سرمایه چگونه بر عملکرد بانک در طول بحران مالی تاثیر می گذارد؟

This paper empirically examines how capital affects a bank’s performance (survival and market share) and how this effect varies across banking crises, market crises, and normal times that occurred in the US over the past quarter century. We have two main results. First, capital helps small banks to increase their probability of survival and market share at all times (during banking crises, market crises, and normal times). Second, capital enhances the performance of medium and large banks primarily during banking crises. Additional tests explore channels through which capital generates these effects. Numerous robustness checks and additional tests are performed.

The recent financial crisis raises fundamental issues about the role of bank equity capital, particularly from the standpoint of bank survival. Not surprisingly, public outcries for more bank capital tend to be greater after financial crises, and post-crisis reform proposals tend to focus on how capital regulation should adapt to prevent future crises. Various such proposals have been put forth recently (e.g., Kashyap et al., 2008, BIS, 2010, Acharya et al., 2011, Admati et al., 2011, Calomiris and Herring, 2011 and Hart and Zingales, 2011). An underlying premise in these proposals is that externalities exist due to the safety net provided to banks and, thus, social efficiency can be improved by requiring banks to operate with more capital, especially during financial crises. Bankers, however, often argue that holding more capital would jeopardize their performance and lead to less lending. The academic literature suggests that this bankers' perspective needs to be more nuanced (e.g., Aiyar et al., 2012, Jiménez et al., 2012 and Osborne et al., 2012), but has pointed out some negative consequences of more capital as well (e.g., Diamond and Rajan, 2001). Given the divergent views in the literature, the issue of the effects capital has on bank performance, the magnitude of these effects, and how they might differ across different types of crises and normal times boils down to an empirical question, one that we confront in this paper. In particular, the goal of this paper is to empirically examine the effects of bank capital on two dimensions of bank performance—probability of survival and market share—during different types of financial crises and normal times. Survival and market share are two key performance issues that concern bank managers. Bank survival is central not only in strategic decisions made by banks, but also in decisions made by regulators concerned about banking stability. Market share is an important goal for most firms (e.g., Aghion and Stein, 2008), and banks often assess their performance relative to each other on this basis. Knowing how bank capital affects bank performance, both during financial crises and normal times, is also of paramount importance for regulators contemplating micro- and macro-prudential banking regulation.2 In particular, comprehending whether higher capital has a significant effect on a bank's survival likelihood and how this effect differs depending on bank size and the nature of the crisis are important details for regulators who are weighing the level and other specifics of capital requirements to achieve a desired level of banking stability. Even though the battle for market share is a zero-sum game, it matters to regulators because it affects bank behavior. For example, if higher capital impeded a bank's pursuit of market share, it might encourage higher leverage and greater banking fragility, something of concern to regulators. These issues also matter for how banking theory evolves, because it helps bring about a better appreciation for the reasonableness of assumptions about the channels through which bank capital affects various aspects of bank performance. Most theories predict that capital enhances a bank's survival probability. Holding fixed the bank's asset and liability portfolios, higher capital mechanically implies a higher likelihood of survival. A deeper justification is provided by incentive-based theories such as Holmstrom and Tirole (1997), Acharya, Mehran, and Thakor (2011), Allen, Carletti, and Marquez (2011), Mehran and Thakor (2011), and Thakor (2012). In these models, either capital strengthens the bank's incentive to monitor its relationship borrowers, reducing the probability of default, or it attenuates asset-substitution moral hazard, or it lessens the attractiveness of innovative but risky products that elevate the probability of financial crises. However, some theories suggest that under certain circumstances increasing bank capital could be counterproductive because it perversely increases bank risk taking (e.g., Koehn and Santomero, 1980 and Besanko and Kanatas, 1996). Nonetheless, the reviews in Freixas and Rochet (2008) suggest that the scales are tilted in favor of the prediction that capital has a salutary effect on the probability of survival. The view that capital strengthens a bank's competitive position in asset and liability markets, which can also improve its odds of survival, is also buttressed by the empirical evidence in papers such as Calomiris and Mason (2003) and Calomiris and Wilson (2004). Recent banking theories also suggest a positive relation between capital and market share (e.g., Allen et al., 2011 and Mehran and Thakor, 2011). The empirical evidence suggests that higher-capital banks are able to compete more effectively for deposits and loans (e.g., Calomiris and Powell, 2001, Calomiris and Mason, 2003, Calomiris and Wilson, 2004 and Kim et al., 2005), providing some support. In contrast, the literature on the interaction between a nonfinancial firm's leverage and its product-market dynamics argues that more highly-levered firms compete more aggressively for market share, suggesting that the relation between capital and market share could be negative (e.g., Brander and Lewis, 1986). Thus, while existing theories provide valuable insights that guide the testable hypotheses we formulate in this paper, the predictions they produce conflict in some cases, pointing to the need for empirical mediation. Moreover, even when the theories strongly predict an effect in one direction, much is to be learned from documenting the sizes of various effects and how these vary in the cross section of banks, which again calls for empirical analysis. Furthermore, the theories generally do not distinguish between financial crises and normal times and do not distinguish between banks of different size classes, although these distinctions are important from a policy perspective and for the empirical tests in this paper. For both survival and market share, we take our cue from the theories and formulate hypotheses that allow us to assess whether capital helps or hurts. The hypotheses are tested using data on virtually every US bank from 1984:Q1 until 2010:Q4. We examine small banks (gross total assets, or GTA, up to $1 billion), medium banks (GTA exceeding $1 billion and up to $3 billion), and large banks (GTA exceeding $3 billion) as three separate groups, because the effect of capital likely differs by bank size (e.g., Berger and Bouwman, 2009).3 We also recognize that not all financial crises are alike. A crisis that originates in the banking sector could differ in impact from one that originates in the capital markets. To make this distinction, we define banking crises to be those that originated in the banking sector; and market crises to be those that originated outside banking in the financial markets. We study the effect of capital during banking crises, market crises, and normal times. We also examine the effect of capital during individual crises. We test the effect of capital on bank survival using logit regressions. We regress the log odds ratio of the probability of survival on the bank's precrisis capital ratio interacted with a banking crisis dummy, a market crisis dummy, and a normal times dummy. Recognizing that coefficients on interaction terms in nonlinear models cannot be interpreted in the same way they are in linear models (Norton, Wang, and Ai, 2004), we use marginal effects to determine the effects of capital on survival. Our results indicate that capital enhances the survival probability of small banks at all times and, in the case of medium and large banks, only during banking crises. The survival regressions also include a broad set of control variables to mitigate potential omitted-variable problems. The controls include not just proxies for risk and opacity, size and safety net protection, location, and profitability, but also for competition [including multi-market contact as in Evans and Kessides (1994); Degryse and Ongena (2007)], ownership (Berger et al., 2000, Giannetti and Ongena, 2009 and Mehran et al., 2011), and organizational structure and strategy (e.g., Degryse and Ongena, 2005, Degryse and Ongena, 2007, Bharath et al., 2007 and Degryse et al., 2009). We test the effect of capital on market share by defining market share in terms of the bank's share of aggregate gross total assets. We regress the percentage change in market share on the bank's average precrisis capital ratio interacted with the banking crisis, market crisis, and normal times dummies, and a set of control variables similar to the one mentioned above. Our results indicate that capital helps to increase market shares for small banks at all times and for medium and large banks only during banking crises. We perform a variety of robustness checks. First, we use regulatory capital ratios instead of the equity-to-assets ratio to define capital. Second, we drop banks that could be considered to be too-big-to-fail (TBTF) to see if our large-bank conclusions are driven by the dominance of a few very large institutions. Third, we use an alternative cutoff to separate medium and large banks to examine the sensitivity of our results to the manner in which banks are classified by size. Fourth, we measure precrisis capital ratios averaged over the four quarters before the crisis or one quarter before the crisis instead of averaging them over the eight quarters before the crisis. Fifth, while the theories suggest a causal relation from capital to performance, we recognize that in practice both could be jointly determined. Although our main regression analyses use lagged capital to mitigate this potential endogeneity problem, we also address it more directly using an instrumental variable (IV) approach. Our main findings stand up to all these robustness checks. In an extra analysis, we investigate whether banks with higher precrisis capital ratios are able to improve their profitability. In conducting these analyses, we aggregate across banking crises and treat them as a group, aggregate across market crises and treat them as a group, and group normal times together as well. We use this approach because we want to draw some general conclusions about how the role of capital differs across banking crises, market crises, and normal times, while minimizing the impact of the idiosyncratic circumstances surrounding a particular crisis. However, it is also useful to examine each crisis individually because doing so permits a more granular look at how the role of capital changes from one crisis to the next. We therefore also run the regressions separately for each of the five crises and normal times. The small-bank results based on examining crises individually are qualitatively similar to those when crises are grouped together, but the individual crisis analysis does yield additional insights for medium and large banks. Having established the effects of capital on performance, we turn our attention to understanding the channels through which these effects work. The theories suggest higher-capital banks engage in more monitoring and invest in safer assets, so we identify three potential channels through which enhanced monitoring and safer investment policies could affect performance: growth in noncore funding, on-balance-sheet relationship loans, and off-balance-sheet guarantees. Our results also raise new questions that we address through additional analyses designed to develop a more textured understanding of the effects of capital. First, to sharpen our understanding of the survival results, we examine whether the manner of exit for nonsurviving banks [mergers and acquisitions (M&As) without government assistance, M&As with assistance, and outright failure or change of charter] is related to their precrisis capital ratios. Second, we assess whether the impact of capital on market share differs across banks with different growth strategies (organic growth versus growth via M&As). The results of all these analyses can be summarized as follows. First, higher precrisis capital is associated with a higher survival probability for small banks at all times (banking crises, market crises, and normal times) regardless of whether crises are considered individually or grouped together. This result holds whether capital is defined as total equity or as regulatory capital. For medium and large banks, this result holds only for banking crises, in particular the credit crunch of the early 1990s. During the recent crisis, unprecedented regulatory intervention that primarily benefited medium and large banks seems to have substituted for precrisis bank capital. Second, for banks that do not survive a crisis, precrisis capital affects the manner of exit. Banks with higher capital are less likely to exit via a government-assisted M&A and more likely to exit via an unassisted M&A. Third, higher precrisis capital is associated with a gain in market share for small banks at all times and for medium and large banks during banking crises, regardless of whether crises are considered in groups or individually. For small and medium banks, the market-share effect is stronger when growth is organic, whereas for large banks the effect is stronger when growth is via M&As. Fourth, the effects of capital appear to be manifested through all three channels we examine. Banks with higher capital before a crisis show higher growth in noncore funding, on-balance-sheet relationship loans, and off-balance-sheet guarantees during the crisis. Interestingly, this seems to hold for small banks at all times, and for medium and large banks primarily during banking crises, precisely the cases in which capital helps banks survive and improve their market shares. Finally, for small banks, higher precrisis capital is associated with higher profitability at all times, and for medium and large banks only during financial crises. Our approach is a significant departure from the existing empirical literature, which typically does not differentiate among bank size classes and studies either the credit crunch of the early 1990s (e.g., Estrella, Park, and Peristiani, 2000) or the recent subprime lending crisis (e.g., Beltratti and Stulz, 2012, Berger et al., 2012 and Cole and White, 2012). Our results strongly suggest that the distinction among size classes and between different types of financial crises as well as the contrast with normal times is important. Specifically, we find that higher capital seems to unambiguously benefit small banks in all circumstances. It enhances their survival probability and market share during banking crises, market crises, and normal times, whether the individual episodes are grouped together or considered one at a time. Capital helps medium and large banks largely during banking crises, but these results are more circumstance-dependent. Our interpretation is that size could be a source of economic strength for a bank, just like capital, and that each has diminishing marginal value. Hence, capital offers the greatest benefit to small banks, which could be viewed as being endangered at all times, while medium and large banks are challenged mainly during banking crises. This is consistent with the fact that small banks have steadily lost market share to medium and large banks and survived less often since the mid-1980s. The remainder of this paper is organized as follows. Section 2 develops the empirical hypotheses. Section 3 explains our approach, discusses the financial crises and normal times, describes the variables and the sample, and provides summary statistics. Section 4 discusses the main results based on grouping banking crises, grouping market crises, and grouping normal times. Section 5 includes the robustness tests. Section 6 revisits the main results and robustness checks by analyzing individual crises. Section 7 examines three channels through which capital could affect performance. Section 8 contains the additional analyses. Section 9 concludes.