor.org/stable/248292
• Francis, J., LaFond, R., Olsson, P. M., & Schipper, K. (2004). Costs of equity and earnings attributes. The Accounting Review , 79(4), 967-101. https://doi.org/10.2308/accr.2004.79.4.967
  » https://doi.org/10.2308/accr.2004.79.4.967
• Gu, Z., & Chen, T. (2004). Analysts’ treatment of nonrecurring items in street earnings. Journal of Accounting and Economics , 38(1-3), 129-170. https://doi.org/10.1016/j.jacceco.2004.09.002
  » https://doi.org/10.1016/j.jacceco.2004.09.002
• Hribar, P., Mergenthaler, R., Roeschley, A., Young, S., & Zhao, C. X. (2022). Do managers issue more voluntary disclosure when GAAP limits their reporting discretion in financial statements? Journal of Accounting Research , 60(1), 299-351. https://doi.org/10.1111/1475-679X.12401
  » https://doi.org/10.1111/1475-679X.12401
• International Financial Reporting Standards Foundation. (2019). Basis for conclusions on IFRS practice statement 2: Making materiality judgements - General presentation and disclosures. https://www.ifrs.org/projects/completed-projects/2024/primary-financial-statements/ed-primary-financial-statements/
  » https://www.ifrs.org/projects/completed-projects/2024/primary-financial-statements/ed-primary-financial-statements/
• International Financial Reporting Standards Foundation. (2024). IFRS 18: Presentation and disclosure in financial statements. https://www.ifrs.org/supporting-implementation/supporting-materials-by-ifrs-standards/ifrs-18/
  » https://www.ifrs.org/supporting-implementation/supporting-materials-by-ifrs-standards/ifrs-18/
• Landsman, W. R., Miller, B. L., & Yeh, S. (2007). Implications of components of income excluded from pro forma earnings for future profitability and equity valuation. Journal of Business Finance & Accounting, 34(3-4), 650-675. https://doi.org/10.1111/j.1468-5957.2007.02033.x
  » https://doi.org/10.1111/j.1468-5957.2007.02033.x
• Murcia, F. D.-R. (2009). Fatores determinantes do nível de disclosure voluntário de companhias abertas no Brasil [Tese de Doutorado]. Universidade de São Paulo. 2009.
• Petria, N., Capraru, B., & Ihnatov, I. (2015). Determinants of banks’ profitability: Evidence from EU 27 banking systems. Procedia Economics and Finance, 20, 518-524. https://doi.org/10.1016/S2212-5671(15)00104-5
  » https://doi.org/10.1016/S2212-5671(15)00104-5
• Ribeiro, A., Shan, Y., & Taylor, S. (2019). Non-GAAP earnings and the earnings quality trade-off. Abacus, 55(1), 6-41. https://doi.org/10.1111/abac.12150
  » https://doi.org/10.1111/abac.12150
• Sousa, R. A. M., Dantas, J. A., Nascimento, M. R., & Gonçalves, R. S. (2021). Lucro contábil versus resultado recorrente: evidências da persistência na indústria bancária brasileira. Revista Universo Contábil, 17(4), 106-126. https://doi.org/10.4270/ruc.2021428
  » https://doi.org/10.4270/ruc.2021428
• U.S. Securities and Exchange Commission. (2003). Final rule: Conditions for use of non-GAAP financial measures(Release No. 33-8176). https://www.sec.gov/files/rules/final/33-8176.htm
  » https://www.sec.gov/files/rules/final/33-8176.htm

• This is a bilingual text. This article was originally written in Portuguese and published under the DOI https://doi.org/10.1590/1808-057x20262380.pt.
• The article is based on a doctoral dissertation written by Raíssa Aglé Moura de Sousa and supervised by José Alves Dantas in 2025.
• DATA AVAILABILITY STATEMENT
  The entire dataset supporting the results of this study can be made available upon request to the authors.
• GENERATIVE AI DISCLOSURE
  The authors declare that generative artificial intelligence was used in the following stages of the production of this manuscript: - Text refinement: OpenAI for grammatical review and stylistic improvement. - Translation: OpenAI for translating text excerpts. The authors declare that, regardless of the use of the tools mentioned above, all generated content was supervised, verified, and critically validated by humans. The authors assume full and exclusive responsibility for the accuracy of the data, integrity of mathematical/statistical formulas, originality of the text, and the conclusions presented in the published article.

Edited by

• Academic Editor-in-Chief:
  Andson Braga de Aguiar

• Associate Editor:
  Eduardo da Silva Flores

Data availability

The entire dataset supporting the results of this study can be made available upon request to the authors.

Publication Dates

• Publication in this collection
  18 Sept 2026
• Date of issue
  2026

History

• Received
  23 May 2025
• Accepted
  02 July 2025
• Accepted
  23 Feb 2026