The Impact of Basel III on Bank Profitability in the EU: A Study of Bank Size Differences Written by: Rebecka Messner (000110) and Meryem Özkan (981209) Spring 2025 Supervisor: Henrik Petri Master thesis in Finance, 30 credits Graduate School, School of Business, Economics, and Law, University of Gothenburg Keywords: Basel III, Banks, EU, Profitability, GDP Growth, Capital Requirements Abstract The global financial crisis in 2007-2008 highlighted significant weaknesses in banking regulation and triggered the development of Basel III, a comprehensive framework aimed at enhancing financial stability through regulations such as capital and liquidity standards. While its stabilizing objectives are clear, the implications of Basel III on bank profitability, particularly across different bank sizes, remain debated. This thesis investigates the impact of Basel III on the profitability of banks in the European Union from 2010 to 2023, with a specific focus on differences between large and small institutions. Using panel data from 500 EU banks, the study applies a fixed effects (FE) regression and a Difference-in-Differences (DiD) approach to assess how regulatory changes have influenced return on assets (ROA) while controlling for macroeconomic and bank specific variables. The results indicate that Basel III has had a positive effect on overall bank profitability. However, the profitability effects were more significant for smaller banks compared to larger institutions. Acknowledgements This study is a master degree project in Finance at the School of business, Economics and Law at the University of Gothenburg. We are deeply grateful to the university for providing the opportunity to write and accomplish this thesis. The opportunity to engage in a project of this scope and relevance has been both enriching and rewarding. We would like to extend our sincere thanks to our supervisor, Henrik Petri, whose expertise and guidance played an important role throughout the development of this thesis. Table of Contents 1. Introduction 1 1.1 Introduction 1 1.2 Disposition 3 2. Background 4 2.1 Basel 4 2.2 The EU 5 2.2.1 The EU and Basel III 5 3. Literature review 6 3.1 The Basel III Framework 6 3.2 Bank Profitability 7 3.2.1 Macroeconomic Factors and Bank Profitability 7 3.3.2 Basel III’s Effect on Bank Profitability 8 3.3.3 Basel III’s Effect on Profitability Depending on Bank size 9 4. Data and Methodology 11 4.1 Data 11 4.2 Variables 12 4.2.1 Dependent Variable 12 4.2.2 Independent Variables 12 4.2.3 Control Variables 13 4.3 Regression Model 15 4.4 Empirical Model 15 4.4.1 Correlation 15 4.4.2 Model description 16 4.4.3 Heteroscedasticity and Serial Correlation 17 4.4.4 Difference in Difference (DiD) 18 5. Results 19 5.1 Results for Regression Model 1 19 5.2 Results for Regression Model 2 20 5.3 Robustness test 21 6. Analysis and Discussion 23 6.1 Regression Model 1 23 6.2 Regression Model 2 24 7. Conclusion 26 7.1 Limitations and Further Research 27 8. References 28 9. Appendix 36 1. Introduction 1.1 Introduction Economic crises have repeatedly disrupted global markets over the past decades, affecting multiple sectors. Financial crises usually increase demand for immediate and thorough policy actions, leading to major reforms in financial regulations and fiscal policy (Claessens & Kose, 2013). In response, the Basel Committee on Banking Supervision (BCBS), the leading international body for banking oversight, develops regulatory standards and fosters cooperation among financial regulators worldwide in order to strengthen financial stability (BIS, 2018; BIS, n.d.-a). The introduction of Basel III was a direct response to the 2007–2008 financial crisis. The framework includes, among other things, higher capital requirements, stricter leverage ratios, and enhanced liquidity regulations (BIS, 2017). The framework aims to improve the resilience of banks and reduce systemic risks (BIS, 2011), however, the impact of the Basel III framework on bank performance, particularly profitability, remains subject to ongoing academic debate. Existing literature offers differing perspectives on how Basel III has influenced bank profitability. Several studies highlight positive effects, suggesting that stricter capital and liquidity requirements have contributed to increased financial stability and enhanced profitability. For instance, Tipu (2019) and Laporšek et al. (2024) report a positive association between Basel III capital and liquidity buffers and return on assets (ROA), indicating that well capitalized and liquid banks tend to have higher profitability. However, the findings are not consistent across the literature. Le et al. (2023) and Rokiyanto and Siahaan (2024) point to potential negative effects of Basel III, noting that increased capital and liquidity requirements may constrain bank profitability. In addition, earlier literature has explored whether bank size plays a role in how profitability is affected by Basel III. Some evidence suggests that larger banks experience more favorable profitability outcomes under Basel III (Gržeta et al., 2023), while others report that smaller banks may benefit more or that the relationship remains ambiguous (Yamin et al., 2025; Laporšek et al., 2025). These mixed findings highlight the need for further empirical analysis, particularly within a European context. This paper seeks to expand the analysis of the effects of Basel III on the financial performance of banks, with a particular focus on profitability and bank size. The 1 implementation of Basel III has not yet been completed globally, with different regions adopting the framework at varying timelines. As a result, the majority of previous research has primarily examined the effects of Basel II rather than the full implications of Basel III. Studies that have analyzed Basel III impact, such as Tipu (2019), Vousinas (2015) and Rokiyanto and Siahaan (2024), have focused on banks in countries such as Pakistan, Greece and Indonesia. While some research has explored its effects within Europe, studies such as Laporšek et al. (2024) and Gržeta et al. (2023) have primarily concentrated on the early to mid years of the Basel III implementation. Given that the regulatory framework continues to evolve, further research is interesting to assess its long-term impact on European banks, particularly in the later stages of implementation. In addition, a key aspect of this study is the comparison between banks of different sizes to determine if there is a difference in the effect of Basel III on profitability depending on bank size. This study aims to contribute to this growing field by investigating: 1. How does Basel III affect the profitability of banks within the EU? 2. Does Basel III have a differentiated impact on banks profitability based on their size? Using panel data from European banks between 2010 and 2023, this research employs quantitative methods to assess the financial consequences of Basel III across different bank sizes. By providing updated insights, this study seeks to fill gaps in existing literature and offer a deeper understanding of how Basel III continues to shape the European banking sector. We expect that Basel III will have a positive effect on bank profitability, supported by studies showing improved performance from the stronger capital and liquidity standards (Tipu, 2019; Rokiyanto & Siahaan, 2024; Le et al., 2023). We also expect the impact to differ depending on bank size. Prior research indicates that the relationship between Basel III and profitability varies across banks of different sizes (Yamin et al., 2025; Laporšek et al., 2024; Gržeta et al., 2023). The results of this study indicate that Basel III contributes positively to the profitability of banks within the EU. At the same time, the effect is not consistent across institutions of different sizes. Smaller banks appear to benefit more from the regulatory changes, while the 2 profitability gains for larger banks are more limited. These findings point to a size-dependent effect of Basel III within the European banking sector. 1.2 Disposition This thesis is structured as follows: The first section introduces the research topic, the purpose, expectations and the main findings. It also presents the research questions. The second section provides context by explaining the evolution of the Basel frameworks. It also includes background on the EU and its connection to Basel III. Previous research on the subject is reviewed in the third section. The fourth section outlines the methodology and data. The fifth section presents the results of the thesis and an analysis of the result is displayed in the sixth section. Lastly, the final section concludes the thesis and suggests areas for future research. The reference list and appendix are located at the end of the paper. 3 2. Background 2.1 Basel In 1974, the Basel Committee was founded in response to disruptions in the global currency and banking markets, particularly influenced by the collapse of Bankhaus Herstatt in West Germany. The main purpose of the committee was to promote financial stability by enhancing global banking oversight. Over the years, it has developed key international banking regulations, most notably the Basel Accords on capital adequacy, known as Basel I, Basel II, and Basel III. (BIS, n.d.-a) Basel I was introduced in 1988 by the Basel Committee to establish a standardized framework for capital adequacy requirements in internationally active banks. Basel I required, among other things, banks to maintain a minimum capital ratio of 8% against their risk-weighted assets (RWA) by the end of 1992. Building upon Basel I, Basel II was proposed in 1999 and released in 2004 as a capital framework that better reflected the evolving financial landscape. Basel II was introduced as three key pillars (BIS, n.d.-a). The first pillar retained the 8% minimum capital requirement. However, it refined the risk-weighting system, allowing for other risk assessments approaches. The second pillar establishes a supervisory framework for regulators to evaluate a bank’s internal risk management and capital adequacy, ensuring that banks identify, assess, and manage risks effectively. The third pillar aims to enhance market discipline by requiring banks to disclose key information about their risk exposures, capital adequacy, and risk management practices (BIS, 2004). In 2013, Basel III was implemented as a response to the financial crisis in 2008, addressing weaknesses in the Basel II framework that were exposed during the crisis. Basel III introduced stricter capital requirements, enhanced risk coverage, and improved banking regulations to strengthen the resilience of the global financial system. The majority of the framework was phased in between 2013 and 2019, allowing banks time to adjust to the new capital and liquidity requirements. In 2017, the Basel Committee introduced final revisions, further improving the standardization of RWA and imposing stricter controls on banks' internal models. (BIS, n.d.-a) 4 2.2 The EU The European Union (EU) was established after World War II, in 1951, by six European countries as an economic partnership. Over time, it has expanded to include 27 member states that collaborate on political and economic matters (European Commission, n.d.-a). Member states of the EU are obligated to adhere to EU laws and regulations as outlined in the founding treaties, which establish the legal framework for cooperation and decision-making within the union. These treaties, agreed upon by all member states, set out which areas the EU has the authority to regulate and how its laws apply to member states (European Commission, n.d.-b). For instance, EU regulations are directly applicable and binding in all member states. Directives, on the other hand, set out objectives that member states must achieve however it allows them the flexibility to determine how to implement these objectives within their national legal systems. The union also has recommendations although these are not binding (European Commission, n.d.-c). 2.2.1 The EU and Basel III The EU has incorporated the Basel frameworks into its regulatory framework through legislation, primarily presented in the Capital Requirements Regulation (CRR) and the Capital Requirements Directive (CRD). The union has progressively been implementing Basel III since 2013. Over time, liquidity requirements such as liquidity coverage ratio has also been implemented (European Banking Authority, n.d-a). In 2021, the European Commission proposed adjustments to CRR and CRD to fully integrate the final Basel III reforms presented in 2017 (European Commission, 2021.). The regulatory requirements will be applicable to all banks within the EU (European Commission, 2024). However, like all Basel Committee standards, Basel III sets minimum requirements that are expected to be implemented by member jurisdictions within the timeframe set by the Committee (BIS, n.d.-d). 5 3. Literature review 3.1 The Basel III Framework Under Basel II, banks were required to maintain a minimum Total Capital ratio of 8% of their RWA. This Total Capital ratio consisted of Tier 1, Tier 2 and Tier 3 (BIS, 2006). In the capital requirements component of Basel III (BIS, n.d.-c), Common Equity Tier 1 (CET1), which absorbs losses immediately when they occur, was introduced as a separate requirement and set to 4.5% of the bank’s RWA. Additionally, Tier 1 capital, which under Basel III includes both CET1 and the Additional Tier 1 (AT1) capital, was raised from 4% under Basel II to 6% (Gatzert & Wesker, 2012). Tier 2 capital serves to absorb losses in the event of a bank failure, complements Tier 1 capital to constitute the Total Capital ratio. Under the Basel III framework, this ratio remains at 8% of RWA, consistent with the requirement established under Basel II (BIS, n.d.-c). Basel III also introduced The Leverage Ratio (LR) that includes a non-risk based minimum capital requirement of 3% to prevent banks from becoming excessively leveraged, regardless of how risky their assets are (BIS, 2014). In addition to the minimum capital requirements, Basel III introduced the Capital Conservation Buffer and the Countercyclical Capital Buffer. The Countercyclical Capital Buffer, ranging from 0 to 2.5% of RWA is intended to increase bank resilience during periods of excessive credit growth and must be met using CET1 or other fully loss absorbing capital. The Capital Conservation Buffer is set at 2.5% of RWA and requires banks to build up capital in normal times to ensure that losses can be absorbed in times of stress (BIS, 2011). Furthermore, two new liquidity requirements were introduced, namely the Liquidity Coverage Ratio (LCR) and the Net Stable Funding Ratio (NSFR). The LCR ensures that banks are resilient to short-term distresses by ensuring that banks have high-quality liquid assets that easily can be converted into cash and meet the liquidity needs for a 30 calendar day distress (BIS, 2013). The NSFR is designed to promote a stable long-term funding profile by requiring banks to maintain a stable funding structure relative to the on and off-balance sheet assets. This, in order to reduce the risk of failure in case of disruptions on a bank's funding sources (BIS, 2014). These requirements apply for all banks, however the impact tends to be greater for larger banks due to their systemic importance. The Basel Committee acknowledges that the failure of larger banks can have severe effects on the global financial markets and on economic stability. It is therefore more difficult to replace its activities by other banks, which further 6 can cause disruption to the financial markets (BIS, 2023). As a result, larger banks are subject to intensive supervision, including requirements for dedicated risk management and by devoting more time and resources to institutions that are larger, riskier and complex (BIS, 2012). The Basel III framework was critically tested during the COVID-19 pandemic, which placed substantial stress on the global financial system. According to the Bank for International Settlements (BIS) (2021), the banking system remained resilient throughout the pandemic, largely due to the increased capital and liquidity requirements introduced by the reforms. As a result, no internationally active bank failed or required significant public sector funding, which further supports the effectiveness of Basel III enhancing financial stability during systemic shocks. Despite these observed benefits, the macroeconomic impact of Basel III remains a subject of ongoing debate (Cosimano & Hakura, 2011). Angelini & Gerali (2012) conclude that higher capital requirements increases the resilience of banks to shocks, the authors also highlight that higher capital requirements may lead to reduced lending activity, which potentially can slow down economic growth. 3.2 Bank Profitability 3.2.1 Macroeconomic Factors and Bank Profitability Laporšek et al. (2024) highlights the importance of macroeconomic factors that are crucial for banks to manage their risks and to implement effective policies. Such factors include real GDP growth, inflation and real interest rates, which affect bank profitability differently. The authors find a positive relationship between real GDP growth and bank profitability, as well as between long-term interest rates and profitability, but report negative effects between inflation and bank profitability. Similar results were found for European banks, where GDP growth was positive and significant for bank profitability (Borroni & Rossi, 2019; Mirović et al., 2024). In contrast, other studies concluded a negative relationship between bank profitability and real GDP growth due to increases in competition, which further decreases profits (Liu & Wilson, 2010; Islam, 2023). The unemployment rate has also been shown to negatively impact bank profitability due to its effect on loan repayment capacity. A high 7 unemployment rate reduces the borrowers’ ability to repay loans, which in turn can lower bank profitability through increased bank losses (Mirović et al., 2024). Islam (2023) found no significant effect between inflation, real interest rate and bank profitability for commercial banks in the UK. Consistent with this, Davis et al. (2022) find no significant relationship between inflation and bank profitability, which according to the authors can be explained by the fact that banks were not anticipating or forecasting inflation correctly. This study focuses on various types of banks. In contrast, Athanasoglou et al. (2008) reported a positive relationship between inflation and bank profitability for Greek commercial banks, suggesting that the banks forecasted future inflation correctly and adjusted the interest rates accordingly in order to yield higher profits. This relationship was also found for commercial banks in Central and Eastern European countries (Căpraru & Ihnatov, 2014). 3.3.2 Basel III’s Effect on Bank Profitability Profitability is an important element of bank resilience, as it allows financial institutions to absorb economic shocks and maintain stable operations during periods of financial stress (Laporšek et al., 2024). Several studies suggest that Basel III requirements, particularly those related to capital adequacy and liquidity risk, can positively influence bank profitability. Vousinas (2015) suggests that the implementation of the Basel III framework enhances a stable banking system by introducing stricter capital adequacy ratios, which in turn may contribute to improved financial performance. Tipu (2019) examines the broader implementation of Basel III on private commercial banks and finds that enhanced capital and liquidity standards contribute to improved bank profitability. Similarly, Rokiyanto and Siahaan (2024) investigated the effects of both the Tier 1 capital ratios and the NSFR on commercial banks profitability, finding that stronger capital buffers are positively associated with ROA. Laporšek et al. (2024) examine the relationship between Basel III-related variables and profitability among European banks, finding that capital adequacy and liquidity risk are positively associated with profitability. However, some studies present a contrasting perspective, suggesting that certain Basel III requirements may negatively affect bank profitability. Le et al. (2023) argue that increased capital buffers can constrain lending and reduce overall bank performance due to the rising marginal cost of funding. In addition, Rokiyanto and Siahaan (2024) report that while Tier 1 capital ratios support profitability, the NSFR has a negative effect on Return on Assets 8 (ROA). Rădulescu and Banica (2018) report that while capital adequacy under Basel III is positively associated with profitability in some Central and Eastern European countries, the relationship is not consistently observed across the region. Although the majority of the literature reports a positive relationship between Basel III implementation and ROA, the existence of mixed results highlights ongoing uncertainty regarding the overall impact of the framework on bank profitability. To develop a more comprehensive understanding of this relationship, the following hypothesis has been formulated: Hypothesis 1: 𝐻 : 𝐵𝑎𝑠𝑒𝑙 𝐼𝐼𝐼 𝑑𝑜𝑒𝑠 𝑛𝑜𝑡 ℎ𝑎𝑣𝑒 𝑎𝑛 𝑒𝑓𝑓𝑒𝑐𝑡 𝑜𝑛 𝑡ℎ𝑒 𝑝𝑟𝑜𝑓𝑖𝑡𝑎𝑏𝑖𝑙𝑖𝑡𝑦 0 𝑜𝑓 𝑏𝑎𝑛𝑘𝑠 𝑤𝑖𝑡ℎ𝑖𝑛 𝑡ℎ𝑒 𝐸𝑈. 𝐻 : 𝐵𝑎𝑠𝑒𝑙 𝐼𝐼𝐼 𝑑𝑜𝑒𝑠 ℎ𝑎𝑣𝑒 𝑎𝑛 𝑒𝑓𝑓𝑒𝑐𝑡 𝑜𝑛 𝑡ℎ𝑒 𝑝𝑟𝑜𝑓𝑖𝑡𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑜𝑓 𝑏𝑎𝑛𝑘𝑠 𝑤𝑖𝑡ℎ𝑖𝑛 𝑡ℎ𝑒 𝐸𝑈. 1 3.3.3 Basel III’s Effect on Profitability Depending on Bank size The impact of Basel III on bank profitability varies depending on bank size, with the reform generally having a greater impact on larger and systemically important institutions than on medium- to small-sized banks (European Banking Authority, n.d-b). Padgett (2013) argues that smaller banks have less access to resources and will therefore have difficulties with raising the additional capital that the reform requires, while larger banks are inclined to capitalize on the opportunities from the regulations. This is acknowledged by Gržeta et al. (2023), finding positive effects of Basel II and Basel III on profitability and efficiency for both large- and medium-sized commercial banks, while negative effects were found for smaller banks. The study explains these outcomes by highlighting that larger banks are more capable of implementing the Basel framework, whereas smaller banks face greater administrative and compliance related challenges. The analysis focuses on efficiency metrics and centers on medium- to large-sized institutions, with data primarily from the earlier phases of the implementation of Basel III. However, other studies suggest the opposite, indicating that larger bank size may actually reduce profitability. Yamin et al. (2025) who studies commercial banks, find that while Basel III capital adequacy requirements improve financial performance overall, larger banks report lower ROA, potentially due to operational inefficiencies and greater exposure to credit risk. Similarly, Laporšek et al. (2024) find a negative association between bank size and 9 profitability and explain this as a consequence of diseconomies of scale and increased complexity in complying with Basel III regulations. This study examines a wide range of banks across Europe and focuses on the early to mid-stages of the implementation of Basel III, with a particular emphasis on the size–profitability relationship. The existing literature offers differing views on how Basel III affects bank profitability depending on size, in order to understand this relationship, the following hypothesis have been developed: Hypothesis 2: 𝐻 : 𝑇ℎ𝑒𝑟𝑒 𝑖𝑠 𝑛𝑜 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑖𝑛 𝑡ℎ𝑒 𝑒𝑓𝑓𝑒𝑐𝑡 𝑜𝑓 𝐵𝑎𝑠𝑒𝑙 𝐼𝐼𝐼 𝑜𝑛 𝑝𝑟𝑜𝑓𝑖𝑡𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑑𝑒𝑝𝑒𝑛𝑑𝑖𝑛𝑔 𝑜𝑛 0 𝑏𝑎𝑛𝑘 𝑠𝑖𝑧𝑒 𝑤𝑖𝑡ℎ𝑖𝑛 𝑡ℎ𝑒 𝐸𝑈. 𝐻 : 𝑇ℎ𝑒𝑟𝑒 𝑖𝑠 𝑎 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑖𝑛 𝑡ℎ𝑒 𝑒𝑓𝑓𝑒𝑐𝑡 𝑜𝑓 𝐵𝑎𝑠𝑒𝑙 𝐼𝐼𝐼 𝑜𝑛 𝑝𝑟𝑜𝑓𝑖𝑡𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑑𝑒𝑝𝑒𝑛𝑑𝑖𝑛𝑔 𝑜𝑛 1 𝑏𝑎𝑛𝑘 𝑠𝑖𝑧𝑒 𝑤𝑖𝑡ℎ𝑖𝑛 𝑡ℎ𝑒 𝐸𝑈. 10 4. Data and Methodology 4.1 Data The initial dataset was collected from S&P Capital IQ, which provides lists of active, covered companies in the banking sector for the year 2023 across each EU member state. These country specific lists were merged into a single dataset, resulting in a total of 2,847 covered financial institutions. The covered companies in the banking sector include, among others, commercial, investment, savings and cooperative banks (S&P Capital IQ, n.d.-a). To avoid selection bias, a random sample of 500 covered companies was drawn using Matlab. To ensure the accuracy and consistency of the sample, each selected entity was verified in S&P Capital IQ to confirm that it was classified specifically as a bank, thereby excluding other types of financial institutions. In order to conduct this study, the sample is based on a few assumptions. Firstly, due to time and resource limitations, the sample was restricted to 500 banks. The final sample therefore consists of 500 banks operating within EU member countries. Secondly, all banks included in the sample were required to have been continuously active between 2010 and 2023. In addition to this, only banks with sufficient data coverage over the sample period were included. Although most variables were available for the full period from 2010 to 2023, the variable Net Interest Margin (NIM) contained missing values for the year 2010. As a result, the dataset is treated as an unbalanced panel data. All data for the variables used in the analysis were collected from S&P Capital IQ where the monetary variables were reported in thousands of euros. The finalized dataset comprises 6628 observations. The distribution of the sample is displayed below: Table 1: Sample Distribution Country Number Of Banks Percent (%) Austria 110 22 Belgium 9 1.8 Bulgaria 4 0.8 Czech 4 0.8 Cyprus 2 0.4 Denmark 17 3.4 Finland 40 8 France 32 6.4 11 Germany 122 24.4 Greece 3 0.6 Hungary 5 1 Ireland 2 0.4 Italy 45 9 Luxembourg 11 2.2 Malta 6 1.2 Netherlands 8 1.6 Poland 10 2 Portugal 25 5 Romania 7 1.4 Slovakia 3 0.6 Spain 20 4 Sweden 15 3 TOTAL 500 100 Source: S&P Capital IQ 4.2 Variables 4.2.1 Dependent Variable Since this study examines the impact of Basel regulatory frameworks on the profitability of banks, ROA is selected as the dependent variable. ROA is widely used in banking literature as a measure of profitability (e.g., Mamatzakis & Bermpei, 2016; Adebisi & Matthew, 2015; Kadioglu & Ocal, 2017), as it captures a bank’s ability to generate earnings from its total assets, where a higher ROA indicates more efficient asset utilization (Linh & Trang, 2019). It is calculated as net income divided by total assets (Terraza, 2015). 4.2.2 Independent Variables BASEL is the independent variable in the first regression model and is used to estimate the effect of Basel III implementation on bank profitability and is presented as a dummy variable. Basel III was first implemented in 2013 and is still being implemented (BIS, n.d.-e; BIS, n.d.-a), hence the dummy is equal to one for the time period 2013-2023 and zero otherwise. 12 BANKSIZE × BASEL is an interaction term included as an independent variable in regression model two, designed to examine whether the impact of Basel III on profitability varies depending on bank size. Including an interaction term allows for testing whether the relationship between the regulatory changes introduced by Basel III and profitability is conditional on the size of the bank, thereby capturing potential differences in how banks of varying sizes respond to the framework’s requirements. 4.2.3 Control Variables The variables GDP, INFLATION and UNEMPLOY are included as macroeconomic control variables, as they are known to influence bank profitability through various macroeconomic channels (Mirović et al., 2024; Laporšek et al., 2024; Borroni & Rossi, 2019). These variables were selected because they reflect overall economic conditions and influence the environment in which banks operate. This distinction ensures that changes in bank performance caused by general economic conditions are separated from those caused by regulatory reforms, allowing the BASEL dummy variable to more accurately capture the impact of Basel III. The variable GDP is the annual real GDP growth and is based on the reported change in real GDP by expenditure and is expressed as the percentage change from the previous year (S&P Capital IQ.-b). Given the variation in GDP measurements used in previous studies, this study will use real GDP as it accounts for inflation, providing an accurate measure of economic health (Callen, n.d.). INFLATION is the consumer price inflation and is expressed as the percentage change in the consumer price index from the previous year (S&P Capital IQ, n.d.-b). The variable UNEMPLOYMENT represents the unemployment rate and is based on the reported total labor force currently unemployed, expressed as the percentage of the total labor force that is unemployed (S&P Capital IQ, n.d.). The variables BANKSIZE, SOLVENCY, COSTINCOME, and NIM are included as bank-specific control variables, as they are known to influence bank profitability (Gržeta et al., 2023); Laporšek et al., 2024; Linh & Trang., 2019; Le et al., 2023). These particular variables were selected because they are not directly linked to Basel III regulatory components. This approach ensures that the effect of Basel III is not mistakenly captured by the control variables, allowing the BASEL dummy variable to accurately reflect the 13 regulatory impact. Consequently, variables explicitly related to Basel III requirements such as capital ratios, equity levels or liquidity measures were intentionally excluded. BANKSIZE represents the size of the bank, measured as the natural logarithm of its total assets. The natural logarithm is applied to reduce skewness and account for non-linear effects of size on profitability (Benoit, 2011). SOLVENCY is the debt to asset ratio. COSTINCOME represents the bank's cost-to-income ratio, expressed as operating expenses as a percentage of operating income (S&P Capital IQ, n.d.-c). NIM is the net interest margin, measured as the ratio of net interest income to average earning assets. If data on average earning assets is unavailable, S&P Capital IQ reports the ratio using average financial assets instead (S&P Capital IQ, n.d.-c). Table 2: Descriptive Statistics Variable Measurement Mean SD Min Max ROA % 0.478 0.618 -2.13 2.57 BASEL Dummy 0.786 0.410 0.00 1.00 GDP % 1.516 3.070 -8.95 8.67 INFLATION % 2.359 2.328 -0.48 10.33 UNEMPLOY % 7.629 3.657 2.98 24.44 BANKSIZE LN of total assets 14.151 2.262 10.72 20.57 SOLVENCY Ratio (Decimal) 0.074 0.141 0.00 0.69 COSTINCOME % 65.677 15.882 25.44 121.55 NIM % 1.926 0.870 0.24 5.13 All variables, except the dummy variable, are winsorized at the 1st and 99th percentiles to reduce the influence of extreme outliers, which can affect the robustness of the analysis (Frey, 2018). 14 4.3 Regression Model Two different regression models were created in order to test the hypothesis. The first regression investigates the relationship between BASEL and ROA while controlling for factors that can have an impact on ROA. The error term is included in order to capture the variation in the dependent variable that the model may not explain (Dougherty, 2011). Regression 1: Model for testing hypothesis 1: 𝑅𝑂𝐴 = β + β 𝐵𝐴𝑆𝐸𝐿 + β 𝐺𝐷𝑃 + β 𝐼𝑁𝐹𝐿𝐴𝑇𝐼𝑂𝑁 + β 𝑆𝑂𝐿𝑉𝐸𝑁𝐶𝑌 + β 𝐵𝐴𝑁𝐾𝑆𝐼𝑍𝐸 𝑖𝑡 0 1 2 𝑖𝑡 3 𝑖𝑡 4 𝑖𝑡 5 𝑖𝑡 + β 𝐶𝑂𝑆𝑇𝐼𝑁𝐶𝑂𝑀𝐸 + β 𝑁𝐼𝑀 + β 𝑈𝑁𝐸𝑀𝑃𝐿𝑂𝑌 + ϵ 6 𝑖𝑡 7 𝑖𝑡 8 𝑖𝑡 𝑖𝑡 The second regression model investigates the interaction effect between BASEL regulations and bank size on ROA. This allows for testing whether the impact of BASEL regulations on bank performance varies depending on the size of the bank. As with the first model, relevant control variables are included, and the error term captures the variation not explained by the independent variables (Dougherty, 2011). Regression 2: Model for testing hypothesis 2: 𝑅𝑂𝐴 = β + β 𝐵𝐴𝑆𝐸𝐿 + β 𝐺𝐷𝑃 + β 𝐼𝑁𝐹𝐿𝐴𝑇𝐼𝑂𝑁 + β 𝑆𝑂𝐿𝑉𝐸𝑁𝐶𝑌 + β 𝐵𝐴𝑁𝐾𝑆𝐼𝑍𝐸 𝑖𝑡 0 1 2 𝑖𝑡 3 𝑖𝑡 4 𝑖𝑡 5 𝑖𝑡 + β 𝐶𝑂𝑆𝑇𝐼𝑁𝐶𝑂𝑀𝐸 + β 𝑁𝐼𝑀 + β 𝑈𝑁𝐸𝑀𝑃𝐿𝑂𝑌 + β (𝐵𝐴𝑁𝐾𝑆𝐼𝑍𝐸 × 𝐵𝐴𝑆𝐸𝐿) + ϵ 6 𝑖𝑡 7 𝑖𝑡 8 𝑖𝑡 9 𝑖𝑡 𝑖𝑡 4.4 Empirical Model To assess the relationships and robustness of the variables included in the study, a series of statistical tests were conducted. These include correlation analysis, model specification tests, heteroscedasticity testing and a Difference-in-Differences (DiD) approach. All tests were performed using Matlab except for the regressions which were conducted in Stata. 4.4.1 Correlation Checking for correlation helps identify strong relationships between variables that could affect the stability and reliability of a regression model, as high correlations can reduce the clarity of the model and the accuracy of its prediction (Costa, 2017). Accordingly, a correlation matrix was created as displayed in Table 3. Correlation coefficients are generally interpreted as low when below 0.4, moderate between 0.4 and 0.8, and high when exceeding 15 0.8 (Shi & Conrad, 2009). As shown, none of the independent variables exhibit excessively high correlation. Still, moderate correlations such as the one between ROA and COSTINCOME at -0.5609 and the one between BANKSIZE and SOLVENCY at 0.5671 could potentially influence the interpretation of results. To examine this, alternative regression models were estimated with and without COSTINCOME and SOLVENCY. The models remained statistically significant regardless, so all variables were included. Table 3: Correlation Matrix VARIABLES ROA BANKSIZE NIM COST SOLVENCY INFLATION GDP UNEM BASEL INCOME PLOY ROA 1.0000 BANKSIZE -0.0267 1.0000 NIM 0.2745 -0.2447 1.0000 COSTINCOME -0.5609 -0.2146 -0.1773 1.0000 SOLVENCY -0.0751 0.5671 -0.2336 -0.1222 1.0000 INFLATION 0.0803 0.0367 0.0847 -0.1073 -0.0436 1.0000 GDP 0.0576 0.0397 0.0011 -0.0451 0.0033 0.1822 1.0000 UNEMPLOY -0.0402 -0.0621 0.0089 -0.0279 0.0380 -0.2474 -0.0651 1.0000 BASEL 0.0712 0.0658 -0.1461 -0.0198 -0.0835 -0.0191 -0.0140 -0.1050 1.0000 4.4.2 Model description To determine the appropriate model, various tests were conducted. As an initial step, it was tested whether the pooled OLS model was suitable for the sample using the Breusch-Pagan LM test. Pooled OLS is frequently used as an initial model in panel data analysis due to its simplicity. It relies on strong assumptions such as homogeneity across individuals and over time, the absence of unobserved heterogeneity, and no correlation between individual effects and explanatory variables (Baltagi & Griffin, 1984; Baltagi, Bresson & Pirotte, 2008). Given these assumptions, it was important to evaluate whether the pooled model was suitable, as violations could lead to biased and inconsistent estimates (Baltagi & Griffin, 1984; Baltagi, Bresson & Pirotte, 2008). The Breusch-Pagan LM test serves this purpose by testing for the presence of random effects (RE). If the null hypothesis of no RE is rejected, it suggests that the pooled OLS model is misspecified and that a panel data model, such as random or fixed effects, would provide a more appropriate specification (Baltagi, Feng & Kao, 2012). The test 16 resulted in a p-value of 0.000, leading to the rejection of the null hypothesis and indicating that the pooled OLS model is unsuitable. The next step was to determine whether a fixed effects (FE) or RE model would be more appropriate. A FE model controls for all unobserved characteristics that do not vary over time and does not rely on the assumption that these effects are uncorrelated with the explanatory variables. In contrast, a RE model assumes such unobserved effects are random and uncorrelated with the regressors, allowing for more efficient estimation under correct specification. While RE models are more efficient and allow for the inclusion of time-invariant variables, they may yield biased results if the assumptions do not hold. It is also worth noting that in RE models, interaction terms can become insignificant as the RE model captures both within- and between-variation (Spineli & Pandis, 2020). FE models, although less efficient, are more robust in the presence of unobserved heterogeneity and are preferred when causal interpretation is the main goal. (Clarke et al., 2010) To determine whether the FE or RE model was more appropriate for the dataset, the Hausman test was conducted. This test compares the FE and RE estimates to assess whether the individual specific effects are correlated with the regressors. The null hypothesis assumes no correlation, in which case the RE model is preferred. If the null hypothesis is rejected, it suggests that the RE assumptions do not hold and the FE model provides more reliable and consistent estimates (Hausman, 1978). In this case, the Hausman test returned a p-value of 0.0000, leading to the rejection of the null hypothesis. Therefore, the FE model was preferred over the RE model. 4.4.3 Heteroscedasticity and Serial Correlation The FE model, when applied to the dataset, was tested for heteroscedasticity and serial correlation. The Wald test was used to assess whether the variance of the error terms differs across banks, which would indicate the presence of heteroscedasticity. The null hypothesis of the Wald test states that the error variance is constant across all cross-sectional units. A rejection of this hypothesis suggests that heteroscedasticity is present in the model (Baum, 2001). In this case, the Wald test yielded a p-value of 0.000, leading to the rejection of the null hypothesis and indicating that heteroscedasticity is present. The Wooldridge test was applied to detect serial correlation in the residuals of the panel data model, as serial correlation can bias standard errors and lead to inefficient estimates. The null hypothesis for 17 this test is that there is no autocorrelation in the errors (Drukker, 2003). The test returned a p-value of 0.000, resulting in the rejection of the null hypothesis and confirming the presence of serial correlation in the model. To account for these issues, clustered standard errors were used to ensure robust and valid inference (Born & Breitung, 2016). 4.4.4 Difference in Difference (DiD) To test Hypothesis 2, this study applies a DiD model, which estimates the effect of exposure by comparing changes over time between a treatment group and a control group (Dimick & Ryan, 2014). Specifically, the model estimates whether the impact of Basel III varies depending on bank size by using the interaction term BANKSIZE x BASEL. This allows for comparison between the changes in bank profitability before and after the implementation of Basel III across banks with different sizes. As outlined in Section 3.3.3, larger banks are expected to be more affected by Basel III and therefore serve as the treatment group, while smaller banks, being less impacted by the policy change, act as the control group. To ensure the validity of the DiD regression with unbalanced panel data, the Wald test is used to test for heteroskedasticity and the Wooldridge test for serial correlation, similar to Regression 1. The Wald test returned a p-value of 0.000, therefore rejecting the null hypothesis of no heteroskedasticity in the model. Similarly, the null hypothesis of no serial correlation is rejected based on the Wooldridge test. To account for these issues, clustered standard errors are used to ensure robust and valid inference (Born & Breitung, 2016). 18 5. Results 5.1 Results for Regression Model 1 Table 4: Regression Output for Model 1 Variable Coefficient Clustered SE P-value Constant 2.5044 0 .7239 0.001*** BANKSIZE -0.0401 0.0471 0.395 NIM 0.1072 0.0252 0.000*** SOLVENCY -0.7754 0.2478 0.002*** COSTINCOME -0.0209 0.0013 0.000*** INFLATION -0.0123 0.0036 0.001*** GDP 0.0081 0.0019 0.000*** UNEMPLOY -0.0382 0.0078 0.000*** BASEL 0.0950 0.0216 0.000*** F-score 66.39 0.000*** R2 (Within) 0.3000 R2 (Between) 0.3615 R2 (Overall) 0.3237 *** p < 0,01; ** p < 0,05; * p < 0,1 The results for Model 1 are presented in Table 4. As can be seen, all variables except for BANKSIZE are significant at the 1% level and the whole model has a F-score with 1% level significance. The variable of interest, BASEL, is statistically significant at the 1% level, with a coefficient of 0.0950. This suggests that during the Basel III period (from 2013 to 2023) banks experienced on average an increase in ROA of 0.0950 percentage points compared to the pre-Basel III period. The null hypothesis that Basel III does not affect the profitability of banks within the EU can therefore be rejected. 19 5.2 Results for Regression Model 2 Table 5: Regression Output for Model 2 Variable Coefficient Clustered SE P-value Constant 2.3904 0.7341 0.001*** BANKSIZE -0.0318 0.0478 0.505 NIM 0.1092 0.0253 0.000*** SOLVENCY -0.8353 0.2544 0.001*** COSTINCOME -0.0209 0.0013 0.000*** INFLATION -0.0119 0.0036 0.001*** GDP 0.0079 0.0019 0.000*** UNEMPLOY -0.0393 0.0077 0.000*** BASEL 0.3569 0.1208 0.003*** BANKSIZE x BASEL -0.0184 0.0081 0.025** F-score 59.27 0.000*** R2 (Within) 0.3013 R2 (Between) 0.3444 R2 (Overall) 0.3140 *** p < 0,01; ** p < 0,05; * p < 0,1 The results for Model 2 are similar to Model 1 where all variables except for BANKSIZE are statistically significant as well as the whole model at the 1% level. The interaction variable, BANKSIZE x BASEL is significant at the 5% level with a negative coefficient of -0.0184. This means that for every unit increase in bank size, the effect of Basel III on ROA decreases by 0.0184 percentage points, which means that smaller banks benefit more from Basel III than larger banks in terms of profitability. Based on this, the null hypothesis that there is no difference in the effect of Basel III on profitability depending on bank size within the EU can be rejected. 20 5.3 Robustness test To assess the reliability of the main regression results, a series of robustness checks were conducted. These checks ensured that the findings are not sensitive to a specific model or the inclusion of certain independent variables (Lu & White, 2014). The Breusch-Pagan LM test rejected the use of pooled OLS for our data while the Hausman test indicated that a FE model was appropriate than a RE model. Tests for heteroskedasticity and serial correlation by using the Wald and Wooldrigde test confirmed the presence in our data, leading to the use of clustered standard errors in the regressions (Born & Breitung, 2016). In the first robustness check, the regression for Model 1 was re-estimated by dropping NIM and GDP. The results, presented in Appendix 9.2, show that the variables remained significant and the signs of the coefficients were consistent with the main model. BASEL, the variable of interest also remained significant. However, BANKSIZE became significant at the 10% level suggesting that NIM and GDP may have been absorbing some of its variation. The overall model was still significant at the 1% level. Similar results were obtained for Model 2, where the variable of interest also retained its significance which further supported the robustness of the findings (Appendix 9.3). Secondly, to test the robustness of our FE results, we re-estimated Model 1 using the RE model to examine whether our findings are sensitive to the model specification. The results are shown in Appendix 9.4. All variables, including BASEL, remained statistically significant while BANKSIZE became significant at the 10% level. Hence, the RE results support the robustness of the main findings. For Model 2, BANKSIZE is still insignificant. However the interaction variable BANKSIZE x BASEL became insignificant even though the whole model is significant. Despite the insignificance of the interaction variable, the overall model remained significant indicating robustness. Furthermore, we compare regular, robust and clustered standard errors in order to evaluate the sensitivity of our key variables under different error structures. In Model 1, the coefficients and the significance of the variables remained unchanged, except for BANKSIZE, which with regular standard errors became significant at the 5% level (Appendix 9.6). Similar patterns were observed in Model 2 (Appendix 9.7), where BANKSIZE also became significant at the 10% level with regular standard errors. These 21 findings suggest that the results in the main model are consistent and robust to different standard errors. As the sample includes a relatively high number of German banks, a final robustness check is conducted by first excluding them from the sample and then only running the sample with German banks to ensure that the results are not disproportionately influenced by a single country. For Model 1, the results are shown in Appendix 9.8. Without German banks in the sample, BASEL retained its positive coefficient and its significance. The signs of all remaining coefficients are consistent with the main model. Similar results were obtained for Model 2 where the coefficients remained stable in both sign and significance. The interaction term BANKSIZE x BASEL retained its significance at the 10% level. When restricting the sample to only including German banks, all variables in Model 1 and Model 2, except for COSTINCOME and BANKSIZE becomes insignificant (Appendix 9.9). Notably, the sign of SOLVENCY changes, while the direction of all other coefficients remains consistent. The overall statistical significance of the models still held at the 1% level. The insignificance of the variables can be explained by the fact that the sample size is smaller with only German banks. As a conclusion, these findings did indicate that the main results are not solely driven by German banks and therefore adds further support to the robustness of the model. 22 6. Analysis and Discussion 6.1 Regression Model 1 The first regression model investigates the average impact of Basel III on bank profitability, measured by ROA. The variable of interest, BASEL, is a dummy variable equal to 1 for years following the introduction of Basel III (2013-2023) and 0 otherwise. As expected, the coefficient, BASEL, is positive at 0.0950 and statistically significant at the 1% level. Since the p-value is less than 5%, the null hypothesis is rejected, indicating that on average, banks experienced an increase in ROA compared to the pre-Basel III period. The results align with several studies mentioned in the literature review that report a positive relationship between Basel III capital and liquidity reforms and bank profitability (Vousinas, 2015; Tipu, 2019; Rokiyanto & Siahaan, 2024; Laporšek et al., 2024). Some possible explanations for this result, as suggested by the authors, are that stronger capital and liquidity requirements improve financial performance, reduce risk and thereby contribute to increased profitability of banks. However, while these explanations are consistent with our findings, it is not possible to confirm that these explanations directly apply to our result. One interpretation of this positive relationship is that the Basel III framework reduced risk, where the capital conservation and countercyclical buffers along with the LCR and NSFR improved the sector's ability to withstand financial stresses as observed during the Covid-19 pandemic (BIS, 2021). As a result, the requirements could have improved bank profitability over the long term horizon by minimizing volatility and enhancing confidence in the banking system. While the results align with several studies as mentioned above, it also differs with other findings in the literature that suggest the reforms negatively impact bank performance. For instance, Le et al. (2023) argues that the increased capital buffers can reduce the overall bank performance due to rising marginal funding costs. Similarly, Rokiyanto and Siahaan (2024) found that although the capital ratios support profitability, the NSFR may negatively affect bank profitability. Rădulescu and Banica (2018) also report inconsistent findings across countries in Central and Eastern Europe. These mixed results of Basel III impact could be explained by country-specific factors such as different bank structure, macroeconomic conditions and different bank regulations. Although all EU banks are subject to the same Basel III framework under CRR and CRD, the implementation and supervision practices can vary across countries, which further can contribute to differences in the impact. 23 6.2 Regression Model 2 For regression model 2, the thesis explores if the implementation of Basel III affected bank profitability differently depending on bank size. The interaction term BANKSIZE x BASEL, which is the variable of interest in this model, had a negative coefficient at -0.0184 and was statistically significant at the 5% level. The null hypothesis that there is no difference in the effect of Basel III on profitability depending on bank size within the EU can further be rejected. The negative coefficient indicates that while Basel III had an overall positive impact on bank profitability, as reflected in the significant and positive coefficient of the BASEL variable, the benefit was not uniform across banks of different sizes. Specifically, the profitability gains from Basel III diminished as bank size increased, suggesting that smaller banks experienced relatively stronger improvements in ROA. This finding contributes to the ongoing academic debate on whether the effects of Basel III are size dependent. Our findings align with those of Yamin et al. (2025) and Laporšek et al. (2024) who document a negative relationship between bank size and profitability, particularly under the Basel III framework. These studies argue that larger banks may face greater operational complexity and diseconomies of scale, which can limit their ability to convert regulatory compliance into profitability gains. Although the results of these studies align with ours, we cannot definitively claim that the underlying explanations are entirely consistent. However, the recurring evidence of size-related profitability constraints under Basel III suggests that operational complexity and diseconomies of scale are plausible contributing factors, suggesting further investigation. In contrast, the findings of Gržeta et al. (2023) suggest that larger and medium-sized banks benefit more from Basel II and III, while smaller banks are negatively affected. The study explains this outcome by pointing to the greater ability of larger banks to implement the Basel framework, whereas smaller banks face more administrative and compliance related challenges. The divergence in findings across studies highlights the complexity of the framework's impact and underscores that the relationship between regulation and profitability is not straightforward. External factors such as geographic region, the timing of implementation, and institutional characteristics likely play a significant role. A closer look at the differences in study design supports this view. Laporšek et al. (2024) analyze a wide range of European banks during the early to mid stages of the implementation, focusing specifically on the link between size and profitability, and report a negative relationship between bank size and 24 profitability. In contrast, Gržeta et al. (2023) examine earlier stages of the Basel III implementation, with a stronger emphasis on efficiency and focus mainly on medium- to large-sized institutions. The findings suggest that larger banks benefit more from Basel reforms, while smaller banks are negatively affected. These variations in sample composition, time period, and methodological approach can strongly influence the observed outcomes. 25 7. Conclusion This thesis set out to investigate the impact of Basel III on bank profitability, specifically examining whether the framework affected large and small banks differently. In order to do so, two research questions were formulated: “How does Basel III affect the profitability of banks within the EU?” and “Does Basel III have a differentiated impact on banks profitability based on their size?”. Based on prior literature, it was expected that the stricter regulatory framework would have a positive impact on bank profitability and that the effect would differ depending on bank size. The results for Model 1 showed that with statistical significance, the implementation of Basel III increased bank profitability, as measured by ROA. This suggests that the regulation, despite its stricter capital and liquidity requirements, improved the overall financial performance of EU banks, potentially due to increased stability and better risk management. This finding supports the first hypothesis and aligns with several studies in the literature that reports a positive relationship between Basel III reforms and profitability. Model 2, which introduced an interaction term between bank size and the Basel III period, examined whether this effect varied by bank size. The results indicate a negative and significant interaction between bank size and Basel III, implying that smaller banks experienced greater profitability compared to larger banks. As bank size increases, the positive effect of Basel III on ROA diminishes. This finding supports the second hypothesis and suggests that the impact of Basel III is not uniform across banks of different sizes which aligns with several studies in the literature. In summary, the findings support that Basel III has affected bank profitability within the EU and have enhanced it, particularly for smaller banks. By including different bank types and examining the overall impact of Basel III on bank profitability within the EU, this adds new perspective on the debate on the relationship between financial regulation and profitability and suggests that a well-designed regulatory framework may support both economic stability and performance. 26 7.1 Limitations and Further Research While the study provides important insights into the effect of Basel III on bank profitability in the EU, there are some limitations that should be acknowledged. Firstly, the distribution across countries is uneven in the sample. Germany accounts for over 24% of the total sample, while smaller EU countries such as Romania and Ireland contribute with less than 10 banks each. Although the robustness checks confirmed that the main findings remained consistent after excluding German banks, a more even distribution of banks across EU member states would enhance the representativeness of the sample and allow for more generalizable conclusions. Therefore, future research could address this by constructing a more country-balanced panel dataset. Secondly, even though the sample of this thesis is relatively large compared to most of the literature, it would still be interesting for future research to increase the sample size in order to analyze if the effect is persistent. Lastly, another limitation concerns the construction of the BASEL dummy variable,, coded as 1 for the period 2013–2023 and 0 otherwise. This binary treatment assumes an immediate and uniform impact of Basel III implementation starting from 2013. While this approach is similar to what other studies have done, it does not account for the fact that Basel III was introduced slowly and at different times across EU countries and banks. An alternative construction for further research is to use bank-level data, in other words use specific information such as capital ratios, liquidity measures, or the year a bank began complying with Basel III, since some banks may have followed Basel III earlier or more completely than others. Furthermore, the model does not incorporate lagged variables such as a lagged BASEL term or lagged ROA. This limits the ability to capture delayed effects of regulation or persistence in profitability over time. Including lagged terms in future research may better reflect the time it takes for regulatory changes to impact financial performance, particularly as banks adjust their strategies and capital structures. Thus, future work could benefit from including a lagged structure. 27 8. References Adebisi, J. F., & Matthew, O. B. (2015). The impact of non-performing loans on firm profitability: A focus on the Nigerian banking industry. American Research Journal of Business and Management, 1(4), 1–9. https://bhlss.wordpress.com/wp-content/uploads/2017/02/2-impact-of-npl-on-profitability.pdf Angelini, P. and Gerali, A., 2012. Banks’ reactions to Basel-III. Bank of Italy Temi di Discussione (Working Paper) No. 876. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2118496 Athanassoglou, P. P., Brissimis, S. N., & Delis, M. D. (2008). Bank-specific, industry-specific and macroeconomic determinants of bank profitability. Journal of International Financial Markets, Institutions and Money, 18(2), 121–136. https://doi.org/10.1016/j.intfin.2006.07.001 Baum, C.F., 2001. Residual diagnostics for cross-section time series regression models. The Stata Journal, 1(1), pp.101–104. https://doi.org/10.1177/1536867X0100100108 Baltagi, B.H., Bresson, G. and Pirotte, A., 2008. To pool or not to pool? In: L. Mátyás and P. Sevestre, eds. The Econometrics of Panel Data. 3rd ed. Berlin: Springer-Verlag, pp.517–539. https://doi.org/10.1007/978-3-540-75892-1_16 Baltagi, B.H., Feng, Q. and Kao, C., 2012. A Lagrange Multiplier test for cross-sectional dependence in a fixed effects panel data model. Journal of Econometrics, 170(1), pp.164–177. https://doi.org/10.1016/j.jeconom.2012.04.004 Baltagi, B.H. and Griffin, J.M., 1984. Short and long run effects in pooled models. International Economic Review, 25(3), pp.631–645. https://doi.org/10.2307/2526223 Benoit, K. (2011). Linear Regression Models with Logarithmic Transformations. Retrieved from: https://links.sharezomics.com/assets/uploads/files/1600247928973-from_slack_logmodels2.p df [Accessed 05.03.2025] 28 BIS. (2011). Basel III: A global regulatory framework for more resilient banks and banking systems. Retrieved from: https://www.bis.org/publ/bcbs189.pdf [Accessed 05.03.2025] BIS. (2018). Basel Committee Charter. Retrieved from: https://www.bis.org/bcbs/charter.htm [Accessed 05.03.2025] BIS. (2017). Basel III: Finalising post-crisis reforms. Retrieved from: https://www.bis.org/bcbs/publ/d424.htm [Accessed 05.03.2025] BIS. (n.d.-d). Basel III: international regulatory framework for banks. Retrieved from: https://www.bis.org/bcbs/basel3.htm [Accessed 05.03.2025] BIS. (2014). Basel III leverage ratio framework and disclosure requirements. Retrieved from: https://www.bis.org/publ/bcbs270.pdf [Accessed 05.03.2025] BIS. (2013). Basel III: The liquidity coverage ratio and liquidity risk monitoring tools. Retrieved from: https://www.bis.org/publ/bcbs238.pdf [Accessed 05.03.2025] BIS. (2013). Basel III: The net stable funding ratio. Retrieved from: https://www.bis.org/bcbs/publ/d295.pdf [Accessed 05.03.2025] BIS. (n.d.-e). Basel III transitional arrangements, 2017-2018. Retrieved from: https://www.bis.org/bcbs/basel3/b3_trans_arr_1728.pdf [Accessed 05.03.2025] BIS. (2012). Core principles for effective banking supervision. Basel Committee on Banking Supervision. Retrieved from: https://www.bis.org/publ/bcbs230.htm [Accessed 05.03.2025] BIS. (n.d.-c). Definition of capital in Basel III - Executive Summary. Retrieved from: https://www.bis.org/fsi/fsisummaries/defcap_b3.pdf [Accessed 05.03.2025] BIS. (2021). Early lessons from the Covid-19 pandemic on the Basel reforms. Retrieved from: https://www.bis.org/bcbs/publ/d521.pdf [Accessed XXX. 2025] 29 BIS. (n.d.-a). History of the Basel Committee. Retrieved from: https://www.bis.org/bcbs/history.htm [Accessed XXX. 2025] BIS. (2004). International Convergence of Capital Management and Capital Standards. Retrieved from: https://www.bis.org/publ/bcbs107.pdf [Accessed XXX. 2025]. BIS. (2006). International Convergence of Capital Measurement and Capital Standards. A revised framework – Comprehensive version. Bank for International Settlements. Retrieved from: https://www.bis.org/publ/bcbs128.pdf [Accessed XXX. 2025] BIS. (2023). Scope and definitions (SCO). Basel Committee on Banking Supervision. Retrieved from: https://www.bis.org/basel_framework/standard/SCO.htm Born, B. and Breitung, J., 2016. Testing for serial correlation in fixed-effects panel data models. Econometric Reviews, 35(7), pp.1290–1316. https://doi.org/10.1080/07474938.2014.976524 Borroni, M. and Rossi, S., 2019. Banking in Europe: The Quest for Profitability after the Great Financial Crisis. Cham: Palgrave Macmillan. h ttps://doi.org/10.1007/978-3-030-15013-6 Callen, T. (n.d). Gross Domestic Product: An Economy’s All. Retrieved from: https://www.imf.org/en/Publications/fandd/issues/Series/Back-to-Basics/gross-domestic-prod uct-GDP? [Accessed 05.03.2025] Căpraru, B., & Ihnatov, I. (2014). Banks’ profitability in selected Central and Eastern European countries. Procedia Economics and Finance, 16, 587–591. https://doi.org/10.1016/S2212-5671(14)00844-2 Clarke, P., Crawford, C., Steele, F. and Vignoles, A., 2010. The choice between fixed and random effects models: Some considerations for educational research. IZA Discussion Paper No. 5287. Bonn: Institute for the Study of Labor (IZA). https://ftp.iza.org/dp5287.pdf 30 Claessens, S., & Kose, M. A. (2013). Financial crises: Explanations, Types and Implications (IMF Working Paper No. 13/28). International Monetary Fund. https://www.imf.org/external/pubs/ft/wp/2013/wp1328.pdf Cosimano, T.F. and Hakura, D.S., 2011. Bank behavior in response to Basel III: A cross-country analysis. IMF Working Paper No. WP/11/119. Washington, D.C.: International Monetary Fund. Costa, V. (2017). Correlation and Regression. In: Naghettini, M. (eds) Fundamentals of Statistical Hydrology. Springer, Cham. https://doi.org/10.1007/978-3-319-43561-9_9 Davis, E. P., Karim, D., & Noel, D. (2022). The effects of macroprudential policy on banks' profitability. International Review of Financial Analysis, 80, 101989. https://doi.org/10.1016/j.irfa.2021.101989 Dimick, J. B., & Ryan, A. M. (2014). Methods for evaluating changes in health care policy: The difference-in-differences approach. JAMA, 312(22), 2401–2402. https://doi.org/10.1001/jama.2014.16153 Dougherty, C. (2011). Introduction to econometrics. Oxford university press, USA. Drukker, D.M., 2003. Testing for serial correlation in linear panel-data models. The Stata Journal, 3(2), pp.168–177. https://doi.org/10.1177/1536867X0300300206 European Banking Authority. (n.d.-b). Basel III reforms: impact study and key recommendations. Retrieved from: https://www.eba.europa.eu/sites/default/files/Basel%20III%20reforms%20-%20Impact%20st udy%20and%20key%20reccomendations.pdf [Accessed 05.03.2025] European Banking Authority. (n.d.-a) The Basel framework: the global regulatory standards for banks. Retrieved from: https://www.eba.europa.eu/activities/basel-framework-global-regulatory-standards-banks [Accessed 05.03.2025] 31 European Commission. (2021). Banking Package 2021: new EU rules to strengthen banks' resilience and better prepare for the future. Retrieved from: https://ec.europa.eu/commission/presscorner/detail/en/ip_21_5401 [Accessed 05.03.2025] European Commission. (2024). Basel III. Retrieved from: https://finance.ec.europa.eu/news/basel-iii-2024-07-25_en [Accessed 05.03.2025] European Union. (n.d.-b). Areas of EU action. Retrieved from: https://commission.europa.eu/about/role/law/areas-eu-action_en [Accessed 05.03.2025] European Union. (n.d.-a). Facts and figures about the European Union. Retrieved from: https://european-union.europa.eu/principles-countries-history/facts-and-figures-european-uni on_en [Accessed 05.03.2025] European Union. (n.d.-c). Types of legislation. Retrieved from: https://european-union.europa.eu/institutions-law-budget/law/types-legislation_en [Accessed 05.03.2025] Frey, B. (2018). Winsorizing. In The SAGE encyclopedia of educational research, measurement, and evaluation(Vols. 1–4). SAGE Publications. https://doi.org/10.4135/9781506326139 Gatzert, N. and Wesker, H. (2012). A comparative assessment of Basel II/III and Solvency II. The Geneva Papers on Risk and Insurance - Issues and Practice, 37(4), pp.539–570. https://doi.org/10.1057/gpp.2012.3 Gržeta, I., Žiković, S., & Tomas Žiković, I. (2023). Size matters: Analyzing bank profitability and efficiency under the Basel III framework. Financial Innovation, 9(1), 43. https://doi.org/10.1186/s40854-022-00412-y Hausman, J.A., 1978. Specification tests in econometrics. Econometrica, 46(6), pp.1251–1271. https://doi.org/10.2307/1913827 32 Islam, M. R. (2023). The Impact of Macroeconomic Factors on Profitability of Commercial Bank in the UK. International Journal For Multidisciplinary Research, 5(1), 1-19. https://doi.org/10.36948/ijfmr.2023.v05i01.1406 Kadioglu, E., Telceken, N., & Ocal, N. (2017). Effect of the asset quality on the bank profitability (SSRN Scholarly Paper No. 3971785). Social Science Research Network. https://ssrn.com/abstract=3971785 Laporšek, S., Švagan, B., Stubelj, M., & Stubelj, I. (2025). Profitability Drivers in European Banks: Analyzing Internal and External Factors in the Post-2009 Financial Landscape. Risks, 13(1), 2. https://doi.org/10.3390/risks13010002 Linh, D. H., & Trang, V. K. (2019). Impact of capital on profitability of banks: Evidence from Vietnamese commercial banks. Journal of Economics and Business, 2(2). 10.31014/aior.1992.02.02.94 Le, T. N. L., Nasir, M. A., & Huynh, T. L. D. (2023). Capital requirements and banks performance under Basel-III: A comparative analysis of Australian and British banks. The Quarterly Review of Economics and Finance, 87, 146–157. https://doi.org/10.1016/j.qref.2020.06.001 Liu, H. and Wilson, J.O.S., 2010. The profitability of banks in Japan. Applied Financial Economics, 20(24), pp.1851–1866. https://doi.org/10.1080/09603107.2010.526577 Lu, X. and White, H., 2014. Robustness checks and robustness tests in applied economics. Journal of Econometrics, 178(1), pp.194–206. https://doi.org/10.1016/j.jeconom.2013.08.016 Mamatzakis, E., & Bermpei, T. (2016). What is the effect of unconventional monetary policy on bank performance? Journal of International Money and Finance, 67, 239–263. https://doi.org/10.1016/j.jimonfin.2016.05.005 Mirović, V., Kalaš, B., Milenković, N., Andrašić, J. and Đaković, M., 2024. Modelling profitability determinants in the banking sector: The case of the Eurozone. Mathematics, 12(6), p.897. https://doi.org/10.3390/math12060897 33 Padgett, S. (2013). The negative impact of Basel III on small business financing. Ohio State Entrepreneurial Business Law Journal, 8, 183–210. https://heinonline.org/HOL/LandingPage?handle=hein.journals/eblwj8&div=12&id=&page= Rădulescu, M., & Banica, L. (2018). The profitability and capital adequacy in central and eastern European countries in the light of the Basel III requirements: a forecast approach. Ekonomický časopis, 66(5), 479-502. https://www.zbw.eu/econis-archiv/bitstream/11159/3937/1/1688758070.pdf Rokiyanto, R. J., & Siahaan, A. (2024). Analysis of the Basel III liquidity and minimum capital requirements towards bank performance: Evidence from commercial banks in Indonesia. In Proceedings of the 5th International Conference on Global Innovation and Trends in Economy 2024 (INCOGITE 2024). Atlantis Press. 10.2991/978-94-6463-585-0_19 Shi, R. and Conrad, S.A., 2009. Correlation and regression analysis. Annals of Allergy, Asthma & Immunology, 103(4 Suppl 1), pp.S34–S40. Spineli, L.M. and Pandis, N., 2020. Fixed-effect versus random-effects model in meta-regression analysis. American Journal of Orthodontics and Dentofacial Orthopedics, 158(5), pp.770–772. https://doi.org/10.1016/j.ajodo.2020.07.016 S&P Capital IQ. (n.d.-a). Bank Covered by SNL. CIQ Pro: Sweden | Banks Covered by SNL [Accessed 05.03.2025] S&P Capital IQ. (n.d.-b). Economic & Demographic data. CIQ Pro: Germany | Economic & Demographic Data [Accessed 05.03.2025] S&P Capital IQ. (n.d.-c). Financial Highlights. CIQ Pro: Aareal Bank AG | Financial Highlights [Accessed 05.03.2025] Terraza, V. (2015). The effect of bank size on risk ratios: Implications of banks’ performance. Procedia Economics and Finance, 30, 903–909. https://doi.org/10.1016/S2212-5671(15)01340-4 34 Tipu, M. (2019). Impact of Basel III on profitability & capital adequacy of private commercial banks in Bangladesh and banking generation-wise implementation status of Basel Accord in Bangladesh. Trent University. http://dx.doi.org/10.2139/ssrn.3688727 Vousinas, G. L. (2015). Supervision of financial institutions: The transition from Basel I to Basel III. A critical appraisal of the newly established regulatory framework. Journal of Financial Regulation and Compliance, 23(4), 380–394. https://doi.org/10.1108/JFRC-02-2015-0011 Yamin, N., Asif, M., Hasnat, M. A., Islam, M. M., Nasiruddin, M., Khandakar, H., & Rahman, M. A. (2025). The impact of Basel III implementation on the financial performance of the banking industry in Bangladesh. Banks and Bank Systems, 20(1), 98–108. http://dx.doi.org/10.21511/bbs.20(1).2025.09 35 9. Appendix Appendix 9.1 Countries and covered companies in the banking sector within the EU Country Number Of Covered Companies Percent (%) Austria 313 11 Belgium 35 1.23 Bulgaria 19 0.67 Czech 10 1.05 Croatia 24 0.84 Cyprus 17 0.6 Denmark 57 2 Estonia 12 0.42 Finland 106 3.72 France 244 8.57 Germany 1178 41.39 Greece 19 0.67 Hungary 21 0.74 Irland 17 0.6 Italy 271 9.52 Latvia 15 0.53 Lithuania 9 0.32 Luxemburg 77 2.71 Malta 14 0.49 Netherlands 42 1.48 Polen 29 1.02 Portugal 93 3.27 Romania 25 0.88 Slovakia 13 0.46 Slovenia 14 0.49 Spain 80 2.81 36 Sweden 72 2.53 TOTAL 2847 100 Source: Capital IQ Appendix 9.2. Robustness test for Model 1 Regression Output for Model 1 without NIM and GDP Variable Coefficient Clustered SE P-value Constant 3.5097 0.6671 0.000*** BANKSIZE -0.0865 0.0464 0.063* SOLVENCY -0.7730 0.2568 0.003*** COSTINCOME -0.0229 0.0012 0.000*** INFLATION -0.0078 0.0034 0.022** UNEMPLOY -0.0366 0.0077 0.000*** BASEL 0.0639 0.0208 0.002*** F-score 70.41 0.000*** R2 (Within) 0.2880 R2 (Between) 0.2945 R2 (Overall) 0.2784 *** p < 0,01; ** p < 0,05; * p < 0,1 Appendix 9.3. Robustness test for Model 2 Regression Output for Model 2 without NIM and GDP Variable Coefficient Clustered SE P-value Constant 3.4381 0.6750 0.000*** BANKSIZE -0.0805 0.0470 0.088* 37 SOLVENCY -0.8274 0.2668 0.002*** COSTINCOME -0.0228 0.0012 0.000*** INFLATION -0.0077 0.0034 0.024** UNEMPLOY -0.0377 0.0076 0.000*** BASEL 0.2815 0.1159 0.015** BANKSIZE x BASEL -0.0156 0.0081 0.055* F-score 61.84 0.000*** R2 (Within) 0.2890 R2 (Between) 0.2772 R2 (Overall) 0.2677 *** p < 0,01; ** p < 0,05; * p < 0,1 Appendix 9.4. Robustness test for Model 1 Regression Output for Model 1 with RE Variable Coefficient Clustered SE P-value Constant 2.0394 0.18997 0.000*** BANKSIZE -0.0159 0.0089 0.076* NIM 0.1039 0.0195 0.000*** SOLVENCY -0.5199 0.1568 0.001*** COSTINCOME -0.0211 0.0011 0.000*** INFLATION -0.0089 0.0029 0.002*** GDP 0.0083 0.0019 0.000*** UNEMPLOY -0.0249 0.0056 0.000*** BASEL 0.1068 0.0206 0.000*** Wald-test 622.81 0.000*** 38 R2 (Within) 0.2969 R2 (Between) 0.4354 R2 (Overall) 0.3619 *** p < 0,01; ** p < 0,05; * p < 0,1 Appendix 9.5. Robustness test for Model 2 Regression Output for Model 2 with RE Variable Coefficient Clustered SE P-value Constant 1.8979 0.2329 0.000*** BANKSIZE -0.0061 0.0122 0.617 NIM 0.1051 0.0196 0.000*** SOLVENCY -0.5379 0.1571 0.001*** COSTINCOME -0.0211 0.0012 0.000*** INFLATION -0.0088 0.0029 0.003*** GDP 0.0082 0.0019 0.000*** UNEMPLOY -0.0251 0.0055 0.000*** BASEL 0.2715 0.1215 0.025** BANKSIZE x BASEL -0.0116 0.0082 0.158 Wald-test 625.74 0.000*** R2 (Within) 0.2976 R2 (Between) 0.4337 R2 (Overall) 0.3614 *** p < 0,01; ** p < 0,05; * p < 0,1 39 Appendix 9.6 Robustness test for Model 1 Regression Output for Model 1 with Regular Standard Errors Variable Coefficient SE P-value Constant 2.5044 0.2896 0.000*** BANKSIZE -0.0401 0.0190 0.035** NIM 0.1072 0.0126 0.000*** SOLVENCY -0.7754 0.1059 0.000*** COSTINCOME -0.0209 0.0005 0.000*** INFLATION -0.0123 0.0025 0.000*** GDP 0.0081 0.0016 0.000*** UNEMPLOY -0.0382 0.0032 0.000*** BASEL 0.0950 0.0152 0.000*** F-score 327.90 0.000*** R2 (Within) 0.3000 R2 (Between) 0.3615 R2 (Overall) 0.3237 *** p < 0,01; ** p < 0,05; * p < 0,1 Regression Output for Model 1 with Robust Standard Errors Variable Coefficient Robust SE P-value Constant 2.5044 0.7239 0.001*** BANKSIZE -0.0401 0.0471 0.395 NIM 0.1072 0.0252 0.000*** SOLVENCY -0.7754 0.2478 0.002*** COSTINCOME -0.0209 0.0013 0.000*** INFLATION -0.0123 0.0036 0.001*** GDP 0.0081 0.0019 0.000*** 40 UNEMPLOY -0.0382 0.0078 0.000*** BASEL 0.0950 0.0216 0.000*** F-score 66.39 0.000*** R2 (Within) 0.3000 R2 (Between) 0.3615 R2 (Overall) 0.3237 *** p < 0,01; ** p < 0,05; * p < 0,1 Appendix 9.7 Robustness tests for Model 2 Regression Output for Model 2 with Regular Standard Errors Variable Coefficient SE P-value Constant 2.3904 0.2913 0.000*** BANKSIZE -0.0318 0.0192 0.097* NIM 0.1092 0.0126 0.000*** SOLVENCY -0.8353 0.1074 0.000*** COSTINCOME -0.0209 0.0005 0.000*** INFLATION -0.0119 0.0025 0.000*** GDP 0.0079 0.0016 0.000*** UNEMPLOY -0.0393 0.0032 0.000*** BASEL 0.3569 0.0802 0.000*** BANKSIZE x BASEL -0.0184 0.0055 0.001*** F-score 293.18 0.000*** R2 (Within) 0.3013 R2 (Between) 0.3444 R2 (Overall) 0.3140 *** p < 0,01; ** p < 0,05; * p < 0,1 41 Regression Output for Model 2 with Robust Standard Errors Variable Coefficient Robust SE P-value Constant 2.3904 0.7341 0.001*** BANKSIZE -0.0318 0.0478 0.505 NIM 0.1092 0.0253 0.000*** SOLVENCY -0.8353 0.2544 0.001*** COSTINCOME -0.0209 0.0013 0.000*** INFLATION -0.0119 0.0036 0.001*** GDP 0.0079 0.0019 0.000*** UNEMPLOY -0.0393 0.0077 0.000*** BASEL 0.3569 0.1208 0.003*** BANKSIZE x BASEL -0.0184 0.0082 0.025** F-score 59.27 0.000*** R2 (Within) 0.3013 R2 (Between) 0.3444 R2 (Overall) 0.3140 *** p < 0,01; ** p < 0,05; * p < 0,1 Appendix 9.8 Robustness tests for the overrepresentation of German banks Regression Output for Model 1 when excluding German banks Variable Coefficient Clustered SE P-value Constant 2.6049 0.8185 0.002*** BANKSIZE -0.0347 0.0536 0.518 NIM 0.1075 0.0296 0.000*** SOLVENCY -0.8959 0.2744 0.001*** COSTINCOME -0.0223 0.0014 0.000*** INFLATION -0.0115 0.0044 0.009*** 42 GDP 0.0073 0.0021 0.000*** UNEMPLOY -0.0389 0.0081 0.000*** BASEL 0.1245 0.0271 0.000*** F-score 66.94 0.000*** R2 (Within) 0.3290 R2 (Between) 0.5611 R2 (Overall) 0.4297 *** p < 0,01; ** p < 0,05; * p < 0,1 Regression Output for Model 2 when excluding German banks Variable Coefficient Clustered SE P-value Constant 2.4927 0.8368 0.003*** BANKSIZE -0.0264 0.0549 0.631 NIM 0.1087 0.0297 0.000*** SOLVENCY -0.9361 0.2790 0.001*** COSTINCOME -0.0223 0.0014 0.000*** INFLATION -0.0113 0.0044 0.010** GDP 0.0072 0.0020 0.000*** UNEMPLOY -0.0398 0.0081 0.000*** BASEL 0.3567 0.1328 0.008*** BANKSIZE x BASEL -0.0164 0.0092 0.077* F-score 60.28 0.000*** R2 (Within) 0.3299 R2 (Between) 0.5488 R2 (Overall) 0.4243 *** p < 0,01; ** p < 0,05; * p < 0,1 43 Appendix 9.9 Robustness tests for the overrepresentation of German banks Regression Output for Model 1 with only German banks Variable Coefficient Clustered SE P-value Constant 3.4776 1.4665 0.019** BANKSIZE -0.1743 0.0961 0.072* NIM 0.0422 0.0375 0.262 SOLVENCY 0.1099 0.2903 0.706 COSTINCOME -0.0103 0.0030 0.001*** INFLATION -0.0021 0.0040 0.602 GDP 0.0003 0.0027 0.906 UNEMPLOY -0.0231 0.0143 0.110 BASEL 0.0083 0.0257 0.748 F-score 14.02 0.000*** R2 (Within) 0.1377 R2 (Between) 0.0870 R2 (Overall) 0.0640 *** p < 0,01; ** p < 0,05; * p < 0,1 Regression Output for Model 2 with only German banks Variable Coefficient Clustered SE P-value Constant 3.4137 1.4348 0.019** BANKSIZE -0.1697 0.0942 0.074* NIM 0.0464 0.0397 0.245 SOLVENCY 0.0072 0.2416 0.976 COSTINCOME -0.0102 0.0030 0.001*** INFLATION -0.0018 0.0043 0.678 GDP 0.0001 0.0030 0.978 44 UNEMPLOY -0.0248 0.0162 0.129 BASEL 0.1650 0.2906 0.571 BANKSIZE x BASEL -0.0110 0.0190 0.561 F-score 13.16 0.000*** R2 (Within) 0.1388 R2 (Between) 0.0813 R2 (Overall) 0.0596 *** p < 0,01; ** p < 0,05; * p < 0,1 45