The Performance of Merging Cooperative Banks in Germany

1 Introduction

Over the past decade, a significant number of German banks have exited the market through mergers, which many banks see as the result of excessive regulation and low-for-long interest rates. The shrinking banking landscape thus raises potential concerns that difficult policy measures have driven ex ante good banks out of the market. This paper examines this issue by estimating the determinants of bank mergers before and during the low interest rate environment, thereby identifying the performance characteristics of merging banks in two time periods. This provides information not only on whether merging banks were good or bad performers ex ante, but also on the primary cause of the merger wave: Mergers are likely to be driven by bank inefficiency if target banks perform relatively poorly before and during the regulatory and interest rate changes and eventually exit the market during this adverse period. Conversely, increased merger activity may reflect challenging policies if target banks initially perform well or similarly to their non-merging peers, but underperform in the course of the low interest rate environment and eventually leave the market.

Given the recent increase in the number of mergers, it seems interesting to ask how adverse operating conditions are shaping and impacting the banking industry. The past decade provided for at least two crucial developments that, possibly each individually, but particularly when viewed together, resulted into a difficult surrounding. First and foremost, the low-for-long interest rate policy in the aftermath of the 2007–2009 financial crisis gradually reduced banks’ profitability and net interest margin and effectively deteriorated their financial leeway (e.g. Borio, Gambacorta, and Hofmann 2017; Busch et al. 2022; Claessens, Coleman, and Donnelly 2018; Genay and Podjasek 2014). Existing work suggests that extended periods of low or negative policy rates induce banks to adapt. Banks increase their risk-taking by lowering their loan standards (Maddaloni and Peydró 2011) and raising the portion of risk assets (Delis and Kouretas 2011; Heider, Saidi, and Schepens 2019). Furthermore, banks cut their lending activity (Heider, Saidi, and Schepens 2019), adjust their funding structure, and exchange interest-generating engagements with fee-related and trading activities (Brei, Borio, and Gambacorta 2020). Importantly, the effects are more pronounced for small banks with higher deposit shares (e.g. Claessens, Coleman, and Donnelly 2018; Genay and Podjasek 2014; Heider, Saidi, and Schepens 2019; Kerbl and Sigmund 2016; Meyer 2018). Besides triggering low policy rates, the financial crisis also spurred authorities to thoroughly tighten and expand existing banking regulations. Considering the sheer complexity and impact on the industry, the Basel III reform thereby hardly compares to prior frameworks. In fact, reforms entail substantial costs for banks to establish permissible risk-management systems and comply with novel regulatory standards and metrics (e.g. Bonner and Eijffinger 2015; Dietrich, Hess, and Wanzenried 2014; Handorf 2014; King 2013). Since regulatory burden can manifest in various aspects, such as increased spending on administration and staff training, however, no reliable quantification exists with respect to banks’ incurred cost of regulation (Cochrane 2014; Hoskins and Labonte 2015). The literature shows that banks transpose and respond to regulatory claims for higher capital and liquidity requirements by cutting their lending (De Nicolò, Gamba, and Lucchetta 2014; Mésonnier and Monks 2015) and risk-weighted assets (Gropp et al. 2019) and by employing different strategies based on their profitability (Andrle et al. 2017; Cohen and Scatigna 2016).

Taken together, banks have faced a difficult surrounding for more than a decade, evoking changes in asset composition and funding structure. Given the relatively sharp decline in the number of independently operating EU banks, however, adverse conditions not only seem to have sparked notable adaptations in financial profiles. They likely also contributed to an accelerated consolidation process in which a substantial number of banks left the market through mergers.^[1] The underlying reasoning is intuitive. Strenuous periods require an effective allocation of resources for banks to remain competitive. Banks failing to cope with detrimental circumstances are then either forced to wind up or merge with other banks. What remains unclear, however, is whether adverse conditions induced ex ante inefficient banks to exit the market in the spirit of Schumpeter (1939) (hereafter the ‘efficiency-view’), or whether they encouraged even initially well-performing banks to leave the market by negatively affecting their lending and deposit business. The uncharted issue of elevated bank exits steps into the line of questions related to the effects of ultra-loose monetary policies and stringent regulations, and is of paramount importance for comprehending developments in the banking landscape. If recent policy measures primarily turn small, albeit initially well performing, banks into merger targets, then this should be of interest to authorities. Likewise, it is important to know if consolidation occurs among low performing banks, which could help reduce over-capacities, enhance profitability, and improve resilience against shocks (ECB 2019). Given these unresolved questions surrounding the recent merger activity, this paper examines the relative performance and other key characteristics of banks that exited the market through mergers in recent years.

We do so by considering the merger dynamics in the German cooperative banking industry between 2014 and 2019. We select the German banking market because it is one of the largest in the EU, while the cooperative sector offers an abundant number of independently but similarly operating banks. This naturally mitigates concerns about potential differences in unobserved time-invariant bank characteristics (Mood 2010), and simultaneously satisfies crucial homogeneity assumptions for well-established efficiency measurement techniques (Dyson et al. 2001). The chosen time span reflects our intention to investigate critical periods. On the one hand, the Capital Requirements Regulation enters legal force in 2014, gradually imposing stricter standards on banks’ capital and liquidity management. On the other hand, stipulated refinancing rates and standing facilities continue their declining path, further reducing banks’ interest expenses but also their overall profitability. For our empirical analyses, we draw on the rich literature on bank merger determinants. As such, we first estimate a series of standard multinomial logistic models for the period 2014 to 2019, linking various performance measures to the probability of becoming a target or an acquirer. This allows important insights into consolidating banks’ performance characteristics at that juncture. Subsequently, we consider an earlier time period from 2010 to 2012, when interest rates prevail at common levels and novel regulations still remain to be poured into binding law. We reiterate our analyses in similar fashion but this time relating banks’ average performance over the period 2010 to 2012 to the probability of becoming a future target at some point between 2014 and 2019. The idea is to test whether performance, as measured before the low interest environment, significantly determines the probability of becoming a future target. If so, we can rule out the notion that the targets were initially well-performing banks that became underperforming due to the adverse policy actions and eventually disappeared from the market as a result. We provide two main findings. First, banks that perform relatively worse before and between 2014 and 2019 are more likely to exit the market during this period. Controlling for bank size and related merger determinants, the probability of becoming a target increases with a higher cost-income ratio, lower growth rates, and poorer asset management. Second, depending on the performance measure of choice, acquirers perform either similar or worse, but in no case better, than the reference group with no merger occurrence. Consolidation thus tends to occur among low performing banks, where large and well-capitalized banks take over their small and inefficient peers. Based on the findings in this paper, we conclude that the recent merger wave primarily arises from bank inefficiencies: Banks that perform relatively worse over the 2010–2012 period are more likely to become targets in the low interest rate environment.

In addition to studies analyzing the effects of banking regulations and low-for-long interest rate policies, our paper relates to several other streams of literature, including research on bank merger motives and determinants (e.g. Beccalli and Frantz 2013; Focarelli, Panetta, and Salleo 2002; Goddard, McKillop, and Wilson 2009; Hadlock, Houston, and Ryngaert 1999; Hannan and Pilloff 2009; Hannan and Rhoades 1987; Hernando, Nieto, and Wall 2009; Huhtilainen, Saastamoinen, and Suhonen 2022; Koetter et al. 2007; Lanine and Vander Vennet 2007; Moore 1996, 1997; Pasiouras, Tanna, and Gaganis 2011; Wheelock and Wilson 2000; Worthington 2004), bank distress and failure (e.g. Berger and Bouwman 2013; Berger, Imbierowicz, and Rauch 2016; Cole and Gunther 1995; Cole and White 2012; DeYoung and Torna 2013; Estrella, Park, and Peristiani 2000; Wheelock and Wilson 2000), and the cleansing effect of crises (Spokeviciute, Keasey, and Vallascas 2019). Our paper closely connects to Spokeviciute, Keasey, and Vallascas (2019), which seems to be the only perceivable work investigating the cleansing effects of financial crises within the banking industry. The work predicts the probability of bank failure or acquisition by interacting crises with banks’ cost efficiency. They find that the savings and loan crisis in the mid 1980s and early 1990s increases the exit probability for less efficient US commercial banks more than for the group of efficient banks. However, the 2007–2009 financial crisis escalates the exit probability regardless of banks’ cost efficiency. The work thus provides mixed evidence on the efficiency-view that crises encourage less efficient banks to drop out of the market. Our paper differs from the existing literature in that we examine the link between recent merger activity and (monetary) policy measures. By showing that consolidation occurs among low performing banks, we provide an interesting avenue for research analyzing the belief that “in systems with many weak-performing small banks, consolidation within their domestic system could improve performance” ECB (2019), p. 107. Related, we also extent the literature on bank merger predictability.

2 Mergers Made in Germany

The German cooperative banking industry has experienced an extraordinary increase of mergers over the past years. This is illustrated in Figure 1(a), which depicts the annual number of banks exiting the market along with the marginal lending facility as a measure of the ECB’s interest rate policy. Importantly, mergers accelerate after 2011, and peak in 2017 when 57 targets are taken over by 40 acquiring banks. These numbers can be put into perspective by comparing the recent merger activity with the situation around the 2007–2009 financial crisis. Although the crisis is known for its substantial impact on the banking industry, relatively few banks engaged into mergers during that period. Considering the entire time span of our sample from 2013 to 2020, merger dynamics induce an overall decline in the quantity of independent cooperative banks by nearly 25 %.^[2]

Figure 1:

The evolution of cooperative bank mergers in Germany. Figure (a) shows recent developments in German cooperative bank merger activity along with the lending facility. Figure (b) depicts the average net interest margin and capitalization for this industry. Shaded areas respectively indicate the 2007–2009 financial crisis and the 2014–2019 low-interest environment as considered in our analysis. The data are taken from the Bundesbank: “Bankstellenstatistik” and “Zeitreihen-Datenbanken”.

The driving motive behind these elevated merger rates could arguably be linked to the low interest environment, which exposes banks to historically low earnings. Such linkage is supported by the simple fact that mergers occur in a time when interest rates concurrently exhibit a declining path. Lower market rates, in turn, come along with a reduction of banks’ net interest margin, as shown in Figure 1(b) by the solid line. In this regard, banks might merge to encounter their decreasing margins by realizing economies of scale or scope (Amel et al. 2004). However, low interest rates may not be the only motive. The recent merger wave also falls in a period in which authorities depart towards a stricter banking regulation. Extensive regulation in terms of the Basel III framework could then impede growth and burden banks with costs (e.g. Dietrich, Hess, and Wanzenried 2014; Mésonnier and Monks 2015), although these costs may be absorbed, at least to some extent, by the compound structures of the German cooperative system. While the quantification of such regulatory costs is difficult, the impact of greater (capital) requirements can be shown clearly by an upward sloping capital ratio, as depicted in Figure 1(b) by the dashed line. Thus, banks might also merge to comply with regulations at fewer costs, for instance by profiting from a diversification of their loan portfolio (Amel et al. 2004). Apart from regulatory motives, mergers can also be based on management restructuring goals and the establishment of a new corporate culture (Gindele et al. 2019). In their recent study, Gindele et al. (2019) explicitly state that mergers among cooperative banks are primarily driven by structural changes in management, rather than attempts to improve profitability. Although the analysis is limited to banks in the German state of Baden-Württemberg between 2009 and 2016, these management motives make it generally unclear whether the target banks are underperforming or distressed.

As for the current wave of mergers, low interest rates seem to be the dominant motive. Drawing on banks’ websites, press releases, local newspaper articles that comprise interviews with bank executives, and related sources, banks mainly attribute their decision to merge to the low interest environment and an extensive set of regulations, but also to a costly layout of a digital infrastructure. More specifically, among the 228 targets subject to our empirical analysis, 211 banks justify their merger by referring to a combination of at least two out of these three elements. For the remaining 17 banks, we find no information. An exemplary statement reflecting key merger motives is taken from a press release by the Volksbank Untere Saar eG and the Vereinigte Volksbank eG Saarlouis – Losheim am See – Sulzbach/Saar:

As for the main reasons of the merger, Soester names the exuberant regulatory requirements, the altered customer behavior due to an increasing digitalization, and the impact of the ongoing low interest environment, which causes declines in earnings year after year. (Press release, accessed on the 11.01.2023 at 11:30)

Another example concerns the merger activity between the VR-Bank Hunsrück-Mosel eG and the Vereinigte Volksbank Raiffeisenbank eG:

The merger is mainly aimed at meeting the challenges of the low-interest policy and the increasing supervisory regulation. (Website, accessed on the 11.01.2023 at 11:30)

Since most statements resemble each other to a great extent, both examples can be taken as representative of the general reasoning in the population. While it could be true that some banks mask other decisive merger intentions behind statements such as the two presented, the disadvantageous nature of the low interest environment is evident. Even if there are cases where unobserved factors ultimately seal the merger, such as excessive losses leading to distressed mergers, it seems likely that such factors are again closely related to the state of profitability, or to the costs of regulation and digitalization, although the latter may again be absorbed by the prevailing network structures in the cooperative sector.

Given the high number of mergers, paired with the fact that most banks attribute their merger decision to the fierce operating situation, it appears natural to ask if such environment encourages rather low performing banks to exit the market. To obtain a first, motivating glimpse on the differences between targets and non-merging banks, Table 1 contrasts both groups with regard to crucial bank (performance) characteristics. Beginning with Panel A, which considers all bank-year observations from our primary data as discussed in the subsequent section, we observe a significant difference in total assets. Target banks are on average only half the size of non-merging banks. Moreover, non-merging banks exhibit greater growth rates, are seemingly better capitalized, profit from significantly lower NPL-ratios, and achieve more favorable cost-to-income balances. We neither observe differences in the profitability, nor in the regional economic output as proxied by the Gross Regional Product per capita. Since Figure 1(b) reveals trends in banks’ capitalization and net interest margin, however, it is likely that comparisons among related variables are flawed. This is because non-merging banks, by definition, remain throughout the sample period and are thus more exposed to these trends. To achieve a less skewed picture, Panel B additionally considers the 2017 sub-sample which is the year with the most mergers. Specifically, Panel B compares the pre-merger period of the 57 banks that become a target in 2017 with the same period (2013–2016) for the group of non-merging banks. As expected, differences in both capitalization metrics turn statistically negligible, while differences in the profitability measures revolve significant.

Overall, these preliminary findings indeed motivate the efficiency-view since targets exhibit relatively weaker profiles when compared to their non-merging peers. This is relevant from an aggregate perspective as the German banking system is often said to be overbanked, fragmented, and unprofitable (see, e.g. a speech by Dombret 2016, or Koetter et al. 2006). If recent mergers among low performing banks have accelerated the ongoing consolidation process within the industry, they might as well have reshaped the German banking market towards a less fragmented and more profitable system.^[3] Given that Table 1 also suggests notable differences in bank size, however, descriptive analyses appear insufficient to draw definitive conclusions. Instead, the underlying setting necessitates multivariate analyses to disentangle performance effects from the potential influence of other bank characteristics.

3 Data

We work with two distinct data sets. For our primary analysis, we start by gathering information on all German cooperative bank mergers between 2013 and 2020. To this end, we first screen and compare the annual directory of credit institutions, provided by the Bundesbank. The directory lists all banks that exist by the end of a corresponding year and serves as a first measure to identify target banks. We hence identify those banks as potential targets, which initially appear in the directory and stop to do so at some future point in time. Subsequently, we verify these banks as targets by using the German firm register (‘Unternehmensregister’). The firm register is a state run, publicly accessible online archive, which comprises annual statements and other qualifying information such as notices on merger activities. Banks are obliged to report their statements and related events that require disclosure to the register according to § 325 in conjunction with § 340 l of the German Commercial Code. Based on the merger notices in the firm register, we retrieve precise information on the merger year, the target bank (‘übernommenes Institut’), and the acquiring bank (‘übernehmendes Institut’). We scrutinize all mergers by examining the annual statement of each acquiring bank with regard to the financial incorporation of the respective target bank(s). Overall, we identify 228 relevant targets that are taken over by 169 acquirers within the period of 2014–2019. The remaining banks from the directory of credit institutions consequently constitute the set of non-merging entities. Note that these numbers as well as all further analyses exclude serial acquirers, banks that acquire another bank and then become a target, and banks that engage in merger activities of any kind in 2013 or 2020. These restrictions aim to address merger comparability concerns and allow for a meaningful sample split into target banks (228), acquirers (169), non-merging banks (599), and others (69). Next, we link the merger information with bank-level financial data, which we retrieve from bank disclosure reports and the German firm register. This allows us to control for a broad set of factors that have previously been shown to influence mergers. To account for regional differences in banks’ operating environment, we complement the data set with economic and demographic indicators at the county level (‘Kreisebene’). These indicators are taken from the federal and state statistical offices (‘Regionalatlas der Statistischen Ämter des Bundes und der Länder’)^[4] and comprise the gross regional product, unemployment rate, disposable income, total population and population density, and the share of the retired population (age over 65).

To understand how targets perform before the changes in regulations and interest rates, we construct a secondary sample as we lack bank disclosure reports for the years before 2013. As such, we first draw on the Bankscope data base of the Bureau van Dijk to obtain annual accounting data at the bank-level. We keep all German cooperative banks that operate between the period 2010 to 2012, and average all bank-specific variables for each bank over this period. As before, we then complement the accounting data with (averaged) macroeconomic and demographic measures at the county level. The starting point in 2010 is chosen to avoid confounding effects with the 2008 financial crisis. We limit the ending point to 2012 to mitigate distorting effects related to banks’ anticipation of the Basel III implementation. This selection leaves a sample of 1.101 German cooperative banks, of which we identify 218 banks as future targets. More precisely, we know with certainty that these 218 banks become a target at some point between 2014 and 2019.^[5] We also attempt to trace future acquirers, which proves difficult as acquirer identification often changes post-merger. Therefore, we abstract from future acquirer identification and instead distinguish future targets from the group of “other cooperatives” comprising all non-future target banks. This allows us to measure future target banks’ performance relative to all other cooperatives for the time before the low interest environment. We report summary statistics for both samples in the Appendix in Table A1. We also break down merger activity by year in Table A2.

4 Methodology

We examine the link between bank performance and the probability of becoming a target or an acquirer by first estimating a multinomial logit model based on the period 2014 to 2019, as in Hannan and Rhoades (1987), Focarelli, Panetta, and Salleo (2002), Worthington (2004), Koetter et al. (2007), Hernando, Nieto, and Wall (2009), Pasiouras, Tanna, and Gaganis (2011), Beccalli and Frantz (2013), Spokeviciute, Keasey, and Vallascas (2019). In contrast to the simple exercise in Table 1, such a model allows us to explore the role of bank performance while controlling for a wide range of factors. This is important as targets tend to be smaller than their non-merging peers, making it virtually unclear if performance actually is a merger determinant when controlling for size and other bank characteristics. Specifically, we predict event Y = i, taking values 0, 1, and 2 if a bank in a given year does not merge, becomes a target, or becomes an acquirer, respectively, based on lagged covariate vector x. For a number of i = 0, …, J groups, we denote the model as:

where P(Y = j|x) is the conditional probability of becoming a target (j = 1), or an acquirer (j = 2), given covariate vector x. Bank-year observations with no merger incident constitute the reference group, P(Y = 0|x), to which all other groups are compared. Covariates are lagged by one year and comprise one bank performance measure at a time to avoid inflated standard errors among respective measures. To alleviate potential distortions due to omitted variables, x further supplies a set of time varying, bank-specific factors that have been shown to determine bank mergers. Moreover, time-varying economic and demographic indicators at the county level capture regional differences in banks’ operating environment. Lastly, we recognize that our investigation window is characterized by a series of (unconventional) monetary policy interventions. Therefore, we also consider year dummies to control for potential portfolio re-balancing effects and other unobserved year effects (Tischer 2018).

Estimated coefficients for these variables per group, β _j , yield the change in the logged ratio of the probability of being a target (acquirer) to the probability of being in the reference group, given a unit change in the predictor variable. Put differently, the relationship holds ln[P(Y = j)/P(Y = 0)] = β _j x, which is somewhat difficult to interpret. To facilitate understanding, we discuss exponentiated coefficients representing relative risk ratios (RRR), as described in Hosmer and Lemeshow (2000) and applied by Koetter et al. (2007). The RRR allow for sensible interpretation: All else equal, a one unit increase in the predictor variable changes the probability of being in group j relative to the probability of being in the reference group by e β j − 1 × 100 % ], where e β j is the RRR of the underlying variable for group j.

The multinomial approach offers insights into how targets and acquirers perform relative to their non-merging peers during periods of low interest rates and increased regulations. We therefore understand consolidating banks’ characteristics at that juncture. However, even if the target banks turn out to be poor performers, this analysis is not sufficient to support the efficiency view. Since adverse conditions likely affect banks heterogeneously with respect to their size, they could be the main driver for target banks’ financial atrophy and ultimately their market exit. Therefore, we must understand if targets perform worse initially and exit the market later on during disadvantageous periods.

The next aim then is to relate the probability of becoming a target between 2014 and 2019 to the average performance prior to the emergence of adverse factors. If ex ante performance significantly predicts future targets, this could show that adverse conditions are not the primary cause for target banks’ weak profile in the more recent years. Here, we consider the 2010–2012 period as a suitable window to measure ex ante performance for two reasons. First, the starting point offers sufficient temporal distance to the financial crisis, allowing banks to digest the most immediate impacts of the crisis. Second, the ending point ensures that banks still operate under earlier regulations with less stringent requirements and at rather common policy-rates. Hence, we relate the probability of being a future target to banks’ average performance over the period 2010 to 2012. We estimate a standard binary logistic model (e.g. Hernando, Nieto, and Wall 2009; Huhtilainen, Saastamoinen, and Suhonen 2022):

where P(Y = FutureTarget) is the probability that a bank becomes a target in the future. FutureTarget is a dummy equal to unity if the bank is acquired at some point between 2014 and 2019, and zero otherwise. The covariate vector x is the same as in equation (1), except for the consideration of year effects. Likewise, vector x is shown with a bar to indicate averaged values. Estimated coefficients yield the change in log odds given a unit-change in the predictor variable. As before, we thus discuss exponentiated coefficients to measure changes in terms of odds ratios (Hosmer and Lemeshow 2000).

Before proceeding with a discussion of our variable selection, we finally address an issue concerning the influence of unobserved mergers during the period 2010 to 2012. Since we lack acquirer information over this period, observations with realized mergers enter the reference group and may distort estimations in (2). In particular, acquisitions between 2010 and 2012 may affect the likelihood of becoming a target later on. We tackle this issue by considering two variations of (2). In the main estimations shown, we proxy acquirers by applying a threshold level of 20 % on asset growth. In practice, this means that we remove banks with annual asset growth in excess of 20 %. We find this value appropriate, given that 99 % of the non-merging banks in our 2014–2019 sample exhibit annual growth rates of less than 16 %, while 75 % of the acquirers in their merger year show growth rates greater than 27 %. This procedure removes 45 pseudo acquirers. In unreported estimations, we then test the robustness by keeping all sample banks. We find that either approach yields similar results.

5 Results

This section addresses our empirical results. We begin by estimating variations of equation (1) for the period 2014 to 2019 to understand if bank performance significantly associates with the probability of turning into a target or an acquirer. Subsequently, our focus shifts to an earlier point in time, when interest rates prevail at somewhat common levels, and new regulations just begin to unpack their influence on the industry. The idea is to analyze how target banks perform prior to low market rates and fiercer regulations. This could show that adverse conditions are not the primary cause for target banks’ low performance. Instead, initially low performing banks may have a greater probability to exit the market later on.

6 Conclusions

This paper examines merger dynamics in the German cooperative banking industry between 2014 and 2019 to determine whether recent mergers are due to bank inefficiencies or to challenging policy measures. Motivated by the unusually large number of bank mergers in a period with low-interest rates and demanding regulations, we specifically test merging banks’ relative performance before and during this period. Using standard bank merger methodology, the results consistently indicate that targets are small and underperforming banks, while acquirers stand out in terms of size and capitalization. Thus, the evidence in this paper supports the efficiency view that ex ante inefficient banks increased their merger activity during the low interest rate environment.

In addition to these findings, authorities may wonder if bank merger benefits outweigh the drawbacks from customers’ perspective. Although we leave this topic to future research, we do point at potential concerns related to the regional accessibility of financial services, personal advisory, and consumer surplus. However, at least for the mergers under investigation, we do not find that mergers per se pose a threat to financial access and personal advisory. In most of the cases, acquirers maintain targets’ branch network along with their personnel. Related issues might rather arise from the digitalization-side, automating various services and, thereby, obsoleting traditional branches.

Based on the findings in this paper, we finally conclude that the spectrum of bank merger motives may extend beyond existing hypotheses, such as the inefficient management hypothesis, stating that acquirers take over their inefficient peers to improve management (e.g. Jensen and Ruback 1983; Manne 1965; Palepu 1986). In this sense, elevated merger rates during crises or adverse conditions could signal waves of consolidation, with weakly performing banks joining forces to stay in business for the long term.