Loan loss provisions and income smoothing in banks: the role of trade openness and IFRS in BRICS

Abstract

Purpose

This paper empirically investigates whether trade openness (TO) in Brazil, Russia, India, China and South Africa (BRICS) countries affects how banks might employ loan loss provisions (LLPs) to smooth out their earnings and how adopting the International Financial Reporting Standards (IFRS) can mitigate it.

Design/methodology/approach

The analysis includes 78 commercial banks from five BRICS nations and spans 2014 through 2020. To test these hypotheses, the authors utilized a fixed-effect and two-step system panel generalized methods of moments (GMM) estimator.

Findings

TO positively affects income smoothing (earnings management) across BRICS commercial banks. The effect is clearer in banks that make financial reports under the IFRS. Path analysis reveals that the effect of TO is driven by nonperforming loans (NPLs). Additionally, the IFRS restricts earnings management in the BRICS banking sector when a better institutional environment is present. The authors found that accounting rules (IFRS) and enforcement (better institutional settings) interact to enhance earnings’ quality.

Practical implications

The relationship between TO and bank earnings management practices is important for understanding the complex interplay between trade and finance and ensuring financial stability, investor confidence and regulatory compliance. This study recommends better regulations and governance mechanisms for financial reports in emerging nations like BRICS. Additionally, macro-prudential regulators and banking supervisors should work closely to ensure transparent TO decisions with improved discipline, institutional quality and regulatory support to enhance bank stability.

Originality/value

The study finds evidence of bank income smoothing in the BRICS and introduces TO as a determinant. It also identifies the evolving role of IFRS in the presence of higher institutional quality and TO, thereby expanding the financial reporting literature.

Citation

Biswas, S., Bhattacharya, S.N., Jin, J.Y., Bhattacharya, M. and Sadarangani, P.H. (2024), "Loan loss provisions and income smoothing in banks: the role of trade openness and IFRS in BRICS", China Accounting and Finance Review, Vol. 26 No. 1, pp. 76-101. https://doi.org/10.1108/CAFR-03-2023-0037

1. Introduction

Banking is at the core of most emerging economies’ financial systems, as bank-based financing is more prevalent than market financing. Government-controlled banks dominate the banking space in emerging economies, and the relevant governments’ socioeconomic objectives typically determine the flow of bank credit to selected industries. Such lending frequently takes precedence over commercial priorities. As a result, firm distress may impact the financial systems’ stability when banks have to acknowledge more loan losses. Banks maintain provisions to cover loan losses, and an expected rise in loan losses requires banks to create more loan loss provisions (LLPs). LLPs are crucial to the banking industry because they directly impact profitability, which in turn affects retained earnings and future capital. They also serve as a tool for signaling and providing information about credit risk to interested stakeholders because LLPs are included in financial reports. In addition, bank managers can exercise discretion in determining the LLPs (Fonseca & Gonzalez, 2008; Hou, Wang, Lian, & Li, 2021; Kanagaretnam, Lobo & Mathieu, 2003, Kanagaretnam, Lim, & Lobo, 2010). As a result, management may use LLPs to transmit private information to the market, exert discretion over reported net income and choose the size and timing of LLPs. This may lead to income smoothing (Bouvatier, Lepetit, & Strobel, 2014; Curcio, De Simone, & Gallo, 2017; Osma, Mora, & Porcuna-Enguix, 2019). Income smoothing reduces the transparency of financial reporting and increases earnings opacity. The accounting standard setters view income smoothing as earnings management (Ozili, 2017; Walker, 2013).

Earnings management is the intentional manipulation of financial statements to achieve predetermined targets. The extant literature reveals that earnings management is executed primarily for opportunistic purposes (Healy & Wahlen, 1999; Katmon & Farooque, 2017; Siregar & Utama, 2008). Earnings management through discretion in LLPs can contribute to systemic crash-and-distress risk (Bushman & Williams, 2015; Ma & Song, 2016) and negatively affect bank stability (Bushman, 2014). Therefore, bank managers’ opportunistic income-smoothing efforts raise issues regarding the quality of reporting accounting information and bank supervision. Banks face risk despite operating in a controlled environment because earnings fluctuate depending on how well their clients businesses do. This frequently prompts bank managers to smooth earnings (income smoothing) to enhance perceptions of risk among their stakeholders. Other explanations for managerial attempts to smooth income focus on: (1) regulatory constraints and accounting practices or (2) agency or compensation problems (Bouvatier et al., 2014; Ozili & Outa, 2018).

The economic environment can also intensify managerial incentives to smooth income because LLPs reflect banks’ credit risk, which is affected by external factors such as the economic environment (Sinkey & Greenawalt, 1991). Trade openness (TO) has been one of the significant economic policies in most emerging economies in recent decades; however, TO is an under-researched macroeconomic factor in the income-smoothing literature, we argue that higher TO can incentivize banks to smooth earnings via LLPs. Higher TO increases the demand for bank finance to facilitate increased domestic production facilities, working capital and international trade. On the one hand, it should improve productivity and economic growth through improved resource allocation and provide diversification opportunities for banks (Bui & Bui, 2020). On the other hand, TO is also known for its destabilizing effects by exposing the domestic economy to international business cycles, resulting in higher consumption, income and price volatility. It exposes a bank’s domestic borrowers to more vulnerable economic conditions, which may result in variability of bank earnings. When a bank’s loan portfolio performance is expected to be affected by such a volatile environment, bank managers may initiate income smoothing through LLPs to mitigate the adverse impact on bank earnings. Also, when the product market becomes more competitive due to open trade, there is an increase in demand for bank finance, leading to financial development in the form of a competitive banking industry (Ashraf, 2018). The credit market competition can force banks to lower loan interest, affecting bank revenue. Moreover, banks may slacken lending criteria to maintain their market dominance, which would result in loans of poor quality showing up on bank balance sheets. Volatile economic situations make it more likely for loans with low credit quality to appear on bank balance sheets, which affects nonperforming loans (NPLs) (Cifter, Yilmazer, & Cifter, 2009). The NPLs directly impact LLPs and bank earnings. Because managers have incentives (e.g. compensation/bonus, contractual obligations, avoiding political costs, concealing excessive risk-taking behavior and others) and considerable discretion over LLPs, they can manipulate LLPs to increase or decrease reported income depending on the incentives.

While TO promotes income smoothing, it also creates the need for improved accounting standards, which can harmonize the accounting rules for external reporting. The implementation of improved accounting standards can enhance the transparency of financial reports. The International Financial Reporting Standards (IFRS) are considered superior accounting standards and can significantly reduce managerial judgment (Amidu & Issahaku, 2019). The IFRS refers to the comprehensive set of accounting standards established by the International Accounting Standards Board (IASB), which various countries have adopted and integrated in their firm’s reporting structure over time to resolve diversity in the practice of reporting transactions. When firms initially adopted it, it was a significant regulatory shift that affected companies across the globe and its benefits and costs were initially unclear. Due to a lack of information, the debates at the time regarding the effects of the IFRS adoption were primarily conjectural remarks (e.g. Ball, 2006). With the evidence of over 10 years of the IFRS reporting adopted by banks in many countries, we expect that the change in the recognition and measurement of banks’ main operating accruals affects earnings management by banks. The IASB updates the IFRS regulations frequently to reflect evolving business practices, technological improvements and economic concerns [1].

The IFRS is expected to enhance the uniformity and comparability of financial reports, which should bring transparency to accounting information. Therefore, implementing the IFRS can offset the incentives created by TO for managers to smooth earnings. Since TO creates global stakeholders, uniform accounting standards can help investors and regulators across the globe better interpret the numbers and identify red flags. However, the impact of IFRS adoption on earnings management has been inconclusive (Brüggemann, Hitz, & Sellhorn, 2013; Hasan, Hossain, Rekabder, Molla, & Ashif, 2022; Soderstrom & Sun, 2007). While early adopters of the IFRS needed transparency in their reporting for external financing, mandatory adopters faced covert options and subjective estimates during implementation, leading to income smoothing after the IFRS adoption. Countries faced varied issues during implementation, and the impact of IFRS adoption is heterogeneous.

Here we provide evidence of income smoothing in the financial reports of Brazil, Russia, India, China and South Africa (BRICS) [2] banking sector using a fixed-effect model. We also explore the role of TO in affecting such attempts and how the IFRS can mitigate it [3]. In addition to the main analysis, we explore whether TO increases earning smoothing in banks through LLPs, which remains valid to alternative measures of TO, model specification and other cross-sectional analyses. We used two-step system-generalized methods of moments (GMM) models with Windmeijer’s finite sample correction to address the potential endogeneity issues with explanatory variables and avoid possible downward biased standard errors. Our results remain robust across the models even when we consider institutional interventions by including institution quality as an additional variable to rule out possible alternative explanations of the results. The results are consistent in sensitivity analysis when we checked whether the overrepresentation of certain countries in our sample has a bearing on the final results. Our empirical findings suggest that TO promotes income smoothing in banks; the positive impact of TO is more pronounced for the IFRS adaptor banks. Ke, Lennox, and Xin (2015) showed that a weak institutional environment leads to lower-quality audits in China. We augment this argument by showing that banks adopting the IFRS effectively mitigate earnings management only in a strong institutional environment in BRICS.

Our analysis is of interest to policymakers, banking regulators and accounting standard setters in BRICS for the following reasons: first, the evidence of earning smoothing in the banking sector of BRICS helps assess the banking industry’s income smoothing practices using loan loss provisioning. As bank provisions are an important aspect of micro-prudential surveillance, understanding the extent of such practices is important for the stability of the banking system. The 2008 global financial crisis showed how susceptible the banking sector can be in the face of an economic crisis when systemic risk increases and income smoothing negatively affects banking systemic risk (Liu & Ryan, 2006; Li, Ma, & Wu, 2021). Second, we show that TO is positively linked to earnings smoothing and NPLs act as a mediators. TO induces credit demand to drive economic growth and banks often relax credit standards to enhance credit support, raising the risk of NPLs and thus the requirement for loan loss provisioning and, as a result, earnings smoothing. BRICS contributes around 18% of the global trade and banks play a crucial role in facilitating cross-border trade. Therefore, ensuring the integrity of the banking system is crucial for financial stability. This insight can be used by policymakers in emerging economies when framing rules for specific loan loss provisioning and expected loss-based provisioning. Third, we provide broad evidence that implementing the IFRS is not effective enough to curb income smoothing in the BRICS region, and we extend this finding in the presence of TO. Fourth, our study shows that institutional quality and IFRS complement each other in discouraging earnings management in BRICS banks.

The rest of the paper proceeds as follows: In Section 2, we discuss the background and develop our hypothesis. In Section 3, we describe the data, the variables and the methodology. Section 4 presents our findings, while Section 5 concludes.

2. Background and hypothesis development

2.1 Theoretical background

The agency theory proposed by Jensen and Meckling (1976) and the positive accounting theory (PAT) proposed by Watts and Zimmerman (1978) explain earnings management behavior. The agency theory emphasizes the principal-agent relationship, an information gap between insiders and outside stakeholders, leading to income smoothing due to the agency relationship (Acharya & Lambrecht, 2015). The PAT suggests discretionary accounting choices where managers might be interested in reporting higher profits or smooth earnings because of incentives, political costs (e.g. tax, regulatory and political scrutiny) and debt covenants.

The agency theory has been widely used in banking research to explain excessive risk-taking behavior by bank managers for speculation, hedging, competition and incentives (Boyd and DE Nicoló, 2005; Girotti & Salvadè, 2022; Minhat & Abdullah, 2016). Using the lens of the agency theory, Galdi, De Moura, and França (2021) explained earnings management in Brazilian banks using LLPs, while AlQudah, Azzam, Haija, and AlSmadi (2020) explained the monitoring role of diverse ownership structures in mitigating earnings management in Jordanian banks. Using the agency theory to explain the board’s role in inhibiting bankers’ opportunistic discretions, Fan, Jiang, Zhang, and Zhou (2019) explained an inverted U-shaped relationship between the number of women on boards and bank earnings management in the USA, while Sadaa, Ganesan, and Ahmed (2023) noted moderating effects of ownership structure and gender diversity on the board on earnings management in Iraqi banks. Chaity and Islam (2022) used the agency theory to comprehend the relationship between bank efficiency and earnings management of Bangladeshi banks. Alam, Ramachandran, and Nahomy (2020) studied the impact of governance and agency on earnings management behavior by Islamic and conventional banks while observing that religiosity does not discourage earnings management within financial institutions, especially Islamic banks.

Under the PAT, managers make less conservative accounting choices (debt–equity hypothesis) or more conservative ones (political costs hypothesis) to create benefits for the firm. However, they could also adopt a less conservative and less volatile accounting procedure (bonus plan hypothesis) to their benefit instead of the firm’s interests. Kwak, Lee, and Eldridge (2009) and Kwak, Lee, and Mande (2009) investigated the relationship between bank size and earnings management in Japanese banks using the political cost hypothesis. Watts and Zimmerman (1978) assert that because larger and more profitable corporations are subject to greater public scrutiny than smaller ones, larger corporations are more vulnerable to social and public pressures. As a result, larger companies use earnings management strategies that reduce income in order to avoid political costs. Lassoued, Attia, and Sassi (2018) used the political cost hypothesis to explain the relationship between institutional and blockholder ownership and income-decreasing earnings management in banks across the Middle East and North Africa (MENA) region. Nevertheless, in a study by PAT has been successfully used to explain South African banks’ earnings management behavior (Ozili & Outa, 2018), European banks (Ozili & Thankom, 2018) and Vietnamese banks (Thinh & Thu, 2020), while Ozili (2019) finds no empirical support for PAT in African banking institutions.

In this study, we have used the agency theory and PAT to explain the earnings management practices of the BRICS banking sector. Information asymmetry, a pervasive issue across all industries, is particularly prominent in the banking sector (Levine, 2004). In the banking sector, crucial information regarding loan quality often remains obscure, permitting bankers to hide problematic loans for extended periods (Hossain, 2008). Furthermore, LLPs are one of the most notable discretionary components in every bank’s financial statement. Managers frequently use this information asymmetry to engage in discretionary behavior using LLPs. Banks’ principal-agent relationship dynamics provide a fertile ground for managers to manipulate financial results through earnings management. The PAT theory provides insights into the underlying motives driving such opportunistic actions. Our study specifically applied the PAT theory to elucidate how TO affects the practice of income smoothing in the banking industry, as outlined in section 2.3.

2.2 Loan loss provisions and income smoothing

A growing amount of research has shed light on the practice of income smoothing through LLPs. Anandarajan, Hasan, and McCarthy (2007) revealed that Australian banks employ LLPs to smooth their income and this tendency is more noticeable among listed banks than among their unlisted counterparts. Furthermore, their research found that implementing the Basel Accord has enhanced earnings management practices within banks. Kanagaretnam et al. (2003) and Danisman, Demir, and Ozili (2021) echoed these findings by documenting instances of income smoothing using LLPs within the USA banks. Moreover, this behavior was notably prevalent during times of uncertainty, as pointed out by Danisman et al. (2021). Ghosh (2007) and Vishnani, Agarwal, Agarwalla, and Gupta (2019) noted income smoothing practices among Indian banks and observed that banks listed on well-known stock exchanges use income smoothing tactics more aggressively than their unlisted counterparts.

Chen, Emanuel, Li, and Yang (2021) studied the impact of regulatory changes (Basel III) on the income smoothing behavior in Chinese banks and did not find a difference in the earnings management behavior after the regulatory change. This finding is contradicted for Russian Banks by Nikulin and Downing (2021), who found evidence of earnings management by banks before and after the regulatory changes. In the context of South Africa, Ozili and Outa (2018) found mixed evidence of income smoothing using LLPs by South African banks. According to the findings of this study, banks in South Africa do not use LLPs to engage in earnings management when they are weakly capitalized and have a huge stock of NPL. However, their study finds that South African banks engage more in income smoothing during the economic boom, given the higher capital and profitability.

Internal factors and opportunities often determine income smoothing by banks (Bouvatier et al., 2014; Ozili & Thankom, 2018). External environmental factors, such as institutional quality and macroeconomic conditions, are also argued to influence income smoothing (Doan, Lin, & Doong, 2020; Pinto, Gaio, & Gonçalves, 2020; Salem, Usman, & Ezeani, 2021). Bank managers can exercise discretion in LLPs to smooth reported earnings to reach a particular earnings threshold. LLPs are by far the most crucial accruals adjusted by banks (Beatty & Liao, 2014; Kanagaretnam et al., 2003) and are usually significant relative to equity capital and net income (Healy & Wahlen, 1999). The fundamental justifications for discretionary provisioning are the ambiguity and subjectivity involved in estimating projected losses (Biswas, Bhattacharya, Bhattacharya, & Sadarangani, 2022; DeBoskey & Jiang, 2012). The incentive to engage in income smoothing becomes stronger if managed earnings maximize shareholder wealth and managerial compensation (Ozili & Outa, 2018). Banks can hold extra reserves when profits are higher by charging more LLPs. Bank managers can charge fewer LLPs to boost profits when business is slow. We therefore hypothesize:

H1.

LLPs and profit before tax and provision are positively related in BRICS.

2.3 The role of trade openness

According to the openness hypothesis of financial development, a country’s financial sector will grow more quickly if integrated into global trade and capital markets (Rajan & Zingales, 2003). However, exposure to the world market has benefits and drawbacks. Some scholars claim that trade or economic openness could harm the banking system’s stability (Bui & Bui, 2020; Hellmann, Murdock, & Stiglitz, 2000). With increased TO, the regional economy may become more vulnerable to the global economic cycle and experience income volatility and unpredictability (Ashraf, Arshad, & Yan, 2017). TO can stimulate competition in the financial sector because increased product market demand necessitates reforms, leading to the entry of new players. Competing banks would then lower their credit standards to amass more loans in the face of such fierce competition, which raises the NPL. A rise in NPL is more likely to happen as a nation grows more open (Bui & Bui, 2020). Higher NPL adversely impacts the profitability of banks through LLPs. Therefore, as per the PAT theory, bank managers are incentivized to increase earnings by exercising their discretion over LLPs in order to maintain their compensation and avoid debt covenant violations.

On the other hand, TO brings diversification benefits. The banks can diversify their loan portfolio between international and domestic clients (Ashraf et al., 2017; Bui & Bui, 2020). Moreover, a bank’s international clients are more competent and resilient than its domestic clients, thereby increasing bank earnings (reducing NPL). As per the political cost hypothesis of the PAT theory, such higher earnings can attract social, regulatory and political scrutiny. According to Watts and Zimmerman (1978), businesses that are subject to more regulations choose accounting practices that result in lower earnings since they incur higher political costs. In highly regulated banking sectors, income-reducing accruals are likely to be used for managing earnings, particularly in state-owned banks that are common in the BRICS market. Therefore, when profits are higher, banks are incentivized to reduce earnings so that the earnings are never too high or too low (Kanagaretnam, Lobo, & Mathieu, 2004; Ozili, 2019; Ozili & Outa, 2018). This influences the market perceptions of earnings volatility, since TO can create high-income volatility. Therefore, TO can incentivize managers to either inflate or deflate earnings, in other words, TO can promote income smoothing in banks (Figure 1). In line with this argument, we extend our first hypothesis as follows.

H2.

Trade openness promotes income smoothing in banks via LLPs.

2.4 The role of International Financial Reporting Standards (IFRS)

According to Jensen (2005), “Indeed, earnings management has been considered an integral part of every top manager’s job for at least the last two decades. But when managers smooth earnings to meet market projections, they are not creating value for the firm; they are both lying and making poor decisions that destroy value…” p.(8). When earnings are managed, the financial reports do not show an accurate and fair view of the firm’s affairs. The empirical literature on financial reporting argues that the IFRS can significantly reduce earnings management in banks (e.g. Amidu & Issahaku, 2019; Leventis, Dimitropoulos, & Anandarajan, 2011; Ozili & Outa, 2019). Compliance with the IFRS is expected to ensure transparency and better financial reporting quality (Kurauone et al., 2021). Leventis et al. (2011) observed a decrease in income smoothing among European listed banks using LLPs, followed by the IFRS implementation. Amidu and Issahaku (2019) found that the IFRS implementation improves African banks’ earnings quality. Manganaris, Spathis, and Dasilas (2016) observed that the IFRS had strengthened the value relevance of conservative and non-conservative banks in European countries, which is further influenced by the institutional environment in which the banks operate.

Similarly, Agbodjo, Toumi, and Hussainey (2021) found that the IFRS improved the quality of accounting information in Islamic banks. However, there can be differences in the mandatory and voluntary adoption of the IFRS, suggesting that financial reporting quality depends on the enforcement quality of the IFRS (Van Tendeloo & Vanstraelen, 2005). Ozili and Outa (2019) found mixed evidence of the impact of IFRS on the earnings management of Nigerian banks driven by enforcement quality. Their study found evidence of income smoothing during Nigerian banks’ voluntary adoption of the IFRS. Ozili and Outa (2018) found that the IFRS implementation does not discourage earnings management in South African banks.

Although the extant literature shows mixed evidence, some scholars argue that the level of earnings management can be reduced by tightening accounting standards, thereby improving reporting quality (Ewert & Wagenhofer, 2005). The IFRS can limit the range of accounting techniques available, reducing managerial discretion (Ashbaugh & Pincus, 2001). The IFRS mandates more disclosures, reducing the information gap between inside and outside stakeholders (Leuz & Verrecchia, 2000). In addition, the IFRS mandates accounting measurements and recognition that more accurately represent a company’s underlying economic performance, providing more pertinent data for investment decisions (Barth, Landsman, & Lang, 2008). In line with these arguments, implementing the IFRS can shrink the earnings management of banks in the BRICS countries. Therefore, we extend our first hypothesis as follows.

H3.

Banks’ adoption of the IFRS reduces income smoothing via LLPs.

3. Methodology

3.1 Sample selection and data collection

Our primary data set consists of bank-level variables and is obtained from Bloomberg. The macroeconomic variables like GDP and TO for each country representing the BRICS group are obtained from the World Bank. We considered all the banks in BRICS nations for the period 2014–2020 and were constrained to remove banks with insufficient data availability. The final sample comprises 78 commercial banks from five BRICS countries and the sample banks account for more than 90% of the assets of the commercial banks identified in the Bloomberg database for BRICS (Table 1). The process for calculating the percentage of total assets can be found in the footnote of Table 1. To improve estimation efficiency, we adjust for outlier observations by winsorizing the continuous variables at the 1% and 99% levels.

3.2 Empirical methodology

To test H1, we used a fixed-effect panel regression model (Equation 1) with robust standard errors that controls for unobserved heterogeneity in the model.

(1)LLPit=β0+β1PBTPit+β2CARit+β3SIZEit+β4CHLOANit+β5CHNPLit+β6BENPLit+β7GDPt+β8GIt+εit

Table 2 shows the description of the variables.

A positive coefficient of PBTP indicates income smoothing using LLPs (Bouvatier et al., 2014; Chen et al., 2021; Danisman et al., 2021). We have included the control variables following earlier research to disentangle the non-discretionary components of LLPs from its discretionary components – capital ratio (Jin, Kanagaretnam, Liu, & Lobo, 2019; Ozili & Thankom, 2018; Ozili, 2019), bank size (e.g. Curcio & Hasan, 2015; Pandey, Tripathi, & Guhathakurta, 2022), change in loans (Aristei & Gallo, 2019; Pandey et al., 2022), NPL (Pinto et al., 2020; Soedarmono, Pramono, & Tarazi, 2017; Vishnani et al., 2019), change in NPL (Pandey et al., 2022) and per capita GDP growth (Alam et al., 2020; Miller, Moussawi, Wang, & Yang, 2021).

LLPs are also affected by regulatory capital, and prior research has found evidence of capital management using LLPs (Bryce, Dadoukis, Hall, Nguyen, & Simper, 2015; Othman & Mersni, 2014; Ozili & Outa, 2018). A bank with a lower capital ratio will keep a higher LLP by using discretion to compensate for its low regulatory capital; therefore, a negative coefficient can be expected. We have controlled the bank size because banks with more extensive business operations may maintain higher LLPs proportionate to their higher operations volume (Anandarajan, Hasan, & Lozano-Vivas, 2003; Ozili, 2019). NPL is related to the specific provisions a bank earmarks for actual losses, and when banks anticipate higher actual losses, they increase the specific provisioning component of LLPs (Ozili & Thankom, 2018). A higher starting balance of NPL will necessitate a higher LLP for the ongoing year, implying that larger loans have remained nonperforming for extended periods. An increase in the volume of NPL in the current year indicates that more LLPs are needed to process them. We have used the change in total loans to control overall credit portfolio default, and a positive relationship is expected with LLPs. An increase in total loans raises the likelihood of higher default and therefore, banks make provisions to account for this contemporaneous credit risk (Biswas et al., 2022; Ozili & Thankom, 2018). We have used the per capita real GDP growth rate to control the procyclicality and macroeconomic effects. Bank provisioning also depends on economic cycles. Banks usually lend more to boost revenue during economic booms, with the expectation that customers will be able to repay, given that businesses are doing well during these booms. Therefore, banks usually set aside less LLPs during economic booms, whereas the borrower’s repayment capacity suffers during economic downturns. As a result, banks become more prudent and charge more LLPs (Miller et al., 2021; Soedarmono et al., 2017). Therefore, similar to the extant literature, we expect an inverse link between LLPs and GDP. We have also controlled country-specific institutional differences by using country-specific governance indicators. The country-specific institutional differences are captured using the Worldwide Governance Indicators developed by Kaufmann, Kraay, and Mastruzzi (2011). Following Lassoued et al. (2018) and Quttainah, Song, and Wu (2013), we have constructed an index of the institutional quality of each country by using six governance indicators (Table 2).

To test H2, we use Equation (2) where we include a TO variable in equation (1) as an interaction term with the income-smoothing variable, i.e. PBTP. There are mainly two aspects of TO de facto and de jure (Gräbner, Heimberger, Kapeller, & Springholz, 2021; Rahman, Begum, Ashraf, & Masud, 2020). The de facto measure of TO is captured using an outcome-oriented indicator that deals with products and services imported and exported by a country and its trading partners, reflecting a country’s real level of economic integration with the rest of the globe. An economy with high TO imports and exports significant goods and services and generally actively participates in the global market. The country’s trade policies and restrictions make up the de jure aspect of TO. Reduced trade restrictions, fewer non-tariff barriers, and lower tariffs all contribute to a more open trade policy that makes moving products and services across international borders easier. The de-jure metrics (tariff and non-tariff barriers) are based on an assessment of a nation’s legal system; they reflect a nation’s willingness to be open as reflected by the current regulatory environment (Gräbner et al., 2021). We concentrated on de facto measures of TO in this research because they are expected to affect LLPs through their effect on bank credit. It is measured as the sum of exports and imports as a percentage of GDP (Ashraf et al., 2017; Alamgir Hossain, Moudud-Ul-Huq, & Kader, 2020). The model is represented as follows:

(2)LLPit=β0+β1PBTPit+β2PBTPit*TOt+β3TOt+β4CARit+β5SIZEit+β6CHLOANit+β7BENPLit+β8CHNPLit+β9GDPt+β10GIt+εit

To test our third hypothesis H3, we divided the sample into two groups: one group consisted of banks adopting the IFRS and the other with banks yet to adopt the IFRS. We tested our models on these two groups and discuss the results in sub-section 4.2. Figure 2 presents the conceptual model.

4. Results and discussion

4.1 Summary statistics and correlation

The descriptive statistics of the variables are reported in Table 3. The average LLP is 0.99% of total assets. While more than 50% of the assets comprise loans and advances, the provisioning percentage appears relatively low. This indicates that banks in BRICS have maintained a smaller volume of provisions and do not rely entirely on LLP to manage credit risk. While LLPs are higher for Brazil, Russia and India, they are substantially lower for China and South Africa, suggesting that bank provisions vary significantly among the BRICS nations. The variable PBTP has a mean of 2.16%, implying that banks in our sample earn 2.16% of total assets before making a provision for loan losses. While profits vary across banks (the maximum PBTP and minimum PBTP are 5.64% and −0.70%, respectively), the profitability of individual countries does not vary much during the sample period and remains close to the mean value. The mean CAR ratio is 11.38% for the whole of BRICS, which is more than the minimum-required ratio as per the Basel norms (6% for Tier-1 capital, and the total capital is 8%) across the BRICS nations. Although the variation in bank size is observed from the minimum and maximum values of the SIZE variable, the country-wise bank size also remains close to the average bank size (25.11) of the entire BRICS nations. South Africa has the highest TO, while Brazil has the least TO during the study period. The total loans remained more than 50% of the total assets for banks in BRICS during the sample period. Even though total loans remain high, the average change in total loans is negative for Brazil and Russia, indicating a contraction in credit supply by most banks in these two economies during our study period. The mean value of beginning NPLs is 1.77% of total assets, and NPL varies significantly across the banks. Russia and India have a higher volume of beginning NPL than Brazil, China and South Africa. The overall change in NPL is positive, indicating the need for more provisioning across the banks. However, the change in NPL is negative for Brazil and Russia. This indicates a contraction of credit supply, which is evident in the negative change in total loans for these two economies. The high standard deviation of the GDP growth rate indicates this region’s diverse and volatile macroeconomic condition.

As per the correlation matrix reported in Table 4, the main explanatory variable, PBTP, exhibits a positive and statistically significant (at 1%) coefficient with LLP. The issue of multicollinearity is mitigated since none of the correlation coefficients between the explanatory variables exceed 0.50 (Van Anh, 2022). The variance inflation factor (VIF) for the explanatory variables is much below the commonly used threshold of 10, indicating no multicollinearity problem (Alkebsee, Habib, & Li, 2023).

4.2 Regression results and discussion

The choice of a fixed-effect estimator is supported by the Hausman test (Table 5). The significant income-smoothing variable (PBTP) suggests that banks manage earnings using LLPs across BRICS. When profit before provision is higher, banks smooth earnings by keeping excess reserves to meet earnings expectations when profits are below these expectations. The income-smoothing hypothesis is consistent with previous studies (e.g. Chen et al., 2021; Hou et al., 2021; Nikulin & Downing, 2021). The significant negative coefficient of TO implies that higher TO reduces bank risk-taking and, therefore, banks charge less LLPs. This finding is consistent with Rahman et al. (2020) in the context of BRICS. The positive and significant coefficient of the interaction term PBTP*TO indicates that TO encourages earnings management through income smoothing in banks across BRICS. In other words, TO reduces earnings quality by promoting earnings management via LLPs.

The negative and significant capital management variable CAR shows that banks raise LLPs to meet overall capital requirements when capital is lower. This confirms the capital management hypothesis and is consistent with previous studies (e.g. Ghosh, 2007; Ozili & Outa, 2018). The loan growth or change in loans (CHLOAN) is significant and negative, implying that incremental loans are of improved quality and therefore, banks charge less LLPs. This is consistent with the findings of Ozili (2019). Another possible reason for this can be the contraction of credit by banks, which is evident in our sample (e.g. Brazil and Russia). The reduced credit supply will attract fewer provisions. Although the change in loans is negative in the second model but remains statistically insignificant, the beginning NPL and change in NPL are positively significant in the two models, suggesting that the higher the NPL ratio, the greater the LLPs. Similarly, a positive change in NPL indicates that more loans have entered the NPL category and therefore, more provisions are required. This finding is consistent with those of Ozili and Outa (2018) and Pandey et al. (2022).

The GDP growth rate is significant and negative, confirming LLPs’ procyclical nature. When the economy is flourishing, banks keep fewer LLPs and vice-versa. In other words, banks are reluctant to keep sufficient provisions before economic recessions, which may result in credit contraction during economic downturns. This finding is consistent with some studies within emerging economies (e.g. Ghosh, 2007; Nikulin & Downing, 2021; Ozili & Outa, 2018). Consistent with Chen et al. (2021) and Ghosh (2007), the SIZE variable is insignificant in both models, which shows that bank size is not a prominent factor of LLPs in BRICS and other factors play a more significant role. In addition, the variation in the size of banks in our sample is minimal. The country-specific governance index is positive and statistically significant in the base model. The positive coefficient of GI suggests that better institutional quality makes banks more prudent in risk-taking. However, this finding remains unclear because we did not find a consistent result in the second model. We argue that the existing institutional quality becomes insufficient in the presence of greater TO, making banks less prudent in lending.

We found no discernible differences between banks which adopted reporting the IFRS and those that did not (Columns 3 and 4 of Table 5), suggesting that either the enhanced flexibility allowed by the IFRS or the lax application of IFRS in BRICS allows banks to employ LLPs opportunistically to smooth their reported earnings. This finding adds to the empirical literature, where similar observations are made for South African banks (Ozili & Outa, 2018).

4.3 Endogeneity and robustness check

A strand of research suggests that LLPs are dynamic, and GMM has been used to account for the endogeneity problem (Bouvatier et al., 2014; Danisman et al., 2021; Soedarmono et al., 2017). System-GMM is considered more efficient than standard GMM (Baltagi, 2008; Salem et al., 2021). GMM estimators are advantageous because they are suitable for addressing any potential bias in a dynamic panel (Salem et al., 2021). GMM estimators can address several critical econometric issues such as the endogeneity of explanatory variables, cross-section-specific unobserved heterogeneity, an autoregressive process in the data, autocorrelation and heteroskedasticity within groups. In order to control for the dynamic behavior of LLPs, we have used a two-step system-GMM approach, and our results remain robust across the models (Table 6).

We have estimated the two-step system-GMM models with Windmeijer’s finite sample correction to avoid downward biased standard errors (see Table 6 for the results). LLPs seem to be affected by their past values. Since GMM uses instruments in the model, the consistency of the GMM estimator depends on the validity of the instruments. We used the Hansen J-statistic to find that the instruments are valid. Besides instrument validity, another necessary condition for the consistency of the GMM estimator is that the error terms should be free from the second-order serial correlation. The Arellano and Bond serial correlation test (AR (2)) indicates the absence of a second-order serial correlation (Table 6). In addition to GMM, we have applied another alternate econometric specification to check the consistency of the finding. In banking study, a possible cross-section dependency may arise because of the inter-connectedness of banks through inter-bank loans and common asset base. Although the cross-section dependency is not a severe problem in a micropanel with a large cross-section and fewer periods (Baltagi, 2008), we have further tested our base models with Driscoll–Kraay standard errors approach to deal with cross section dependency and our results (untabulated) remain consistent with fixed effect and GMM estimators.

Furthermore, all key results are robust to the replacement of the TO variable. As an alternative proxy of TO measure, the study has used a simple mean of tariff rate applied on all goods in a country (% of GDP). A higher tariff value indicates lower TO because a country’s more restrictive trade policies are indicated by greater tariffs placed on its products (Rahman et al., 2020; Yanikkaya, 2003). Therefore, a negative relationship is expected between tariff rate and income smoothing. The variable tariff interacts with the income smoothing variable PBTP and the study finds a statistically significant negative coefficient (Table 7). This indicates that when TO is higher (lower tariff rate), the possibility of earnings management is also higher and vice-versa. This finding is consistent with the main proxy of TO.

4.4 Sensitivity test

Using the sensitivity tests, we address the concern about the overrepresentation of certain countries in our sample. First, we excluded banks from China and then reran the analysis with banks from BRIS (Brazil, Russia, India and South Africa). Second, we removed banks from India and then reran the analysis with banks from BRCS (Brazil, Russia, China and South Africa). Our inferences remain unchanged if our analysis is repeated using the sample as mentioned above since the coefficients of PBTP and PBTP*TO remain statistically significant (Table 8).

4.5 Additional analysis

The institutional environment (e.g. regulatory quality, political stability, the rule of law and control of corruption) can significantly affect managerial discretionary and opportunistic behavior (An, Li, & Yu, 2016; Lassoued et al., 2018; Quttainah et al., 2013). We extended the analysis to explore whether stronger institutional quality reduces earnings management via LLPs in banks in BRICS. We created a dummy variable HGI, which equals 1 if the institutional quality (GI) is above the median value and 0 otherwise. The significant and negative coefficient of PBTP*HGI indicates that the stronger the institutional quality, the less income smoothing (Table 9). The stronger institutional quality enhances the effective implementation of IFRS in BRICS, which ultimately discourages the opportunistic behavior of bank managers. This result does not hold in the case of banks without IFRS, suggesting that IFRS and institutional quality complement each other in BRICS. The positive and significant coefficient of PBTP*TO for banks disclosing reports under the IFRS (Table 9) shows that TO promotes a more significant income smoothing in banks that report under the IFRS than local accounting standards.

4.6 Path analysis

Path analysis is used to analyze the mechanism through which TO impacts bank earnings management. For this purpose, we segregated the discretionary component of the LLPs from its non-discretionary components. The income smoothing results in discretionary LLPs (DLLPs), and we have used the LLPs specification from Equation (1) to calculate the DLLPs. The discretionary portion (DLLPs) is captured in the error terms of equation (1).

Since we argued that TO impacts bank earnings management through NPL, we used NPL (scaled by total loans) as a mediator. The results of the path analysis show that the indirect effect (a x b) is significant while the direct effect (c’) is insignificant (Figure 3). Therefore, there is indirect only mediation via the mediator (full mediation). This means that TO impacts earnings management (DLLPs) through NPL.

5. Conclusion

With an improved focus on responsible and sustainable banking practices, bankers have the challenge of balancing the interests of their shareholders, analysts, community and other stakeholders. Sound loan loss provisioning practices are an integral part of the credit risk assessment and valuation process, but subjective estimates allow bankers to use them for income smoothing. The study brings insights into the income smoothing practices of the BRICS banking sector and the role of TO and IFRS in influencing such smoothing practices. The study has used the primary discretionary accruals of banks, i.e. LLPs to investigate the income smoothing practices and the sum of exports and imports as a percentage of GDP to measure the TO of countries. The findings of the study suggest the existence of income smoothing via LLPs in the BRICS banking sector and TO promotes such earnings smoothing, which is evident even in the presence of IFRS. The study’s findings reveal that implementing the IFRS does not discourage income smoothing. However, implementing the IFRS can effectively mitigate earnings smoothing only in a strong institutional environment. The possible explanation to the findings lies in the presence of institutional voids in BRICS. We provide novel results to the bank earnings management literature and highlight that income smoothing in banks is encouraged by increased TO in emerging countries like BRICS and results are robust to a battery of additional tests and are consistent with the existing literature. Businesses in emerging countries are generally bank financed and such countries are known to have institutional voids (Heeks et al., 2021). The presence of such voids helps explain the need for a strong institutional environment for the IFRS adoption to reduce earnings management in banks in BRICS. This has important policy implications for nations such as India and China, where the banking sector is yet to fully integrate IFRS. Before implementing IFRS standards, governments in India and China must improve the current institutional framework to ensure effective enforcement of these standards. While Brazil, Russia, and South Africa have successfully embraced IFRS, the existing empirical literature does not provide a definite view on the usefulness of IFRS in mitigating EM in these countries. The implementation of IFRS 9, which is based on the expected credit loss (ECL) model, gives managers additional leeway in estimating possible loan losses. Indian banking regulator (Reserve Bank of India) has recently shown interest in implementing the ECL model for LLPs. The findings support strengthening each country’s institutional environment, including enforcement and limiting the extent of managerial use of forward-looking information.

Our findings provide several important practical implications for regulators and policymakers. Earnings management practices can undermine regulatory efforts to maintain a stable and transparent financial system as regulators rely on accurate financial reporting to monitor the financial health of banks and ensure compliance with regulations. The findings support that BRICS banks undermine credit risk management by employing LLPs for earnings management. In order to reduce income smoothing for opportunistic purposes, regulators across the BRICS should explore alternative provisioning criteria beyond LLPs to assess credit risk. Through the banking system’s stability, the positive effect of TO on income smoothing has consequences for the nation’s financial stability. BRICS contributes around 18% of the global trade and banks play a crucial role in financing and facilitating cross-border trade. Therefore, ensuring the integrity of the banking system is crucial for economic stability. Our results suggest that macro-prudential regulators and banking supervisors should work closely to ensure that all TO decisions are made under improved discipline and transparency in provisioning, better institutional quality and regulatory support that can promote greater bank stability.

Figures

Figure 1

Relationship between trade openness and income smoothing (earnings management)

Figure 2

Conceptual model

Figure 3

Path diagram

Table 1

Sample distribution

Country	No. of banks (our sample)	No. of obs	No. of banks in Bloomberg*	Percent of total assets*
China	38	266	60	95.16%
India	21	147	34	86.22%
Brazil	9	63	18	98.24%
Russia	6	42	25	89.88%
South Africa	4	28	7	86.86%
Total	78	546	144	Mean = 91.27%
				Median = 89.88%

Note(s): ^*Similar to Bouvatier et al. (2014), we calculated the percentage of total assets of sample banks. Percent of total assets = total assets of our sample banks in 2020/total assets of banks identified in the Bloomberg for BRICS in 2020. We have excluded the banks whose total assets data is missing for the year 2020 while calculating the total number of banks in Bloomberg. Since Bloomberg shows a very small number of South African banks, we validated the adequate sample representation from other available reports. All the values are converted to the common denomination of USD

Source(s): Authors’ own creation

Table 2

Description of variables

Variable	Abbreviation	Measurement
Loan loss provisions	LLP	Loan loss provisions to total assets
Profit before tax and provisions	PBTP	Profit before tax and provisions to total assets
Trade openness	TO	Sum of export and import as a percentage of GDP
Capital adequacy ratio	CAR	Tier-1 common equity to risk weighted assets
Bank size	SIZE	Natural log of total assets
Change in total loans	CHLOAN	Change in total loans to total assets
Beginning nonperforming loans	BENPL	Beginning nonperforming loans to total assets
Change in nonperforming loans	CHNPL	Change in nonperforming loans to total assets
Gross domestic product growth rate	GDP	Per Capita GDP growth rate
Country-specific governance index	GI	Average of six governance indicators (government effectiveness, rule of law, political stability, voice and accountability, regulatory quality and control of corruption)

Source(s): Authors’ own creation

Table 3

Summary statistics

		Panel A: Summary statistics of variables for BRICS
	LLP	PBTP	CAR	SIZE	CHLOAN	BENPL	CHNPL	GDP	GI	TO
Mean	0.0099	0.0216	0.1138	25.1101	0.0283	0.0177	0.0012	3.5044	0.5162	39.5308
Std. Dev	0.0089	0.0098	0.0219	1.8649	0.0757	0.0228	0.0099	4.0811	0.1418	6.8622
Min	0.0003	−0.0071	0.0693	20.7334	−0.2884	0.0017	−0.0465	−8.1649	0.0828	24.3197
Max	0.0486	0.0564	0.1750	29.2619	0.1695	0.1393	0.0366	7.0822	0.9081	59.4995

		Panel B: Country-wise mean of variables
Country	LLP	PBTP	CAR	SIZE	CHLOAN	BENPL	CHNPL	GDP	GI	TO
Brazil	0.0191	0.0316	0.1337	24.1475	−0.0334	0.0217	−0.0031	−1.6235	0.5460	27.3070
Russia	0.0132	0.0272	0.1145	23.8035	−0.0343	0.0306	−0.0028	0.0876	0.1750	48.1895
India	0.0129	0.0199	0.1130	24.1849	0.0229	0.0329	0.0043	3.7131	0.5380	41.8649
China	0.0058	0.0188	0.1076	26.0336	0.0579	0.0058	0.0010	5.6649	0.5150	38.1344
South Africa	0.0061	0.0251	0.1290	25.3239	0.0079	0.0216	0.0021	−1.4527	0.8620	55.0583

Note(s): TO and GDP growth rate are in percentage

Source(s): Authors’ own creation

Table 4

Correlation matrix

	LLP	PBTP	CAR	SIZE	CHLOAN	BENPL	CHNPL	TO	IFRS	GI	GDP
LLP	1.0000
PBTP	0.2766***	1.0000
CAR	−0.1189**	0.3811***	1.0000
SIZE	−0.2484***	−0.0535	−0.0538	1.0000
CHLOAN	−0.4228***	0.0539	0.0308	0.2382***	1.0000
BENPL	0.6209***	−0.0883**	−0.1421***	−0.2128***	−0.4786***	1.0000
CHNPL	−0.0089	−0.0161	−0.0899**	0.0097	0.2736***	−0.2571***	1.0000
TO	−0.2253***	−0.0994**	−0.1230***	−0.0733*	−0.0130	0.0937**	0.0807*	1.0000
IFRS	−0.1627***	0.0385	−0.0740*	0.3093***	0.0800*	−0.3917***	−0.1944***	0.1083**
GI	−0.0205	−0.0788*	0.1831***	0.1491***	0.1563***	0.0291	0.0662	−0.1333***	−0.1009**	1.0000
GDP	−0.4092***	−0.2500***	−0.3173***	0.2125***	0.4229***	−0.3631***	0.2484***	0.1290***	0.1258***	−0.1809***	1.0000

Note(s): ***p < 0.01, **p < 0.05 and *p < 0.1

Source(s): Authors’ own creation

Table 5

Regression results

	(1)	(2)	(3)	(4)
	FE	FE	FE (banks with IFRS)	FE (banks without IFRS)
PBTP	0.1673**	0.1886***	0.1597***	0.1618**
	(0.0763)	(0.0521)	(0.0548)	(0.0758)
PBTP*TO		0.0178***
		(0.0056)
TO		−0.0005**
		(0.0002)
SIZE	−0.0013	−0.0008	−0.0028	−0.0005
	(0.0008)	(0.0007)	(0.0036)	(0.0009)
CAR	−0.0737**	−0.0588*	−0.0868***	−0.0663
	(0.0311)	(0.0295)	(0.0229)	(0.0419)
CHLOAN	−0.0134**	−0.0102	−0.0146**	−0.0121*
	(0.0061)	(0.0068)	(0.0058)	(0.0063)
BENPL	0.2326***	0.2432***	0.1889	0.2625***
	(0.0248)	(0.0258)	(0.1114)	(0.0283)
CHNPL	0.2233***	0.2325***	0.1977**	0.2037***
	(0.0518)	(0.0525)	(0.0928)	(0.0735)
GDP	−0.0004***	−0.0003**	−0.0015***	−0.0002
	(0.0001)	(0.0001)	(0.0003)	(0.0001)
GI	0.0120*	−0.0030	0.0037	0.0235*
	(0.0067)	(0.0086)	(0.0068)	(0.0138)
_cons	0.0387*	0.0396*	0.0890	0.0126
	(0.0202)	(0.0204)	(0.0904)	(0.0230)
Observations	546	546	224	322
R-squared	0.4824	0.5237	0.4350	0.6282
Firm-fixed effects	Yes	Yes	Yes	Yes
Time-fixed effects	Yes	Yes	Yes	Yes
F-stat	21.15***	24.93***	7.37***	48.75***
Hausman test p-value	0.0503	0.0002	0.0856	0.0019

Note(s): All the variables are defined in Table 2. Robust standard errors are in parentheses. ***p < 0.01, **p < 0.05 and *p < 0.1

Source(s): Authors’ own creation

Table 6

Robustness check

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	FE (with lagged LLP)	FE (with lagged LLP)	GMM	GMM	FE-lagged LLP (banks with IFRS)	FE-lagged LLP (banks without IFRS)	GMM (banks with IFRS)	GMM (banks without IFRS)
LLPit−1	0.2284***	0.1806***	0.2324**	0.2041**	0.1665***	0.2678***	0.1871**	0.4714***
	(0.047)	(0.0658)	(0.0975)	(0.0956)	(0.0379)	(0.0925)	(0.0891)	(0.1223)
PBTP	0.1615**	0.2033***	0.2638***	0.2864***	0.1569*	0.2135***	0.2007***	0.2878***
	(0.0645)	(0.0423)	(0.0477)	(0.0472)	(0.0903)	(0.0525)	(0.0627)	(0.0595)
PBTP*TO		0.0168**		0.0073*
		(0.0072)		(0.0044)
TO		−0.0005*		−0.0001
		(0.0003)		(0.0001)
_cons	0.0669***	0.0544**	0.0136**	0.0192***	0.1702	0.058**	0.0074	0.0018***
	(0.0205)	(0.0228)	(0.0057)	(0.0067)	(0.1264)	(0.0274)	(0.0088)	(0.0044)
Observations	468	468	468	468	192	276	192	276
R-squared	0.5021	0.5332	–	–	0.4521	0.6847	–	–
Control variables	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Firm-fixed effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Time-fixed effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Country-fixed effects	No	No	Yes	Yes	No	No	Yes	Yes
F-stat	27.28***	36.89***	254.60***	173.00***	89.13***	83.20***	517.21***	310.12***
Hausman test p-value	0.0000	0.0000	–	–	0.0000	0.0000	–	–
AR (2) p-value	–	–	0.7460	0.5740	–	–	0.8520	0.2910
Hansen test p-value	–	–	0.2560	0.3130	–	–	0.1470	0.2730

Note(s): Consistent with Bikker and Metzemakers (2005) and Ozili and Thankom (2018), the authors estimated the FE with lagged dependent variable to account for banks’ dynamic adjustments to LLPs in expectation of predicted losses on the bank loan portfolio. Our results remain unchanged as the coefficients of PBTP and PBTP*TO are positive and statistically significant. Robust standard errors are in parentheses. ***p < 0.01, **p < 0.05 and *p < 0.1

Source(s): Authors’ own creation

Table 7

Alternative proxy of trade openness

	(1)	(2)
	BRICS	BRICS
LLPit−1		0.1795***
		(0.0479)
PBTP	0.2209***	0.1970***
	(0.0538)	(0.0542)
PBTP*Tariff	−4.5551***	−3.3110**
	(1.1597)	(0.9883)
Tariff	−0.0943***	−0.0841***
	(0.0208)	(0.0226)
_cons	0.0340*	0.0642***
	(0.0203)	(0.0218)
Observations	546	468
R-squared	0.5168	0.5263
Control variables	Yes	Yes
Firm-fixed effects	Yes	Yes
Time-fixed effects	Yes	Yes
F-stats	29.00***	42.02***

Note(s): ***p < 0.01, **p < 0.05 and *p < 0.1

Source(s): Authors’ own creation

Table 8

Sensitivity test results

	(1)	(2)	(3)	(4)
	BRIS	BRIS	BRCS	BRCS
PBTP	0.1507*	0.1916***	0.1409**	0.1372**
	(0.0782)	(0.0497)	(0.0662)	(0.0565)
PBTP*TO		0.0141**		0.0171***
		(0.0053)		(0.0046)
TO		−0.0003		−0.0008***
		(0.0002)		(0.0003)
_cons	0.0536*	0.0455	0.0268	0.0939
	(0.0287)	(0.0283)	(0.0624)	(0.0755)
Observations	280	280	399	399
R-squared	0.517	0.5393	0.3238	0.4209
Control Variables	Yes	Yes	Yes	Yes
Firm-fixed effects	Yes	Yes	Yes	Yes
Time-fixed effects	Yes	Yes	Yes	Yes
F-stats	26.29***	28.01***	4.44***	24.40***

Note(s): The authors avoided the GMM and FE with lagged dependent variable due to an insufficient number of observations in each sub group as our estimator breaks. In addition, we showed that our findings are robust using alternative econometric specifications. The authors avoided sensitivity tests for groups (with and without IFRS) to avoid the limitation of a very small number of observations. Robust standard errors are in parentheses. ***p < 0.01, **p < 0.05 and *p < 0.1

Source(s): Authors’ own creation

Table 9

Results of additional analysis

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	FE (banks with IFRS)	FE-lagged LLP (banks with IFRS)	FE (banks with IFRS)	FE-lagged LLP (banks with IFRS)	FE (banks without IFRS)	FE-lagged LLP (banks without IFRS)	FE (banks without IFRS)	FE-lagged LLP (banks without IFRS)
LLPit−1		0.1033***		0.1100**		0.2408***		0.2643***
		(0.0364)		(0.0513)		(0.0882)		(0.0980)
PBTP	0.1761***	0.2141**	0.1298**	0.1897***	0.0606	0.1369***	0.1870**	0.2329***
	(0.0593)	(0.0905)	(0.0519)	(0.0505)	(0.0748)	(0.0465)	(0.0747)	(0.0795)
HGI	0.0108**	0.0131**			−0.0074***	−0.0075***
	(0.0044)	(0.0052)			(0.0023)	(0.0016)
PBTP*HGI	−0.1597***	−0.2505***			0.2154***	0.1530**
	(0.0380)	(0.0632)			(0.0762)	(0.0652)
TO			−0.0007**	−0.0003			−0.0002	−0.0002
			(0.0003)	(0.0004)			(0.0003)	(0.0002)
PBTP*TO			0.0172***	0.0183***			0.0052	0.0041
			(0.0031)	(0.0038)			(0.0104)	(0.0087)
_cons	0.0988	0.1350	0.1255	0.0973	0.0258*	0.0492**	0.0165	0.0599**
	(0.1020)	(0.1338)	(0.1206)	(0.1588)	(0.0153)	(0.0228)	(0.0234)	(0.0287)
Observations	224	192	224	192	322	276	322	276
R-squared	0.4849	0.5342	0.5075	0.5051	0.6519	0.7098	0.6304	0.6859
Control variables	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Firm-fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Time-fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes

Note(s): Robust standard errors are reported in parentheses. ***p < 0.01, **p < 0.05 and *p < 0.1

Source(s): Authors’ own creation

Notes

1. More details can be found at https://www.ifrs.org/

2. BRICS stands for “Brazil, Russia, India, China and South Africa.” Together, these five emerging economies “represent about 42% of the world’s population, 23% of gross domestic product (GDP), 30% of the territory and 18% of global trade” (Uzma & Nurunnabi, 2021).

3. The IFRS was implemented in Brazil in 2010, in Russia in 2012 and in South Africa in 2005. India decided to converge its accounting standards with the IFRS in a phased manner from 2016. Similarly, Chinese Accounting Standards for Business Enterprises (ASBEs), issued in 2006, shows China’s convergence efforts toward IFRS but India and China are yet to fully converge with IFRS. The application of IFRS to the banking sector aligns with its implementation timeline in Brazil, Russia and South Africa. Unlike China, India has yet to implement the IFRS-converged IndAS within its banking sector.