Introduction

This paper aims to empirically test the validity of the two competing hypotheses regarding the relationship between competition and stability: the competition-fragility hypothesis, which views market competition as reducing stability, and the competition-stability hypothesis, which views market competition as increasing stability. Economic theory often supports the view that perfect information and market competition are associated with good performance, while the concentration of market power is usually synonymous with weak competition and bad performance.

Competition in the banking markets is also often supposed to benefit bank clients, because it increases the available choices and contributes to the lowering of the borrowing rates and the raising of the deposit rates. In addition, it is recognised that the increasing globalization and liberalisation of financial markets, together with the technological changes across the world, have changed the functional and competitive environment of the banking institutions.

On the other hand, and particularly in the aftermath of the subprime crisis, special attention was paid to the consequences of asymmetric information in financial markets. In order to prevent adverse selection and moral hazard problems, it is often recommended to increase transparency and trust, namely through the establishment of long-lasting relationships between the banking institutions and their clients. These lasting relationships based on trust are usually associated with bank market power and may also be considered as a necessary condition for bank market stability.

In the European Union (EU), the global process of liberalization was accompanied by efforts to establish the single European financial market, based on the belief that market competition would increase bank performance. The subprime crisis hit hard the European banking sector and many EU banks faced serious losses and were able to survive mainly due to the exceptional financial support from their national governments and monetary authorities. Shortly after the global financial crisis, some EU countries were also affected by the Euro area sovereign debt crisis, when troubled banks reduced their funding to governments, raising sovereign borrowing costs. The path to overcoming the crisis included an increase in the process of restructuring and consolidation of the EU banking sector, raising questions about the possibility of guaranteeing the stability of the banking institutions, accompanied by (or supported by) healthy competition in the banking market.

The paper considers a relatively large panel of 784 relevant banks from all the EU member states (although the majority of these banks were from the two biggest EU economies: Germany and France) over the years 2006–2021. It empirically tests the validity of the competition-fragility hypothesis or the competition-stability hypothesis and contributes to the literature by providing answers to the following questions:

  1. 1)

    How does bank market competition affect bank market stability in the considered sample of EU banks?

  2. 2)

    Are the results obtained for the whole panel of EU banks in line with the results obtained separately for the sub-panels of German and French banks?

  3. 3)

    How did the different crises that affect the EU banking sector during the considered period influence bank stability?

  4. 4)

    Is it possible to identify other factors (besides competition measures and crisis dummies) that contributed to the stability of the EU banking sector over the years 2006–2021?

Overall, the results obtained in the paper are consistent with the competition-fragility hypothesis. However, they also reveal that the results depend on the proxies used to measure competition in the EU banking sector as well as on potential country-specific performance and their different reactions, for example, to the crises that affected the EU banks during the considered period.

The empirical findings clearly confirm the validity of the competition-fragility hypothesis when bank competition is measured with the Boone indicator. In cases where bank competition is proxied with the Herfindahl–Hirschman Index (HHI), the results are not statistically so robust, but it is still possible to conclude that the increase in bank market concentration contributes to the stability of the German banks included in the sample. The effects of the dummies reflecting the years of the crises that affected the EU during the period 2006–2021 also reveal that the subprime crisis had a strong negative effect on the sample including all EU banks as well as on subsamples including only German or French banks. Overall, these results are confirmed for the dummies representing the other two crises, but also highlight some country-specific differences. Regarding the effects of the dummy representing the sovereign debt crisis, for the panel including only German banks, there is convincing evidence that the stability of German banks increased with this crisis. Moreover, although not so evident, it seems likely that for the sub-panel including only French banks, the dummy representing the recent pandemic crisis had a positive effect on bank stability.

Relevant Literature

Economic literature on the relationship between bank market competition and stability is far from consensus regarding the relationship between competition and stability. Two main competing hypotheses are identified. The competition-fragility hypothesis views competition as reducing stability because it encourages banks to increase risk and operate with low capital buffers, highlighting the potential trade-off between competition and stability. The competition-stability hypothesis views financial consolidation as improving stability, namely through higher capital buffers and a greater degree of diversification.

Supporters of the competition-fragility view include Hellman et al. (2000) who highlighted the inconsistency of interest-rate liberalization and prudential bank behaviour, namely because financial-market liberalization increases competition, eroding profits and contributing to the increase of moral-hazard problems. Marquez (2002) focused on the relevance of access to information, considering that with a high number of competing banks in the market, each bank becomes less informed and potentially less efficient. Dam and Zendejas-Castillo (2006) also found that a higher level of competition induces banks to invest in risky assets, confirming that risks and potential fragility are associated with competition. Beck et al. (2006) analysed the relationship between the market structure of the banking industry and bank fragility, mainly concluding that crises are less likely to occur in economies with more concentrated banking systems. Beck et al. (2013) tested the relationship between bank competition and bank stability with cross-country analyses exploring market, regulatory and institutional features. The results obtained are overall consistent with the competition-fragility hypothesis. However, they clearly reveal large cross-country variation, suggesting that an increase in competition is associated with a larger rise in banks’ fragility mainly in countries with stricter activity restrictions, lower systemic fragility, better developed stock exchanges, more generous deposit insurance, and more effective systems of credit information sharing. Horvath et al. (2016) analysed the relationship between bank competition and bank liquidity creation, concluding that increased bank competition reduces liquidity creation. They interpreted this result in terms of the effect of competition on increasing bank fragility. Ahnert and Martinez-Miera (2021) presented a model to evaluate how different developments in the banking industry, such as changes in competitive intensity or opacity, affect bank fragility, competitive structure, and welfare. Their findings are also in favour of the competition-fragility view, as shocks that increase bank competition or bank transparency contribute to increasing deposit rates, costly withdrawals, and thus bank fragility.

The competition-stability view is supported by several studies. For example, Boyd and De Nicoló (2005) concluded that as competition in bank markets increases, lending rates become lower as well as the probability of borrower default, and this improves bank profitability and stability. Similar conclusions were obtained in Boyd et al. (2009), as well as in De Nicoló and Turk-Ariss (2010), overall supporting the non-existence of an evident trade-off between competition and stability. Schaeck et al. (2009) also found that more competitive banking systems are less prone to experience systemic crisis. In addition, they considered that economic policies promoting bank market competition, if well executed, have the potential to promote stability. Schaeck and Cihák (2014) examined the effect of market competition on bank stability in the belief that competition incentivizes banks to enhance cost efficiency, increasing reallocation from unsuccessful (inefficient) banks to successful banks. They confirmed that competition robustly improves stability through an increase in efficiency, concluding also that bank capital and profitability benefit from increased competition. However, their results are heterogenous and there is clear evidence that fragile banks benefit less from competition. Also in favour of the competition-stability view is Goetz (2018) who suggested that less barriers to entry significantly contribute to the increase of bank stability in the United States, because more competition boosts bank profits and leads to the reduction of individual bank shares of non-performing loans. Martinez-Miera and Repullo (2010) utilized the Boyd and De Nicoló (2005) model, but took into account the degree of market competition, suggesting a U-shaped relationship between competition and the risk of bank failure. They also highlighted the dominant risk-shifting effect of increased competition in highly concentrated markets. More competition leads to lower loan rates, which in turn leads to lower probabilities of loan default, thus corroborating the Boyd and De Nicoló (2005) conclusions. Moreover, Martinez-Miera and Repullo (2010) identified another effect of increased competition, the margin effect, dominating in competitive markets. Lower rates reduce bank revenues from performing loans, decreasing bank profitability and stability. Similar conclusions were obtained by Claessens (2009) considering that competition in financial services is unambiguously good, but the effect on bank stability clearly depends on the degree of competition as excessive competition can compromise financial stability.

Several works support that the two competing hypotheses (the competition-fragility hypothesis versus the competition-stability hypothesis) do not necessarily lead to opposing effects of competition and market power on bank stability. For example, Berger et al (2009) considered that even if market power in the loan market results in riskier loan portfolios, the overall risks may not increase if banks adopt some appropriate risk-mitigating techniques. Their empirical results are consistent with the competition-stability view, since banks with a higher degree of market power have increased loan portfolio risk. However, they also found that banks with a higher degree of market power have less overall risk exposure, therefore supporting the competition-fragility view.

Tabak et al. (2012) also found evidence that the two competing hypotheses can hold simultaneously. They analysed the effects of bank competition on bank risk-taking using data from 10 Latin American countries between 2003 and 2008. They concluded that competition affects risk-taking behaviour in a non-linear way, as both high and low competition levels enhance financial stability, but there is an opposite effect for average competition. Moreover, the paper highlights the relevance of bank size and capitalization, suggesting that banks in competitive markets are less vulnerable, especially the larger ones, since banks with a higher capital ratio are more stable. Liu and Wilson (2013) tested the relationship between competition and stability, concluding that it depends on the initial risk of the considered banks. More precisely, they found that banks with high risks tend to avoid more risk, protecting their franchise values when competition increases, which is consistent with the competition-stability hypothesis. Alternatively, competition has a positive impact on the risk of banks with lower levels of risk, which is in line with the competition-fragility hypothesis.

Analysing the effects of competition on bank stability in the United Kingdom over the period 1994–2013, de-Ramon et al. (2018) concluded that, on average, the competition-fragility hypothesis holds as higher levels of competition tend to lower bank-level stability. However, the results are not homogeneous and are highly dependent on the underlying financial strength of the banks. More precisely, there is evidence that financially weak banks clearly benefit from greater levels of competition, as their bank profitability and capitalisation increase as a result of accelerated competition, which is consistent with the competition-stability hypothesis. Soedarmono et al. (2013) also analysed the influence of competition on stability, but focused on some Asian emerging markets and took into account the effects of crisis periods (over the years 1994–2009). Overall, their conclusions suggested the validity of the competition-fragility view, as they showed that a higher degree of market power in the banking market was associated with higher capital ratios, higher income volatility, and higher insolvency risk of banks. However, their paper also concluded that market power in banking had a stabilising impact during the crisis, in particular during the 1997 Asian crisis. Moreover, the paper suggested that the findings were dependent on the size of the largest banks as well as on the too-big-to-fail policies of the considered countries.

Brei et al. (2020) tested the relationship between bank competition and stability in 33 countries of Sub-Saharan Africa over the years 2000–2015, concluding that there is a U-shaped relationship between bank competition and credit risk. Up to a certain point, higher levels of bank competition are associated with lower credit risks. Beyond this point, more competition increases credit risks. Their findings also suggest that increased competition should be accompanied by policies specifically targeted to financial stability, and that policy makers should encourage bank competition when the banking sector is relatively concentrated, but only up to a certain limit. Similar conclusions were obtained by Cuestas et al. (2020), confirming the existence of a U-shaped relationship between competition and financial stability in a sample of commercial banks in the Baltic countries over the period 2000–2014. This paper also highlights that the manner in which the structure of the banking industry evolves is of critical importance for financial stability, suggesting that policymakers should encourage mergers and acquisitions, when competition is fierce. In contrast, they should prevent augmenting concentration in the already highly concentrated banking markets.

Leroy and Lucotte (2017) empirically investigated the relationship between competition and bank risks among a large sample of European-listed banks over the years 2004–2013 considering both individual and systemic dimensions of bank risk. The findings provide support for the two competing hypotheses. On one hand, competition seems to increase individual risk, as banks stressed with competition take more risks, in line with the competition-fragility view. On the other hand, an increase in market power is associated with more systemic risk and with an increased contribution of financial institutions to the deterioration of the banking system, which supports the competition-stability view. IJtsma et al. (2017) empirically tested the effect of concentration in 25 EU countries, during the 1998–2014 period, considering both bank-level and country-level dimensions of financial stability. They concluded that concentration minimally affects stability at both levels, suggesting that in these EU countries, neither supervisory restructuring, nor normal market-driven mergers, are likely to be substantially harmful to financial stability.

Data and Methodology

Using panel estimates, this paper tests the influence of competition on bank stability considering a relatively large panel including 784 relevant banks of all the 27 EU countries between 2006 and 2021. The banking sector stability was proxied with the estimated Z-score.Footnote 1 Following the usual procedure, the Z-score of bank i in the year t \({(Z}_{i,t})\) was computed with the expression:

$${Z}_{i,t}=\frac{{ROA}_{i,t}+{\left(\frac{E}{TA}\right)}_{i,t}}{{\sigma ROA}_{i,t}}$$
(1)

where:

\({ROA}_{i,t}\)= return on average assets (%), \({\left(\frac{E}{TA}\right)}_{i,t}\)= equity / total assets (%) = capital ratio, and \({\sigma ROA}_{i,t}\)= standard deviation of the return on average assets.

Two different measures were used to represent bank competition: the Boone indicator (which measures competition from an efficiency perspective), and the HHI (a specific measure of market concentration). Following, among others, Dutta and Saha (2021), and taking into consideration the intermediation function of the banking institutions, two Boone indicators were computed: one for bank loans (Bloans) and the other for bank deposits (Bdeposits). As the considered sample included a different number of banks from each of the 27 EU countries and the banking markets of these countries were not homogeneous, the market share of bank i, was related to the subsample of the banks in its home country (and not to the whole sample of 784 banks), in the year t. The Boone indicators are the values of the coefficients b that were obtained through the estimation of the following linear equations:

$$\begin{array}{l}{B}_{loans}:\\ ln{(Market\;share\;of\;the\;loans)}_{i,t} =\alpha +\beta\;ln{(Average\;variable\; cost)}_{i,t}\end{array}$$
(2)
$$\begin{array}{l}{B}_{depositss}:\\ ln{\left(Market\;share\;of\;the\;deposits\right)}_{i,t}=\alpha +\beta\;ln{(Average\;variable \;cost)}_{i,t}\end{array}$$
(3)

where the average variable cost was proxied with the sum of the interest expense and non-interest expense to total loans (in the case of Bloans) or to total customer deposits (for Bdeposits).

The HHI also considers the market share of bank i, in relation to the subsample of the banks of its own country. Therefore, the HHI indicates the level of bank market concentration in each of the 27 EU countries, and it was also computed separately for the market share of bank loans and bank deposits, following the usual definition:

$${HHI}_{loans}= \sum\nolimits_{i=1}^{N}{{(Market\;share\;of\;the\;loans)}_{i,t}}^{2}$$
(4)
$${HHI}_{deposits}= \sum\nolimits_{i=1}^{N}{{(Market\;share\;of\;the\;deposits)}_{i,t}}^{2}$$
(5)

All data reporting on banks’ performance were sourced from the Moody’s Analytics (2022) BankFocus database. The choice of banks considered not only data availability for the period 2006–202,1 but also bank size, as bank size is likely to affect the relationship between competition and stability (e.g., Tabak et al., 2012). Overall, banks with less than 2 billion Euros of total assets in 2021 were excluded from the sample. However, for the EU countries with few banks with a high amount of total assets, the sample includes banks with less than 2 billion Euros of total assets (but not far from 1 billion Euros in 2021). Online Supplemental Appendix (OSA) Table 1 specifies the number of the banks from each of the 27 EU countries included in the sample, and their representativeness, not only in terms of the percentage of total banks included in the whole sample, but also in terms of the percentages of total loans and total deposits to costumers. The information provided in OSA Table 1 clearly highlights the specific situations of the banks from the two biggest EU economies: France and Germany. More precisely, the French banks included in this sample, represent around 16% of the total number of banks, but collect 31% of the deposits and provide 33% of all loans. On the other hand, the German banks represent 41% of the banks included in the sample, but represent only 27% of all collected deposits and 26% of all provided loans.

Additionally, two control variables were included in the estimations: the ratio of net loans to total assets (%), and the natural logarithm of real per capita gross domestic product (GDP). Like all the other variables related to bank performance, the ratio of net loans to total assets was sourced from the Moody’s Analytics (2022) BankFocus database. The values of GDP were sourced from the World Bank (2022) Global Financial Development database. The net loans to total assets ratio provides an indication of bank liquidity, more precisely showingf how much of bank assets are tied to liquid loans. Growth of real GDP per capita is a proxy for the countries’ macroeconomic conditions.

As the paper aims to test the effects on bank market stability of the crises that affected the EU countries during the period 2005–2021, three dummies were included for the main crisis years: the global subprime financial crisis, D1 (2008–2010), the sovereign debt crisis, D2 (2011–2013), and the pandemic crisis, D3 (2020 and 2021). The estimated model is basically the following:

$${{bank stability}_{i,t}={\alpha }_{0}+\alpha }_{1 }{ bank competition}_{i,t}+{\alpha }_{2}{ net loans to total assets ratio}_{i,t} +{\alpha }_{3}{ GDP}_{j,t }+{\alpha }_{4}{ D}_{1}+{\alpha }_{5}{ D}_{2}+ {\alpha }_{6}{ D}_{3}+ \varepsilon { }_{i,t}$$
(6)

where the stability of bank i (i = 1, …784), in year t (t = 2006, …, 2021), was proxied with the natural logarithm of the Z-score. Bank market competition was measured with one of the previously mentioned competition indicators: Bloans, Bdeposits, HHIloans, HHIdeposits. GDP is the natural logarithm of the real per capita GDP of the EU country j (j = 1, …27). D1, D2, and D3 are the crisis dummies; and εi,t is the error term (OSA Table 2 presents descriptive statistics of all the variables included in the estimations and OSA Table 3 displays the pairwise correlations between all these variables).

Equation (6) was estimated first applying fixed and random effects estimations, using the values of the Hausman test to determine, in each situation, whether the fixed effects model or the random effects model is more appropriate. As clearly explained, for example, in Wooldridge (2010) and in Greene (2018), fixed effects estimations facilitate overcoming one important concern in cross-sectional studies: the potential omission of relevant control variables. Fixed effects regressions control for any time-invariant cross-sectional variable and are particularly appropriate to analyse the impact of variables that vary over time. Fixed effects explore very well the relationship between the explanatory variables and the outcome within each cross-sectional unit. A random effects model assumes that explanatory variables have fixed relationships with the response variable across all observations, and that these fixed effects may vary from one observation to another, although usually following a normal distribution. An estimation of random effects provides inference about the specific levels (similar to a fixed effect), but also population-level information.

This paper tests the robustness of the results obtained with fixed and random effects. It also uses dynamic one-step system generalized method of moments (GMM) estimations. GMM deals well with another concern in the context of the considered model: the potential existence of endogenous regressors. Dynamic GMM panel estimations not only address the endogeneity problems, but also reduce the potential bias of the estimated coefficients. The GMM method proposed, among others, by Arellano and Bond (1991), Arellano and Bover, (1995), and Blundell and Bond (1998), uses cross-country information and jointly estimates the equations in first difference and in levels, with first differences instrumented by lagged levels of the dependent and independent variables and levels instrumented by first differences of the regressors.

Empirical Results

The basic model was estimated considering separate equations, each of them including one of the proxies for bank competition: Bloans, Bdeposits, HHIloans, HHIdeposits. All the equations include the same sample of control variables and the crisis dummies. Taking into consideration not only the fact that, in terms of the GDP, Germany is the biggest EU economy and France is the second biggest one, and also the great relevance of the German and French banks included in the sample of banks considered in the estimations, the paper presents results obtained over the years 2006–2021 for the whole panel of 784 banks for 27 EU countries, as well as results obtained for the subsample of the 322 German banks, and for the subsample of the 129 French banks.

Results Using Either Fixed or Random Effects Panel Estimations

Table 1 presents the results obtained using either fixed or random effects, following the indications of the Hausman test values. The values of the F or Wald statistics provide evidence of the overall robustness of the reported results.

Table 1 Panel fixed and random effects estimation results
Full size table

There is robust statistical evidence that, in all situations, an increase in bank market competition (proxied with the Boone indicator, which measures competition from an efficiency perspective) does not favour banking market stability. More precisely, an increase in a country’s market competition, either in bank loans (Bloans) or in bank deposits (Bdeposits), clearly contributes to a decrease in banking stability (measured with the Z-score). These results are clearly consistent with the competition-fragility hypothesis.

Overall, the results obtained for the concentration measure both for bank loans (HHIloans) and for bank deposits (HHIdeposits) are not statistically robust, and do not permit very relevant conclusions about the influence of a country’s bank market concentration on banking stability, either in the whole sample or in the subsample including French banks. However, the results permit the conclusion that an increase of the concentration of the German bank market (both in terms of loans and deposits) seems to have a positive influence on banking market stability. According to the results in Table 1, the contribution of the bank liquidity situation (indicating how much of the bank assets are tied to liquid loans) to bank stability is always statistically robust. In all situations, an increase in bank liquidity has a positive influence on the banks’ Z-scores.

The statistically robust positive influence of the growth of the real GDP per capita on banking stability is also confirmed, but only for the whole panel of the 784 banks from the 27 EU countries. The results obtained for the subpanels of German and French banks are not statistically robust, with the exception of the equation including Bloans as a proxy for bank competition. However, this hold true only for the subpanel of German banks, revealing that in this particular case the increase in the growth of the real GDP per capita decreases the stability of the German banks.

Not surprisingly, the results reported in Table 1 indicate that almost always all the included crisis dummies have a clear and statistically robust negative influence on the stability of the EU banks, including the subpanel of French banks. In addition, these results point to another specific reaction of the German banks. Though the dummies representing the subprime and pandemic crises negatively affected the German banks’ stability, the dummy representing the years of the sovereign debt crisis, D2 (2011–2013) seem to have a positive influence on the stability of the subpanel of the German banks.

Results Using Panel Dynamic One-step System GMM Estimations

The results obtained for the same regressions, but now using dynamic one-step system GMM estimations, are reported in Table 2. The overall robustness of these results is confirmed by the values of the Wald tests and with the tests proposed by Arellano and Bond (1991). More precisely, the autocorrelation of the first order, AR(1), is always validated; and (with few exceptions) the autocorrelation of the second order, AR(2) is not validated. The validity of the instruments is also ensured with the values of the Sargan statistic, which is supposed to be robust to heteroskedasticity and autocorrelation.

Table 2 Dynamic one-step system GMM estimation results
Full size table

The results obtained with GMM estimations are overall in line with the results obtained with random and fixed effects estimations, namely regarding the validity of the competition-fragility hypothesis, when bank competition is measured with the two Boone indicators (Bloans and Bdeposits). Now in all situations, and not only for the subsample of the German banks, there is statistically robust evidence that the higher the bank market concentration, both in terms of HHIloans and HHIdeposits, the higher the bank market stability.

Fully in line with the results obtained using fixed and random effects estimations, the increase in bank liquidity always has a positive influence on bank market stability. Again, the influence of the other control variable, the growth of real GDP per capita, is not unanimous and raises doubts about the positive or negative influence of this variable on the stability of the considered banks.

The results obtained for the crisis dummies overall corroborate the previous conclusions. They almost always have a statistically robust negative influence on the stability of the EU banks, with the clear exception of their evident positive influence on the stability of the German banks, but only for D2, which is the dummy associated with the sovereign debt crisis.

Main Conclusions and Policy Recommendations

This paper contributes to the literature by analysing the relationship between bank market competition and stability in the EU banking sector, over the period 2006–2021. It first considers a panel including 784 banks from all 27 EU member states and then two subpanels: one including only the 322 German banks and the other including only the 129 French banks.

The results obtained using panel fixed and random effects estimations as well as dynamic one-step system GMM estimations, lead to the following conclusions. First, overall, there is a demonstration of the validity of the competition-fragility hypothesis, as higher banking market competition does not contribute to an increase in banking stability in the considered panels of EU banks. This conclusion reveals that the stability of the EU banking sector will not benefit from an increase in market competition, probably because there is already a high level of bank market competition, due not only to the global process of liberalisation but also to the specific process of economic and financial integration, namely associated with the establishment of the single European financial market.

Second, the results obtained for the whole panel including the 784 banks from the 27 EU countries are in line with those obtained for the subpanels including either only the 322 German banks or only the 129 French banks. Nevertheless, it is still possible to identify some country-specific results, namely regarding the relevance of the measure of banking market concentration, the HHI, for the stability of the German banks. More precisely, there is very clear evidence that an increase in the concentration of German loans (HHIloans) and deposits (HHIdeposits) markets has a robust positive effect on the stability of the considered German banks. This specific result reveals that despite the relevant process of European economic and financial integration, and the fact that German banks represent more than 41% of all banks considered in this paper, the behaviour of the German banks does not clearly represent the behaviour of the entire panel.

Third, overall, there is evidence that the identified crises had a negative effect on the stability of the EU banking sector. However, it is still possible to identify differences in the effects associated with specific crisis dummies as well as to the concrete subpanels. It is particularly evident that, contrary to the results obtained for the whole panel of 784 EU banks and in the subpanel of the French banks, there is very robust evidence that the dummy representing the sovereign debt crisis years (2011–2013) has a positive effect on the stability of the German banks. This result confirms that the German banks were not negatively affected by this specific crisis as Germany was not one of the EU countries facing problems associated with high sovereign debts. The findings of this paper also reveal some differences in the effects associated with the pandemic crisis dummy, namely in the subpanel including only the French banks as, at least when using GMM estimations, the results indicate that the stability of the French banking sector increased with this crisis.

Fourth, the results regarding the effects of the two considered control variables overall reveal that increasing the growth of real GDP per capita is not the most appropriate way to ensure banking stability. On the other hand, in all considered panels, there is a very convincing evidence that an increase in the net loans to total assets ratio, indicating improved bank liquidity, has a strong positive effect on the stability of the EU banking sector.

The findings of this paper reinforce the relevance of policymakers’ role and provide room for some recommendations. Banking market competition in the EU is probably already sufficiently high, and should not be reinforced as, overall, an increase in banking competition appears to be detrimental to the stability of the EU banking institutions (at least of the most relevant ones, in terms of their total assets in 2021).

The findings also highlight that, overall, policies encouraging bank mergers and acquisitions and increasing bank market concentration are not the best way to ensure the stability of the EU banking sector. However, the results also reveal some country-specific characteristics. For example, when considering only the subsample of the German banks, a higher bank market concentration is strongly recommended to increase the stability of the German banking sector. The results obtained with the dummies representing the crises that affected the EU countries over the years 2006–2021, highlight important cross-country differences. Despite the process of economic and financial integration in the EU, the member states and their banking institutions still have some individual characteristics that should be taken into account, and do not always recommend the adoption of one-size-fits-all policies.

Further research should also be encouraged, exploring the impact of different policies as well as some relevant exogenous shocks on banking systems, namely the consequences of the current high inflation rates, overall instability, and new challenges that the whole world, and particularly Europe, are facing in these turbulent years.