Deregulation of Bank Entry and Bank Failures

Transcription

1 Deregulation of Bank Entry and Bank Failures Krishnamurthy Subramanian Indian School of Business Ajay Yadav Fuqua School of Business, Duke University April 14, 2012 Abstract Does deregulation of bank entry enhance bank stability or exacerbate bank fragility? Theoretically, the effect of deregulation of bank entry on bank failures is ambiguous. Using the deregulation of entry restrictions in the U.S. states, we show that deregulation enhances bank stability by lowering instances of bank failures. Placebo effects do not account for our results: the deregulation had no effect on thrift failures. Preexisting bank failures in a state did not determine its timing of deregulation, which assures against reverse causal effects. The benefits of deregulation are strongest in environments where market structure changed the most and stem from geographical diversification, operating efficiencies and reduced loan losses. Keywords: Banks, Banking Crises, Bank Failure, Competition, Consolidation, Crisis, Deregulation, Entry JEL Codes: G01, G21, G28, G33 We would like to thank Phil Strahan for his inputs regarding the Garn-St. Germain Act of 1982 and Viral Acharya, N Prabhala, Raghuram Rajan and Luigi Zingales for their valuable comments and suggestions. Please address correspondence to Krishnamurthy Subramanian at Krishnamurthy_Subramanian@isbE.edu. Ajay Yadav would like to thank the Indian school of business for research support. The usual disclaimers apply. Ph.D. (Finance) student at Fuqua School of Business, Duke University starting in Fall 2012

2 Deregulation of Bank Entry and Bank Failures Abstract Does deregulation of bank entry enhance bank stability or exacerbate bank fragility? Theoretically, the effect of deregulation of bank entry on bank failures is ambiguous. Using the deregulation of entry restrictions in the U.S. states, we show that deregulation enhances bank stability by lowering instances of bank failures. Placebo effects do not account for our results: the deregulation had no effect on thrift failures. Preexisting bank failures in a state did not determine its timing of deregulation, which assures against reverse causal effects. The benefits of deregulation are strongest in environments where market structure changed the most and stem from geographical diversification, operating efficiencies and reduced loan losses. Keywords: Banks, Banking Crises, Bank Failure, Competition, Consolidation, Crisis, Deregulation, Entry JEL Codes: G01, G21, G28, G33

3 I Introduction The financial crisis of 2008 highlighted the debilitating effect that a disruption in the banking sector can have on the macro-economy. Not surprisingly, therefore, preserving banking sector stability has occupied center stage in regulatory and policy making circles; the enactment of the Dodd-Frank Act of 2010 represents a case in point. As well, blame for the recent financial crisis has been laid, at least partially, on increased deregulation of the banking sector. For example, Igan, Mishra, and Tressel (2011) and Igan and Mishra (2011) provide evidence that lobbying by the financial industry played a role in making the regulatory environment lax, which allowed the lenders to engage in riskier lending. In this context, an important question arises: does deregulation lead to instability in the banking sector? In this overarching theme, in this paper, we examine the effect of the deregulation ofbankentryonbankstability. In theory, the effect of the deregulation of bank entry on bank stability is ambiguous. On the one hand, theoretical models provide opposing predictions about the effect of bank competition on the stability of banks. On the other hand, deregulation of bank entry may result simultaneously in greater competition in the local banking market as well as greater consolidation among banks in the economy. As Beck (2008) argues, while seemingly opposing trends, consolidation does not necessarily imply less competition, as consolidation can take placeacrossdifferent business lines or markets or create fewer, but more competitive, players. Since the effects of greater competition and greater consolidation on bank failures oppose each other and since the theoretically predicted effect of competition on bank failures remains ambiguous, the effect of deregulation of bank entry on bank failures remains an open question. In this paper, we investigate the effect of the regulation of bank entry on bank failures by exploiting the staggered deregulation of bank entry across the U.S. states since the 1970s. We use data on bank failures from FDIC s Historical Statistics on Banking together with the data on deposits and assets of banks from Bank Call Reports. We employ two main proxies for bank failures: (i) proportion of all banking sector deposits owned by failing banks; and (ii) proportion of all banking assets owned by failing banks. Unlike the total deposits/assets accounted for by failing banks, these normalized measures have the advantage of not being affected by the expansion/contraction of the banking sector in a particular state. Using this data, we investigate whether deregulation of bank entry fosters bank stability or exacerbates bank fragility. Some theoretical models argue that a less concentrated banking sector is more prone to bank failures than a more concentrated one. First, as theoretically modeled by Marcus (1984), Greenbaum and Thakor (1986), Keeley (1990), Besanko and Thakor (1993), Hellmann, Murdoch and Stiglitz (2000), and Matutes and Vives (2000), 1

4 in concentrated banking systems, banks may exercise greater market power and generate greater profits. Since banks value their banking charter considerably in such environments, they may take systematically lower risk and are more likely to survive. Second, banks earn fewer rents in a more competitive environment, which may reduce their incentive to properly screen borrowers, thereby enhancing the likelihood of bank failure (Allen and Gale (2000, 2004) and Boot and Greenbaum (1993)). In contrast, other theories contend that increased competition reduces bank failures. First, Klein (1974), Hayek (1978, 1990) argue that competition among banks enhances efficiencies, strengthens the banking system and thereby decreases bank failures. Second, after deregulation, banks can diversify their assets across different geographies and thereby limit their risks (Gart (1994), Hubbard (1994)). Finally, Boyd and De Nicolo (2005) contend that since borrowers respond endogenously to the high interest rates charged by monopolistic banks, borrowers would assume greater risks, thereby endangering monopolistic banks. As well, competition among banks forces managers to improve the quality of loans by improving screening and monitoring and avoiding less productive loans (Jayaratne and Strahan, 1996). To examine the effect of deregulation of bank entry on bank failures, we exploit the fact that states in the U.S. gradually lifted geographic restrictions on branching in different years. Therefore, we can use states that have not deregulated till a particular point in time as a control group to control for potentially confounding effects at the state level. Furthermore, the panel structure of our dataset also allows us to control for any unobserved differences among states. Figure 1 shows the average year-on-year change in the deposits of liquidated banks over a 25-year window surrounding the year of deregulation. It is quite clear from figure 1 that the percentage of deposits lost due to bank failures decreased a year after deregulation and persisted for almost a decade after deregulation. To examine this before-after change in an econometrically robust manner, we undertake our empirical analysis in the following steps. First, to investigate concerns relating to reverse causality, we examine whether prior bank failures in a given state help to predict the timing of deregulation in that state. While we find that the political factors identified by Krozner and Strahan (1999) continue to predict the timing of branching deregulation, bank failures have no explanatory power by themselves. In fact, even in figure1above, wedonotfind any pre-existing pattern of bank failures before deregulation. After reassuring ourselves against possible reverse causality, we examine the effect of the regulation on future bank failures, specifically up to three years after deregulation. We estimate this effect as a difference-in-difference: the difference in the proportion of deposits (and assets) lost due to bank failures in a state before and after the deregulation, compared to the same difference for states that did not undergo a deregulation during the same period. 2

5 Figure 1: Change in the deposits of liquidated banks before and after branch deregulation We estimate this effect in panel regressions that include state and year fixed effects to control respectively for time-invariant unobserved factors affecting bank failures at the state level as well as time trends. We also include several lags of the growth of per capita GDP in the state. We find that bank failures are negatively correlated with deregulation of bank entry. This correlation, however, could be driven by omitted variables. For example, technological changes that may have led to deregulation could also have led to the reduction in bank failures if such technological changes enabled banks to screen/monitor borrowers better. To address such concerns, we analyze possible placebo effects by examining the effect of bank deregulation on failure of thrifts. Since the Garn-St. Germain Act of 1982 had permitted thrifts to branch across state borders, bank deregulation did not affect the entry barriers faced by thrifts. Furthermore, any economy-wide factors that may have affected bank failures would affect failure of thrifts as well. Therefore, failure of thrifts represent a placebo effect in our setting. We find that branching deregulation did not have any effect on thrift failures, which mitigates concerns regarding omitted variable bias. Having alleviated concerns about reverse causality and omitted variable bias, we conclude that the regulation of bank entry indeed enhanced bank stability by reducing bank failures. The economic effect is large: after deregulation, the decrease in the proportion of deposits lost due to bank failures reduced annualy by 23.6% of its standard deviation. Furthermore, the effect of deregulation on bank failures is quick and does not die down with time. Consistent with the effect of the regulation being greater in environments where the structure of the banking market changed more, we find that intra-state deregulation primarily 3

6 accounts for the decrease in bank failures. As Strahan (2002) explains, only intra-state deregulation had a significant effect on the structure of the banking markets; inter-state deregulation had no effect. Furthermore, we find that the effect of deregulation is centered around the unit banking states. Since these states were subject to the most restrictive environment before deregulation, the structure of banking markets change the most in these states when compared to those that allowed limited branching before deregulation. Next, we study the effect of the deregulation on failures of banks of different sizes. We divide the banks that failed into four quartiles according to their size and investigate the effect in each of these categories separately. We find that post deregulation bank failures declined significantly in each of the four size categories; furthermore, the effects are quite uniform across the different size categories. With respect to the channels through which the regulation reduced bank failures, our evidence together with those found by prior studies suggest three main sources. First, banks benefited from the greater opportunities available for geographic diversification. Byretard- ing the natural evolution of the industry, branching restrictions had reduced the performance of the average banking asset by restricting better banks from lending across different geographies and reaping the benefits from such diversification. Since branching was prohibited altogether in the unit banking states, branching deregulation expanded opportunities for diversification in these states when compared to states that allowed limited branching. Therefore, our finding that bank failures reduced significantly in unit-banking states but not in states that allowed limited branching supports this diversification hypothesis. Second, banks became more efficient post-deregulation as better banks grew at the expense of their less efficient rivals. Jayaratne and Strahan (1998) find that operating costs of banks decreased by about 8% after intra-state deregulation. Stiroh and Strahan (2003) find that although better-performing banks grew faster than under-achievers before intrastate branching was allowed, low-cost, high-profit banks grew even faster once branching restrictions were lifted. Furthermore, the lifting of restrictions on branching enabled banks to exploit economies of scale. Our evidence that deregulation reduced failures for banks of all size categories suggests that these efficiency gains were reaped by banks of all sizes. Third, bank failures reduced due to lower loan losses suffered by banks. Given greater competition following deregulation, most of the reduction in banks costs were passed along to bank borrowers in the form of lower loan rates. Average loan rates fell by about 19 basis points in the short run and by 30 basis points in the long run (Strahan, 2002). As Boyd and De Nicolo (2005) argue, since borrowers respond endogenously to the interest rates charged by banks, the lower interest rates charged may have reduced moral hazard by borrowers. As well, competition among banks forced managers to improve the quality of loans by improving 4

7 screening and monitoring and avoiding less productive loans (Jayaratne and Strahan, 1996). The decrease in loan losses by about 50% after intra-state branching deregulation, as pointed out by Jayaratne and Strahan (1998), is consistent with these forces being at play. We also rule out the following alternative interpretations. First, it could be argued that our results are a manifestation of the mechanical effect of increased mergers and acquisitions post deregulation. In particular, inefficient banks may have been acquired by better run banks resulting in a weeding out of weak banks from the system. Since small banks are more likely to merge or be acquired, the reduction in bank failures should have been greatest in the smallest size category in this case. The fact that our results are equally strong across all size categories stands in contrast to this alternative explanation. Second, our results could be a manifestation of possible lobbying by large banks for deregulation. If large banks lobbied for deregulation since they were most likely to benefit fromthesame(attheexpenseof small banks), then the effect of deregulation should largely be restricted to the large banks. However, we do not find that to be the case either. Third, our results could be due to a secular decrease in economic volatility coinciding with the period of the deregulation. If this were the case, the reduction in bank failures should have been the maximum for the small sized banks since an economy-wide shock is more likely to sink the small sized banks than the larger banks. However, the fact that the effect is strongest for the largest banks is inconsistent with this alternative. The remainder of the paper is organized as follows. Section III presents background information and develops our hypotheses. Section IV describes the data and proxies. Section V describes our empirical methodology and reports the results. Section VI discusses the possible channels for the documented effect. Section VII concludes. II Review of Literature We contribute to the literature examining the effect of the deregulation of bank entry on bank stability. Our study differs from the above studies (Jayaratne and Strahan (1996, 1998), Stiroh and Strahan (2003)) as we examine the effect of banking deregulation on bank failures directly. While the decrease in loan losses or the increases in the quality of bank loans, in bank profitability or the strengthening of the relationship between lagged bank performance and its market share are potential channels through which bank failures may have reduced failures post deregulation, these proxies do not directly capture the risk of bank failures. For example, two banks with similar quality of loan portfolios and magnitude of loan losses may be more or less likely to fail depending on how well the bank is capitalized. In fact, Keeley (1990) finds that greater competition among banks reduces bank capital. If bank capital had reduced significantly due to the deregulation of bank entry, banks failures 5

8 may have increased post-deregulation despite the reduction in loan losses and the operational improvements. Our study shows that bank deregulation indeed enhanced bank stability by reducing instances of bank failures. Our evidence contrasts with that of Keeley (1990), who argues that increased competition decreases bank stability. Using data on bank holding companies and using a bank s Tobin s Q as a measure of the bank s market power, Keeley (1990) finds that the market-value-capitalto-assets ratio of a bank increases with the market power of the bank. Keeley interprets this as evidence that increased competition decreases the bank s charter value and thereby increases the likelihood of bank failures. Our evidence contrasts with that of Keeley (1990) for multiple reasons. First, we examine bank failures directly. Second, the deregulation of bank entry led not only to increased competition but also to greater consolidation among banks (Strahan, 2002). Since the effect of increased consolidation on bank failures would be opposite that of greater competition, the evidence in Keeley (1990) cannot extrapolate to the effect of banking deregulation. Third, change in bank competition also affects borrower behavior and thereby the extent of loan losses, on the one hand, and banks efforts at achieving efficiencies on the other hand. Our study also relates to work examining the effect of branching restrictions on bank failures during the Great Depression era. Studies using state- and county-level data find that states allowing branching had lower failure rates (Mitchener 2005, Wheelock 1995). Studies of individual banks find contrasting results. Calomiris and Mason (2003) and Carlson (2004) find that branch banks were more likely to fail than other banks. In contrast, Carlson and Mitchener (2009) find that unit-banks that had to compete with larger branching networks improved their efficiency and profitabilty and were more likely to survive the Great Depression. Wheelock and Wilson (2000) use Kansas banks during the depression and find that inefficient banks were more likely to fail than efficient banks. 1 Since the deregulation of entry restrictions by U.S. states in the 1980s resulted simultaneously in greater competition in the local banking markets as well as greater consolidation among banks at the state level (Strahan, 2002), the results obtained for the Great Depression era do not necessarily extend to the deregulation of entry restrictions by U.S. states in the 1980s. Moreover, by addressing concerns relating to reverse causality and omitted variable bias, we are able to identify better the effect of the deregulation of bank entry on bank failures. Other prior work examines the impact of removal of entry restrictions on other economic outcomes. Jayaratne and Strahan (1996) analyzed the impact of intra-state branching deregulation on growth and found that states grew faster after deregulation. Black and Strahan 1 In international evidence, Beck, Demirguc-Kunt and Levine (2006) found that probability of crisis increases with increase in the fraction of applicants denied entry as a fraction of total applicant pool. 6

9 (2002) and Kerr and Nanda (2009) explore the impact of these regulations on entrepreneurship: Black and Strahan (2002) found that the rate of new incorporations increased after intra-state deregulation while Kerr and Nanda (2009) find that both entrepreneurship and business closures increased following inter-state deregulation. Black and Strahan (2001) document that the wage gap between male and female executives decreased and women s share of employment in managerial positions increased in the U.S. banking sector following deregulation. We contribute to this literature by providing evidence of the effect of bank deregulation on bank failures. This study also relates to the broader issue of regulation in the banking sector. Kroszner and Rajan (1994) indicate that Glass-Steagal Act of 1933, which led to the separation of commercial and investment banking, constrained the bank s ability to underwrite high quality securities without providing any significant benefits. Also, Ramirez (1999) points out that this regulation might have raised the costs of raising external funds for investment purposes by the corporations. Aharony and Swary (1981) analyzed the impact of the 1970s amendment to the Bank Holding Company Act that allowed banks to participate in the non-banking activities. Their analysis suggests that the amendment had no effect on either BHC s risk or profitability. Demirguc-Kunt and Detragiache (2002) and Demirguc-Kunt and Huizinga (2004) provide cross-country evidence on the adverse effects of deposit insurance. They suggest that explicit deposit insurance lowers market discipline for risk taking and increases the likelihood of banking crises. III III.A Background and Empirical Hypotheses Deregulation of Entry Restrictions by U.S. states The banking sector in the U.S. has been highly regulated historically. Until the 1970s, almost all states restricted within state branching, which meant that a bank cannot expand within the state boundaries by opening new branches. In the 1970s and 1980s, most states passed laws removing the restrictions on the ability of banks to open or acquire new branches. Two classes of restrictions were eased over this period. First, intra-state deregulation allowed banks to expand within the passing state either by acquiring other bank branches or by setting up new bank branches themselves. Second, inter-state branch banking deregulation allowed banks to acquire branches in other states with which their home state had negotiated such a bilateral agreement. Due to the reciprocal nature of these agreements, most states undertook inter-state deregulation in the mid 1980s to early 1990s. These state-level reforms culminated in the Riegle-Neal inter-state Banking and Branching Efficiency Act of 1994, which allowed national inter-state branch banking after

10 III.B Deregulation of Entry and Banking Market Structure The prior literature suggests that branching deregulation led to substantial changes in the structure of banking markets. Consolidation occurred as large multibank holding companies (MBHCs) acquired banks and converted existing subsidiaries into branches (McLaughlin, 1995). Between 1988 and 1997, both the number of banks and the number of banking organizations fell by almost 30%. As a result, the share of total assets held nationwide by the largest eight banking organizations rose from 22.3% to 35.5% (Berger, Demsetz and Strahan, 1999). The share of banking assets held by banks with assets under $100 million decreased from 24% in mid-70s to 15% in the mid-90s. Despite the consolidation, the Herfindahl- Hirschmann index of concentration in local markets has remained stable. As well, according to Black and Strahan (2002), such consolidation did not necessarily reduce competition in the local banking markets. Thus, deregulation increased consolidation at the aggregate level as well as competition at the level of the local banking market. In fact, as Beck (2008) argues, while seemingly opposing trends, consolidation does not necessarily imply less competition, as such consolidation can take place across different business lines or markets or create fewer, but more competitive players. Among the two different forms of deregulation intra-state and inter-state the literature argues that only intra-state deregulation had a significant effect on the structure of the banking market. In contrast, banking structure did not change much after the inter-state banking deregulation. For instance, Amel and Liang (1992) find significant entry into local markets through de novo branching after the removal of intra-state branching restrictions. However, they do not find much change after inter-state deregulation. Similarly, Jayaratne and Strahan (1998) find that operating costs and loan losses decrease sharply after intrastate branching restrictions are lifted; however, the effects of the removal of inter-state branch regulation are not strong. III.C Deregulation of Entry and Bank Failures Theoretically, the effect of the deregulation of bank entry on bank failures is ambiguous. Some studies argue that monopolistic banks earn more profits and reduce the chances of a banking crisis by using these higher profits as capital buffers to be used at the time of some negative macroeconomic shock. In monopolistic banking systems, bank charters are valuable and these high charter values deter bank s management from making risky investments as failure could result in the loss of the valuable charter (see Marcus (1984), Greenbaum and Thakor (1986), Keeley (1990), Besanko and Thakor (1993), Hellmann, Murdoch and Stiglitz (2000), and Matutes and Vives (2000) and Boyd et al. (2004)). However, there is a counterview that unregulated banking reduces bank fragility. Klein 8

11 (1974), Hayek (1978, 1990) argue that competition among banks enhances efficiencies, strengthens the banking system and thereby decreases bank failures. Along these lines, Stiroh and Strahan (2003) find that increased competition forced bank managers to cut down costs, improve the quality of service and enhance efficiency. Those banks that failed to do so shrank or exited the system and were replaced by strong banks, which strengthened the banking system as a whole. These effects together constitute the efficiency channel through which banking deregulation can reduce bank failures. After deregulation, banks could diversify their assets across different geographies and thereby limit their risks (Gart (1994), Hubbard (1994)). We label this the diversification channel. Finally, as pointed out by Boyd and De Nicolo (2005), banking deregulation increases credit market competition, which leads to reduction in loan rates for borrowers and, thereby, increases the ability of the borrowers to repay. Also, decreased loan rates reduce the moral hazard problem on part of the borrowers by reducing their temptation to invest in risky projects. The reduced moral hazard and increased inability to repay should together reduce the chance of default by the borrower. As well, competition among banks forces managers to improve the quality of loans by improving screening and monitoring and avoiding less productive loans (Jayaratne and Strahan, 1996). We label this the loan loss channel. Given the contrasting predictions, we develop the following hypotheses: Hypothesis 1A: Deregulation of bank entry leads to more bank failures. Hypothesis 1B: Deregulation of bank entry leads to less bank failures. As highlighted in Section III.B, intra-state deregulation significantly affected banking market structure while inter-state deregulation did not. Therefore, we predict that: Hypothesis 2: While intra-state deregulation affects bank failures significantly, interstate deregulation does not affect bank failures significantly. Branching deregulation is likely to have more effect on banking structure in unit banking states as these states were subject to the most restrictive environment. Therefore, posted deregulation, banking structure should change more in the unit banking states when compared to the states that allowed limited branching. Therefore, we hypothesize that: Hypothesis 3: Deregulation affects bank failures disproportionately more in states that had unit banking laws than in states that allowed limited branching before deregulation. IV Data and Proxies This section describes our data and proxies. 9

12 IV.A Bank Failures We employ data on bank failures from FDIC s Historical Statistics on Banking (HSOB). For each year, HSOB provides information on the type of failure (liquidation or assistance), charter type (bank or thrift), location (state) as well as the total deposits and assets owned by each institution as of the last Call Report or Thrift Financial Report filed by the institution before the failure. Although data on bank failures is available from 1934, we start our sample in 1976 since data on assets and deposits from the call reports is available only from Also, even though the data is available for later years, following Black and Strahan (2002), we choose to end our sample in The passage of Riegle-Neal inter-state Banking and Branching Efficiency Act (IBBEA) in 1994 allowed nationwide branching for all FDIC approved banks, which make state-level estimates of banking sector variables unreliable. We classify bank failures by the method of its resolution by FDIC. These fall into three categories. In a type I resolution, the institution s charter is preserved and FDIC assists it through open-bank assistance or other means. A type II resolution refers to an assisted merger by FDIC, where the acquiring bank purchases the distressed bank s assets and assumes its liabilities; in this case, the distressed bank s charter is discontinued. Finally, a type III resolution refers to the complete closure of the institution. In this type of resolution, all the assets and liabilities of the institution are liquidated. Thus, type III failures would have a greater detrimental effect on the economy than type II failures. As type I resolutions only involve assistance by FDIC, they do not necessarily rank as a failure. Since theeconomicimpactduetotypeiresolutionsmaybeminimal,weexcludetypeifailures from our analysis. We use two proxies for bank failures: (i) the total deposits accounted for by failed banks in a particular state, year as a proportion of the total deposits owned by all banks in that state-year; and (ii) total assets accounted for by failed banks in a particular state, year as a proportion of the total banking assets in that state-year. The dollar value of total deposits/ assets accounted for by failing banks is inappropriate since it changes due to the banking sector in a state expanding or contracting. Similarly, any measure based on the number or proportion of failing banks is inadequate because it fails to take into account the fact that failure of larger banks may be economically more material than failure of small banks. Figure 2 and 3 show the time trends in bank failures across several states in the U.S. Since the states of Texas, Alaska, Oklahoma, Louisiana, Massachusetts, Connecticut, Tennessee and Florida experienced the largest number of bank failures, we examine the trends in these states individually. Figure 2 displays the trend in proportion of deposits and assets owned by failing banks in these eight states and the rest of the U.S. Except for Louisiana and 10

13 Tennessee, failures were negligible at the start of the sample period and peaked in the late 80s and early 90s. In Louisiana (Tennessee), failures were high in the mid-70s and peaked in the early 80s (late 80s). The largest number of bank failures occurred in the state of Texas. In Figure 3, we find a similar trend in type II and type III failures as well. Table I reports the summary statistics for the variables. IV.B Deregulation of Bank Entry We use the year of inter-state and intra-state deregulation as in Jayaratne and Strahan (1996). Our data is available for fifty U.S. states and the District of Columbia. Following the practice in the literature, we drop Delaware and South Dakota from our analysis; the banking system in these two states is very different from rest of the U.S. due to the presence of credit card banks and laws favouring the credit card industry. 2 IV.C Control Variables Bank failures are less likely in economies that are experiencing significant economic growth as compared to the economies that are not. To control for this effect, we calculate growth rate of per capita Gross State Product (GSP) using data from the Bureau of Economic Analysis and include upto three lags of this variable. V V.A Results Empirical Strategy We investigate whether deregulation of bank entry leads to lower bank failures. Since bank failures may be expected to be largely collinear with other state-level unobserved economic factors, we exploit state-level exogenous differences in the timing of branching deregulation. Specifically, since geographic restrictions on banks were lifted gradually and states deregulated at different times, we can use states that did not deregulate to control for potentially confounding effects and thereby estimate a difference-in-difference: the difference in the level of bank failures in a state before and after the regulation compared to this difference for states that did not undergo a deregulation during the same period. To fix ideas, consider two states that deregulated at different points in time: New York (NY) in 1976 and Massachusetts (MA) in Since MA had not deregulated till 1984, the before-after difference in bank failures in NY due to the deregulation in 1976 when compared against the before-after difference over the same period for MA provides a causal estimate of the effect of the deregulation in NY in This is because the before-after difference 2 As a robustness test, in unreported tests, we repeat our analysis after including these two states and find that all our key results remain unchanged. 11

14 in bank failures in MA provides a reasonable estimate for the counterfactual question: what would have been the bank failures in NY if the deregulation had not happened in 1976? Our empirical strategy unfolds in the following steps. First, if the pre-existing pattern of bank failures in a particular state affects the timing of deregulation in the state, then our analysis would be flawed given concerns of endogeneity. Therefore, we test the validity of this assumption by examining whether prior bank failures affected the timing of branching deregulation. Second, if the timing of the deregulation is uncorrelated with pre-existing bank failures in the state, we can estimate the effect of deregulation on bank failures using a panel regression: j=k X ln(y s,t+k )=β s + β t + β 1 intra st + β 2 inter st + γ j X s,t+j,k =1, 2, 3 (1) where Y s,t+k denotes a measure of bank failure in state s at time t + k and intra st and inter st aredummyvariablesthatequaloneintheyearsafterstates deregulates and zero otherwise. The coefficients, β 1 and β 2, capture the difference-in-difference estimate of the impact of intra-state and inter-state deregulation respectively on bank failures. X st is a vector of time-varying state level variables. β s and β t denote state and year fixed effects that control respectively for state-specific, time-invariant and year-specific, state-invariant unobserved characteristics that affect Y s,t+k. Third, despite the difference-in-difference estimation, omitted variables may affect our results. For example, technological changes that may have led to the deregulation of bank entry, could also account for the reduction in bank failures if such technological changes enable banks to screen borrowers better. To address such concerns, we analyze possible placebo effects by examining the effect of the state-level deregulation on failure of thrifts. Since the Garn-St. Germain Act of 1982 had permitted thrifts to branch across state borders, bank deregulation did not affect the entry barriers faced by thrifts. Furthermore, any technological changes that may have enabled banks to screen better should allow thrifts to do the same as well. Similarly, any economy-wide factors that may have affected bank failures would affect failure of thrifts as well. Therefore, failure of thrifts represent a placebo effect in our setting. Fourth, we focus on the mechanism that leads to this effect. Since we argued that the change in banking structure as a result of deregulation led to the subsequent effect on bank failures, we examine the relative effectofderegulationinenvironments wherethebanking structure change substantially when compared to environments where it did not change very much. In particular, we use the presence of unit banking laws in a particular state, as compared to the states that allowed limited branching before deregulation, to proxy for the j=1 12

15 differing impacts on banking structure. In these tests, we provide third-difference estimates for the effect of deregulation on bank failures. As argued above in Section III.C, intra-state deregulation impacted the structure of banking markets while inter-state deregulation did not. Therefore, we interact the presence of unit banking laws in a particular state with intra-state deregulation: j=k X ln(y s,t+k )=β s + β t +(β 1 UnitBank s + β 2 ) intra st + γ j X s,t+j,k =1, 2, 3 (2) where UnitBank s equals one for unit banking states and zero otherwise. V.B Pre-existing Bank Failures and Timing of Deregulation Our analysis rests on the assumption that the timing of branch deregulation was unaffected by the pre-existing pattern of bank failures. Figure 4 shows that neither the level of bank failures before deregulation nor its rate of change prior to deregulation explains the timing of branch deregulation. In a regression of the year of deregulation on the average proportion of deposits lost due to bank failures before deregulation or on the rate of change in the proportion of deposits lost in the years before deregulation, the t-statistic on these bank failure proxies are and 0.747, respectively. Using a hazard model to study the determinants of deregulation, we provide additional evidence that prior bank failures did not affect the timing of deregulation. We follow Kroszner and Strahan (1999) and employ a Weibul hazard model where the dependent variable is the log of expected time to branch deregulation given that the state has not already deregulated. Once a state deregulates it automatically drops out from the analysis. The results of the hazard model are reported in Table II, where we report the results using the percentage of deposits owned by failing banks to proxy the existing patterns of bank failures. In Table A-1 in the Internet Appendix, we report the results using the percentage of assets owned by failing banks. Since the results are almost identical using the percentage of assets lost, in the main text, we only report the results using the percentage of deposits lost. In all the regressions we control for the growth rate of per capita GDP in the state and political economy factors from Kroszner and Strahan (1999) that where found to influence the timing of deregulation. The political economy factors include: (1) share of small banks among all banking assets in a state; (2) capital ratio of small banks relative to large banks in the state; (3) an indicator that takes value 1 if banks may sell insurance; (4) relative size of insurance markets in states where banks may sell insurance; (5) relative size of insurance markets in states where banks may not sell insurance; (6) share of small firms among firms in a state; (7) percentage of state governments controlled by Democrats; (8) an indicator 13 j=1

16 thattakesavalue1ifthestateiscontrolledbyoneparty;(9)averageyieldonbankloans in the state minus Fed funds rate; (10) an indicator for A state having unit-banking laws; and (11) an indicator that takes the value 1 if state changed insurance powers. We find that prior bank failures did not affectthelikelihoodthatastatethathasnot already deregulated does so in a particular year. Among the political economic factors, first, we find that states where small banks own greater share of the banking assets/deposits deregulated later. Second, states where larger firms operated were more likely to deregulate later. Third, deregulation was likely to be later if the state was controlled by one party. Fourth, states that had unit banking laws were more likely to deregulate later. These results are very similar to those obtained in Kroszner and Strahan (1999). V.C Effect of Deregulation on Bank Failures In this section, we assess the impact of branch deregulation on bank failures. Table III presents the result of the test of regression equation (1). Panels A, B and C respectively show the effect of deregulation on our dependent variables three, two and one year after deregulation. Since deregulation should have a long-run effect on bank failures, we first examine the effect on bank failures three years after deregulation in Panel A. Examining the effect at least one year after deregulation enables us to mitigate concerns related to simultaneity or reverse causality. Each regression includes up to three years of per capita growth rate of GDP apart from state and year fixed effects. Also, we estimate standard errors that are clustered by state to account for possible autocorrelation in the residuals. The results in Table III indicate that while intra-state branching deregulation led to a drastic reduction of bank failures, inter-state deregulation did not have much of an impact. The coefficient on intra-state deregulation dummy is negative and statistically significant in all the specifications in Table III. When we compare the coefficient of interest across panels A, B and C for each specification to examine the dynamic effect of deregulation on bank failures, we find that the effect across time is almost identical. In fact, this pattern was reflected in figure1aswell,whereweobservedthattheeffect of the regulation on bank failures stayed similar and extended till 15 years after deregulation. Thus, we observe that the effect of the deregulation of entry on bank failures manifests immediately (i.e. the year after the regulation) and prevails in the long-run as well. TableA-2intheInternetAppendixconfirms these findings using the proportion of assets owned by failing banks as a proxy for bank failures. The effect on proportion of assets owned by failing banks is similar to the proportion of deposits owned by these failing banks both statistically and economically. In each of the panels in Table III, we find that the coefficient on per-capita growth rate of GDP is negative and significant, which suggests that bank 14

17 failures were lower in states with higher per-capita growth rate of GDP. This result is in line with expectations since high growth implies better business opportunities for banks. The result that bank deregulation is associated with large declines in both proportion of deposits and assets owned by failing banks is consistent with Hypothesis 1B: banking deregulation reduced bank fragility. As well, we find that only intra-state deregulation had a significant effect on bank failures; inter-state deregulation is not associated with any change in bank failures. These results are consistent with Hypothesis 2 and the findings in the literature that banking structure change substantially (did not change) after intra-state (inter-state) deregulation. V.D Omitted Variable Bias and Possible Placebo Effects Despite the difference-in-difference estimation, a residual concern remains that our results are driven by omitted variables. To address this concern, we analyze possible placebo effects using the failure of thrifts as a placebo. As discussed in Section V.A, thrifts are a suitable placebo in our analysis as they are similar to banks in their functioning. However, bank deregulation did not apply to thrifts. In Table IV, we present the results of our placebo test. The Garn-St. Germain Act, which permitted states to branch across state borders, was passed in Therefore, we start our sample for these tests in The specification is as in equation (1), where Y s,t+k now denotes a measure of thrift failures in state s at time t + k. Columns 1-3 show the effect of branching deregulation on proportion of deposits owned by failing thrifts 3, 2 and 1 year after deregulation respectively. The coefficient on branching deregulation dummy is statistically insignificant at any reasonable level of confidence in all the three specifications. Moreover, in column 2, which analyzes the effectofderegulationonproportionofdeposits two years after deregulation, the dummy is positive. Table A-3 in the Internet Appendix shows the result of branching deregulation on proportion of assets owned by failing thrifts. As in the case with deposits, the branching deregulation dummy is not significant in any of the specifications. These results provide strong evidence that our results are not driven by omitted variables and strengthen our main finding that branching deregulation reduced bank fragility. V.E Economic Magnitudes Having alleviated concerns about reverse causality and omitted variable bias, we can assess the economic magnitude of the effect of deregulation on bank failures. Consider column 1 in panel A of Table III showing the effect on type II failures. To interpret the resultsincolumn1notethatthemeanofproportionofdepositsownedbybanksinvolved 15

18 in type II failures is and the standard deviation is The coefficient of on the intra-state deregulation dummy implies that the proportion of deposits owned by failing banks in this category reduced annually by 84.4% (= exp( 1.859) 1) three years after deregulation. However, since the proportion of bank deposits of failing banks equals on average, which is quite small, year-on-year percentage changes are misleading. Instead, we state the economic magnitudes as a percentage of the standard deviation of the proportion of bank deposits of failing banks. Therefore, intra-state deregulation led to a reduction in proportion of deposits owned by failing banks involved in type II failures by 24.2% (=0.844 * / ) of the standard deviation three years after deregulation. Similarly, the coefficients of , and in columns 2, 3 and 4 in Panel A imply that three years after deregulation, proportion of deposits held by banks involved in type III failures and all failing banks fell by 10.1% and 23.6% of the standard deviation respectively. Clearly, the magnitude of reduction in failures is economically significant. V.F Dynamic Impact of Deregulation on Bank Failures In this section, we look at the dynamics of the relationship between deregulation and bank failures by including a series of dummy variables in our basic regression model: Y st = β s + β t + β 1 D 10 st + β 2 D 9 st β 25 D +15 st + st, (3) where the deregulation dummies equal zero except as follows. D j equals one for states in the j th year before deregulation and for j = 10 and equals one for years that are 10 or more before deregulation. D +j equals one for states in the j th year after the deregulation and for j =15 and equals one for years that are 15 or more after deregulation. Y st is defined as before. β s and β t the present state and year fixed effects as before. Year of deregulation is excluded from these regressions and thus the dynamic effect of deregulation is relative to the year of deregulation. Figure 5 plots the results and the 95% confidence intervals using the percentage of deposits lost as the dependent variable. Figure A-1 in the Internet Appendix plots the results using the percentage of assets lost. These figures reaffirm the conclusions of Table II that the decrease in our two proxies of bank failures took place only after the deregulation. The figure shows no clear trend in our dependent variables before the years of deregulation. All our dependent variables decrease immediately after deregulation and continue decreasing with some fluctuations through the next fifteen years. Thus, the effect of deregulation on bank failures is quick and does not die down with time. This suggests that the channels through which bank deregulation affects bank failures operate fast and they continue operating long after bank deregulation. 16

19 V.G Relative Effects in Unit Banking vs. Other States In Table V, we explore the possibility of whether deregulation had different effects on bank failures in states that had unit banking laws as compared to the states that allowed limited branching before deregulation. If deregulation caused the above reduction in bank failures through its effect on the structure of banking markets, then the effect of deregulation should be stronger in unit banking states that were subjected to more restrictive regulations. This is because after the removal of restrictions, the structure of banking markets should change more in these states. Table V presents the results of estimating equation (2). The results in Table V provide evidence consistent with Hypothesis 3. In fact, we find that all the reduction in bank failures following deregulation took place only in states that had unit banking laws prior to deregulation with no significant reduction in bank failures for limited branching states. Table A-4 in the Internet Appendix confirms these findings using the proportion of assets owned by failing banks as a proxy for bank failures. These results confirm that the impact of branching deregulation on bank failures occurred through changes in the structure of banking markets. V.H Relative Effect on Banks of Different Sizes Next, we examine the effect of branch deregulation on banks of different sizes. Figure 6 displays the effect of deregulation on bank failures for each decile based on bank size, where size is measured using the deposits lost. We notice that band deregulation reduced bank failures have banks of all sizes with the effect being statistically significant for the bottom three, fifth, seventh and top deciles. Next, to test for this effect in the regression setting, we divide banks into four quartiles based on their size (as measured by their deposits). Table VI displays the results for the impact of deregulation on failing banks of different sizes, where the rank the failing banks into the four quartiles based on their deposits. The coefficient on branching deregulation dummy is negative and significant for all but the topmost quartile. Next,welookattherelativeeffect of bank deregulation separately for banks of different sizes in the unit banking states vis-à-vis those states that allowed limited branching. The results are reported in Table VII, where we find that the coefficient of the interaction between intra-state deregulation dummy and unit banking dummy is negative and significant whereas the coefficient on branching deregulation dummy is insignificant in all the regressions. This suggests that the reduction in bank failures across all size categories was restricted to the unit banking states. Since the reduction in bank failures centered primarily around the unit banking states (as seen in Table V), we can conclude that the reduction in bank failures was uniform across all sites categories. 17

20 VI Channels In this section, we discuss the possible channels through which branch deregulation could have led to the observed reduction in bank failures. VI.A Evidence on the Efficiency Channel Branch deregulation intensified competition in the local banking markets by allowing new banks to enter freely and existing banks to expand unhindered. This increased competition forced bank managers to cut down on costs, increase efficiency and improve quality of banking services. Jayaratne and Strahan (1998) show that operating costs decreased by 8% after intra-state deregulation. Their analysis also suggests that these cost reductions took place because better run banks grew at the expense of their less efficient rivals. Banks that failed to do so either exited or contracted and their market share was taken up by better run banks. Moreover, removal of restrictions on expansion allowed banks to grow to their efficient size and achieve the benefits of economies of scale. As large banks could achieve greater economies of scale, they were able to reduce their costs more as compared to smaller banks. The evidence in Tables 6 and 7 that the reduction in bank failures spread across all size categories suggests that the operating efficiencies, i.e. cost reductions, and economics of scale possibly manifested across banks of all sizes. VI.B Evidence on the Diversification Channel By prohibiting banks from opening branches, banking regulations had denied them the opportunity to better diversify their assets (Gart 1994, Hubbard 1994). Calomiris (2000) suggests that banking regulations led to banking instability by creating small banks that could not diversify easily, which made banks vulnerable to local macroeconomic downturns and portfolio shocks. By removing the restrictions on branching, bank deregulation gave banks the opportunity to operate in several local markets. This could have led to a reduction in bank failures by helping banks to better sustain the effect of local macroeconomic shocks. Our result in Table V implying that bank failures reduced significantly only for unit bankingstatessupportstheabovelineofreasoning. Asbanksintheunitbankingstateswere allowed to open only one branch, banks in these states were the least diversified. Therefore, by allowing banks to open new branches, branching deregulation led to the greatest increase in diversification opportunities for banks located in these states. Also, the results in Table VIII show that the decline in bank failures in the unit banking states spread across banks of all size categories. These results provide strong support for the diversification channel. 18

21 VI.C Evidence on the Loan Loss Channel Bank deregulation reduced bank failures possibly by decreasing loan losses of banks. Competition forces managers to improve the quality of loans by improving screening and monitoring and avoiding less productive loans. Consistent with such an effect, Jayaratne and Strahan (1996) show that the fraction of non-performing loans, fraction of loans written-off and fraction of loans to insiders, i.e. executives and principal shareholders, declined after deregulation. As well, increased competition forces managers to lower interest rates charged to borrowers. Strahan (2002) documents that average loan rates fell by about 19 basis points in the short run and by 30 basis points in the long run. As Boyd and De Nicolo (2005) argue, a decrease in the interest rate improves borrower s ability to repay and reduces borrower moral hazard thereby reducing loan losses. The decrease in loan losses by about 50% after intra-state deregulation (Jayaratne and Strahan, 1998) is consistent with such a phenomenon. VI.D Mechanical effects of M&A among Banks? We have shown that intra-state deregulation reduced bank failures. After intra-state deregulation, banks grew mainly by acquiring other banks located in the same state and merging them into their own branching network. It could be argued that our results are a manifestation of the mechanical effect of increased M&A post deregulation. Specifically, inefficient banks may have been acquired by better run banks resulting in a weeding out of weak banks from the system. Wheelock and Wilson (2000) explored this channel and found that, on the contrary, high cost inefficiency lowered the probability of a bank being acquired. As well, since small banks are more likely to merge or be acquired, the reduction in bank failures should have been greatest in the smallest size category in this case. The fact that our results manifest across all size categories, as seen in Table VII, contrasts with this alternative explanation. Thus, we can infer that within-state M&A contributed to the benefits that banks generated by diversifying geographically and thereby reduced bank failures. VI.E Benefits from an Active Market for Corporate Control? Banking regulation had possibly weakened the market for corporate control by prohibiting takeovers of inefficient banks by better run banks and thereby worsened the agency problems between bank shareholders and their managers. Deregulation increased the threat of takeover by removing these restrictions and led to improved managerial incentives to operate their banks efficiently (Berger, Kashyap and Scalise, 1995). Consistent with a more active market for corporate control among banks, Hubbard and Palia (1995) find evidence of increase in 19

22 CEO turnover and the sensitivity of CEO compensation to performance after deregulation. Strahan (2002) shows that acquisitions increased sharply following inter-state deregulation but there was no significant increase in acquisitions after intra-state deregulation. Therefore, if the efficiency gains from a more active market for corporate control led to the reduction in failures, inter-state deregulation should have led to a greater reduction in bank failures, which is not what we observe. We therefore infer that a more active market for corporate control did not contribute to lowered bank failures following deregulation. VI.F EffectsduetoLobbyingbyLargeBanks? Could our results be a manifestation of possible lobbying by large banks for deregulation? In Table II, we find that a state was more likely to deregulate earlier if the proportion of larger banks in the state was greater, which may be an outcome of possible lobbying by the large banks. However, if large banks lobbied for deregulation since they were most likely to benefit from the same (at the expense of small banks), then the effect of deregulation should have been largely be restricted to the large banks. However, we do not find that to be the case in Table VII, which alleviate concerns relating to possible lobbying by large banks. VI.G Benefits from Reduced Economic Volatility? Reduced economic volatility due to branching deregulation may have led to lower bank exposure to macroeconomic shocks thereby reducing the likelihood of bank failure. However, the evidence seems to suggest that this channel may not have been material. First, Morgan et. al. (2004) find that inter-state deregulation led to decrease in year-to-year fluctuations in economic growth by integrating of banks across states. If the decrease in economic volatility accounted for the lower bank failures, the results should have been stronger for inter-state deregulation, which is however not the case. Furthermore, reduced economic volatility should affect small banks relatively more as small banks are less diversified and as a result more susceptible to failures following macroeconomic shocks. However, we find in Table VII that bank failures decreased uniformly for banks of all size categories, which contradicts the above prediction. We therefore conclude that the reduced economic volatility was unlikely to have accounted for the lower bank failures after deregulation. VI.H Are our results due to the Savings & Loans crisis? Several banks and financial institutions failed during the Savings and Loans (S&L) crisis in the late 80s and the early 90s. Is it the case that our results are mechanical effect of the S&L crisis? Since banking deregulation by many U.S. states occurred before the late 80s and early 90s, any mechanical effect of the S&L crisis would stack the odds against our finding the evidence that banking deregulation reduced bank failures. Nevertheless, we test 20

23 for the effect of the S&L crisis in an indirect way by splitting our sample into a period before the S&L crisis ( ) and a period after ( ) rerunning our tests for these two samples separately. We find that our results are qualitatively similar for both the samples, which suggests that our results are not an artifact of the bank failures during the S&L crisis. VI.I Are our results driven by bank failures in a few states? Finally, since figure 2 shows that bank failures were most common in TX and AK, we exclude these states from our sample and find that our results continue to hold, which reassures that our results are not driven by these outlier states. In sum, we conclude that banking deregulation lowered bank failures through three main channels: efficiency gains, benefits from geographical diversification and lower loan losses. VII Conclusion Existing theoretical and empirical studies provide conflicting arguments about the possible effects of banking deregulation on the stability of the banking sector. On the one hand, it is argued that banking deregulation leads to excessive competition and thereby reduces bank profits and threatens their stability. This line of reasoning represents a key reason behind the geographic restrictions that persisted in the U.S. banking sector for the better part of the last century. On the other hand, empirical studies find that banking deregulation leads to decreases in operating costs and lower loan losses. However, since bank capital may respond endogenously to deregulation, decreases in operating costs or loan losses may not necessarily translate into reduced instances of bank failures. In this study, we show that deregulation of bank entry indeed enhances bank stability by reducing bank failures. By ruling out possible sources of omitted variable bias and reverse causality, we identify the effect of deregulation on bank failures by exploiting the staggered manner of deregulation by the U.S. states since the 1970s. Our results question a popular narrative relating to the financial crisis that the crisis was an outcome of increased deregulation of the banking sector during the last 20 years. References [1] Aharony, Joseph and Ithzak Swary, Effects of the 1970 Bank Holding Company Act: Evidence from Capital Markets, Journal of Finance, 36 (Sept. 1981), [2] Amel, Dean, State Laws Affecting the Geographic Expansion of Commercial Banks, Manuscript, Board of Governors of the Federal Reserve System, [3] Amel, Dean, and Nellie Liang, The Relationship between Entry into Banking Markets and Changes in Legal Restrictions on Entry, The Antitrust Bulletin, 37(1992),

24 [4] Beck, Thorsten, Asli Demirgüç-Kunt, and Ross Levine, (2006) Bank Concentration, Competition, and Crises: First Results, Journal of Banking and Finance, 30, [5] Berger, Allen N., Rebecca S. Demsetz, and Philip E. Strahan (2000), The Consolidation of the Financial Services Industry: Causes, Consequences, and Implications for the Future, Journal of Banking and Finance, 23, [6] Black, Sandra E., and Philip E. Strahan, 2001, The division of spoils: Rent-sharing and discrimination in a regulated industry, American Economic Review, 91, [7] Black, Sandra E., and Philip E. Strahan, 2002, Entrepreneurship and bank credit availability, Journal of Finance 57, [8] Boyd, J., and G. De Nicolo. 2005, The Theory of Bank Risk-Taking and Competition Revisited. Journal of Finance, 60: [9] Boyd, John H., Gianni De Nicolo, and Bruce D. Smith, 2004, Crises in competitive versus monopolistic banking systems, Journal of Money, Credit and Banking, 36, [10] Boyd, John H., and David Runkle, 1993, Size and performance of banking firms, Journal of Monetary Economics 31, [11] Carlson, Mark. (2004). Are Branch Banks Better Survivors? Evidence from the Depression Era, Economic Inquiry Vol. 42(1), pp [12] Carlson, Mark, and Kris Mitchener, 2007, Branch Banking as a Device for Discipline: Competition and Bank Survivorship During the Great Depression, Journal of Political Economy, Vol. 117 (2), pp [13] Demirguç-Kunt, A. and Huizinga H., 2004, Market discipline and deposit insurance, Journal of Monetary Economics, 51, pp [14] Demirguc-Kunt, A. and Detragiache, E., 2002, Does deposit insurance increase banking system stability? An empirical investigation, Journal of Monetary Economics, 49, pp [15] Demsetz, Rebecca S. and Philip E. Strahan, 1997, Diversification, Size, and Risk at Bank Holding Companies, Journal of Money, Credit and Banking, 29(3), pp [16] De Nicolo, Gianni, 2000, Size, charter value and risk in banking: An international perspective, International Finance Discussion Paper no. 689, Board of Governors of the Federal Reserve System. [17] Demyanyk, Yuliya, Charlotte Ostergaard, and Bent E. Sørensen, 2007, U.S. banking deregulation, small businesses, and inter-state insurance of personal income, Journal of Finance 62, [18] DeYoung, Robert, Iftekhar Hasan, and Bruce Kirchhoff, 1998, The Impact of Out-of-State Entry on the Cost Efficiency of Local Commercial Banks. Journal of Economics and Business, Vol. 22(5), pp [19] Flannery, Mark, 1984, The Social Costs of Unit Banking Restrictions, Journal of Monetary Economics, 13,

25 [20] Huang, Rocco R., 2008, Evaluating the real effect of bank branching deregulation: Comparing contiguous counties across U.S. state borders, Journal of Financial Economics 87, [21] Igan, Deniz, Prachi Mishra, and Thierry Tressel, 2011, A Fistful of Dollars: Lobbying and the Financial Crisis, NBER Macroeconomics Annual. [22] Igan, Deniz and Prachi Mishra, 2011, Three s Company: Wall Street, Capitol Hill, and K Street, IMF Working Paper. [23] Jayaratne, Jith, and Philip Strahan, 1998, Entry restrictions, industry evolution, and dynamic efficiency: Evidence from commercial banking, Journal of Law and Economics XLI, [24] Kerr, William R., and Ramana Nanda, 2009, Democratizing entry: Banking deregulations, financing constraints, and entrepreneurship, Journal of Financial Economics 94, [25] Kroszner & Rajan, 1994 R. Kroszner and R. Rajan, Is the Glass Steagall Act justified? A study of the U.S. experience with universal banking before 1933, American Economic Review, 84 (1994), pp [26] Kroszner, Randall S., and Philip E. Strahan, 1999, What Drives Deregulation: Economics and Politics of the Relaxation of Bank Branching Restrictions? Quarterly Journal of Economics 114, [27] McLaughlin, Susan, 1995, The Impact of Interstate Banking and Branching Reform: Evidence from the States, Current Issues in Economics and Finance, I, Federal Reserve Bank of New York. [28] Mian, Atif, Amir Sufi, and Francesco Trebbi, 2010a, The Political Economy of the U.S. Mortgage Default Crisis, American Economic Review, vol. 100 (5), pp [29] Morgan, Donald, Bertrand Rime, and Philip Strahan, 2004, Bank integration and state business cycles, Quarterly Journal of Economics 119, [30] Keeley, Michael, 1990, Deposit insurance, risk and market power in banking, American Economic Review 80, [31] Martinez-Miera, David and Repullo, Rafael. (2010). Does Competition Reduce the Risk of Bank Failure? Review of Financial Studies, 23(10), pp [32] S. Park and S. Peristiani, Are bank shareholders enemies of regulators or a potential source of market discipline?, Journal of Banking and Finance, 31 (2007), pp [33] Pye, Gordon and Young, Ian. The Effect of Deposit Rate Ceilings on Aggregate Income, Journal of Finance, Vol. 27, No. 5 (December 1972), pp [34] Ramirez, Carlos D, 1999, Did Glass Steagall Increase the Cost of External Finance for Corporate Investment? Evidence from Bank and Insurance Company Affiliations. Journal of Economic History, [35] Stiroh, Kevin J.and Philip E. Strahan, 2003, Competitive Dynamics of Deregulation: Evidence from U.S. Banking. Journal of Money, Credit, and Banking, Vol. 35, pp [36] Wheelock, David, 1995, Regulation, Market Structure, and the Bank Failures of the Great Depression. Federal Reserve Bank of St. Louis Review, Vol. 77, pp

26 [37] Wheelock, David and Paul Wilson, 1995, Explaining Bank Failures: Deposit Insurance, Regulation, and Efficiency. The Review of Economics and Statistics, Vol.77(4), pp [38] Wheelock, David and Paul Wilson, 2000, Why do Banks Disappear? The Determinants of U.S. Bank Failures and Acquisitions. Review of Economics and Statistics, Vol. 82(1), pp [39] J. D. Wolken and F. J. Navratil, 1981, The Economic Impact of the Federal Credit Union Usury Ceiling. The Journal of Finance 36,

27 Figure 2: State-wise distribution of proportion of deposits and assets owned by failing banks This figure shows the year-wise distribution of proportion of deposits and assets owned by failing banks in USA between 1976 and The group represented by Other is the sum of the respective values across all states apart from the ones in first eight groups. 25

28 Figure 3: State-wise distribution of proportion of deposits owned by failing banks depending The figure at the top shows the year-wise distribution of proportion of deposits and assets owned by failing banks of Type II in USA between 1976 and Similarly, the figure at the bottom shows the proportion of deposits and assets owned by failing banks of Type III in USA over the same period. The group represented by Other is the sum of the respective values across all states apart from the ones in first eight groups. 26

29 Figure 4: Timing of bank deregulation and pre-existing pattern of failures Figure (A) shows a scatter plot of the average proportion of bank deposits owned by failing banks prior to bank deregulation and the year of bank deregulation. Figure (B) shows a scatter plot of the average change in the proportion of deposits owned by failing banks prior to bank deregulation and the year of bank deregulation. The t-statistics for the correlations in Figures (A) and (B) are and 0.747, respectively. 27

30 Figure 5: The dynamic impact of deregulation on proportion of deposits owned by failing banks The figure plots the impact of intrastate bank deregulation on the natural logarithm of the proportion of deposits owned by failing banks. Figure a, b, c, and d respectively show the impact of deregulation on failing banks of type 2, failing banks of type 3, failing commercial banks and all failing banks. We consider a 25-year window, spanning from 10 years before deregulation until 15 years after deregulation. The dashed lines represent 95% confidence intervals, adjusted for state-level clustering. Specifically, we report estimated coefficients from the following regression: Log ( ) = α εst The D's equal zero, except as follows: D j equals one for states in the jth year before deregulation, while D +j equals one for states in the jth year after deregulation. We exclude the year of deregulation, thus estimating the dynamic effect of deregulation on the different categories relative to the year of deregulation. A s and B t are vectors of state and year dummy variables that account for state and year fixed effects, respectively. 28

31 Figure 6: The impact of deregulation on different deciles of banks The figure plots the impact of bank deregulation on proportion of deposits owned by failing banks by different deciles. Dark bars represent estimates that are significant at 10% level. Light bars represent statistically insignificant estimates. 29

32 Table I: Summary Statistics The data on deposits and assets involved in bank and thrift failures is taken from FDIC`s Historical Statistics on Banking website and the data on total bank deposits and assets is obtained from Call Report. Growth rate of percapita GDP is calculated using data from Bureau of Economic Analysis and the data on banking regulations is taken from Jayaratne and Strahan (1996). All the dependent variables reported below are winsorized at 1% level. The number of observations equals 49 states (we exclude Delaware and South Dakota) multiplied by 19 years from 1976 to Variable Mean Standard deviation Total deposits involved in type 2 failures (X10 ) Total deposits involved in type 3 failure (X10 ) Total deposits owned by failing commercial banks (X10 ) Total deposits owned by failing banks (X10 ) Total deposits owned by failing thrifts (X10 ) Total bank deposits (X10 ) Total assets involved in type 2 failures (X10 ) Total assets involved in type 3 failure (X10 ) Total assets owned by failing commercial banks (X10 ) Total assets owned by failing banks (X10 ) Total assets owned by failing thrifts (X10 ) Total bank assets (X10 ) Proportion of deposits owned by failing banks of type 2 (X 10 ) Proportion of deposits owned by failing banks of type 3 (X10 ) Proportion of deposits owned by failing commercial banks (X 10 ) Proportion of deposits owned by failing banks (X 10 ) Proportion of deposits owned by failing thrifts (X 10 ) Proportion of assets owned by failing banks of type 2 (X 10 ) Proportion of assets owned by failing banks of type 3 (X 10 ) Proportion of assets owned by failing commercial banks (X 10 ) Proportion of assets owned by failing banks (X 10 ) Proportion of assets owned by failing thrifts (X 10 ) Intra-state banking deregulation post indicator Inter-state banking deregulation post indicator Unit banking indicator Per capita GDP growth rate at 2000 prices Number of observations

33 Table II: Did pre-existing pattern of entrepreneurship in a state affect the timing of intrastate deregulation? This table reports the results from a Weibul hazard model where the dependent variable is the log of the expected time to intrastate deregulation. The sample period is 1977 to 1994 and the sample comprises 36 states that deregulated after States drop from the sample once they deregulate. We divide U.S. into four regions and include dummies for each of the regions. Standard errors are adjusted for state-level clustering and appear in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels respectively. (1) (2) (3) Proportion of deposits with banks involved in type 2 failures (0.024) Proportion of deposits with banks involved in type 3 failures (1.012) Proportion of deposits with banks involved in failures (0.016) Percentage of all banking deposits owned by small banks in the state 8.113*** (1.922) 8.181*** (1.973) 8.058*** (1.922) Capital ratio of small banks relative to large banks in the state ** (4.972) ** (5.556) ** (4.951) Relative size of insurance in states where banks may sell insurance, 0 otherwise (2.273) (2.239) (2.292) Indicator is 1 if banks may sell insurance in the state (0.933) (0.923) (0.953) Relative size of insurance in states where banks may not sell insurance, 0 otherwise (0.867) (0.679) (0.780) Percentage of all firms in the state that are small *** (5.308) *** (4.988) *** (5.486) Share of state government controlled by Democrats (0.200) (0.208) (0.210) Indicator is 1 if state controlled by one party 0.293* (0.163) 0.303* (0.155) 0.298* (0.161) Average yield on bank loans in the state minus Fed funds rate (5.029) (5.133) (5.148) Indicator is 1 if state has unit banking law 0.496*** (0.188) 0.446*** (0.165) 0.482*** (0.185) Indicator is 1 if state changed bank insurance powers (0.259) (0.250) Growth rate of per capita Gross State Product (2000 dollars) (1.222) (1.268) Regional indicator dummies Yes Yes Yes Observations (0.257) (1.206) 31

34 Table III: Impact of Deregulation on Proportion of Deposits Owned by Failing Banks The table shows estimates of the impact of bank branch deregulation on the proportion of deposits owned by failing banks of various categories. The bank deregulation indicator equals one during all years in which a state permits in-state branching and equals zero otherwise. The number of observations in each regression equals 49 states (we exclude Delaware and South Dakota) multiplied by 19 years from 1976 to All regressions control for state and year fixed effects. Growth rate of per capita GDP at 2000 prices is also used as control in all the regressions. Panel A, B and C respectively show the effect of branch deregulation 1, 2 and 3 years after deregulation. Standard errors are clustered at the state level and appear in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels respectively. Log of proportion of deposits owned by failing banks of Type 2 Type 3 All banks (1) (2) (3) Panel A: Effect on Bank Failures three years after Deregulation Intra-state banking deregulation post indicator *** `-0.526* ** (0.612) (0.275) (0.679) Inter-state banking deregulation post indicator (0.685) (0.408) (0.683) Per capita GDP growth, one year after deregulation *** *** (6.795) (2.735) (6.251) Per capita GDP growth, two year after deregulation *** *** (4.774) (1.582) (5.195) Per capita GDP growth, three year after deregulation (5.743) (3.522) (6.020) State, Year FE Yes Yes Yes Observations R-squared Panel B: Effect on Bank Failures two years after Deregulation Intra-state banking deregulation post indicator ** * ** (0.718) (0.304) (0.730) Inter-state banking deregulation post indicator (0.576) (0.301) (0.606) Growth rate of per capita Gross State Product (2000 dollars) *** (6.143) (2.596) *** (5.486) Per capita GDP growth, one year after deregulation *** *** (4.827) (1.603) (5.185) Per capita GDP growth, two year after deregulation (5.393) (3.761) (5.700) State, Year FE Yes Yes Yes Observations R-squared Panel C: Effect on Bank Failure one year after Deregulation Intra-state banking deregulation post indicator ** ** ** (0.670) (0.281) (0.707) Inter-state banking deregulation post indicator * * (0.721) (0.289) (0.727) Growth rate of per capita Gross State Product (2000 dollars) *** (5.966) (1.648) *** (6.095) Per capita GDP growth, one year after deregulation * * (4.726) (3.587) (4.905) State, Year FE Yes Yes Yes Observations R-squared

35 Table IV: Tests Analysing Possible Placebo Effects - Impact of Deregulation on Thrift Failures The table shows estimates of the impact of bank branch deregulation on the proportion of deposits owned by failing thrifts as a proportion of total banking deposits. The bank deregulation indicator equals one during all years in which a state permits in-state branching and equals zero otherwise. Since the Garn-St. Germain Act was passed in 1982, we start our sample from The number of observations in each regression equals 49 states (we exclude Delaware and South Dakota) multiplied by 12 years from 1983 to All regressions control for state and year fixed effects. Growth rate of per capita GDP at 2000 prices is also used as control in all the regressions. Standard errors are clustered at the state level and appear in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels respectively. Log of proportion of deposits owned by failing thrifts (1) (2) (3) 3 years after 2 years after 1 year after Intra-state banking deregulation post indicator (0.963) (1.051) (0.783) Inter-state banking deregulation post indicator (0.833) (0.832) (0.950) Per capita GDP growth, one year after deregulation * (9.187) *** (9.163) * (9.131) Per capita GDP growth, two year after deregulation *** (9.653) ** (7.305) Per capita GDP growth, three year after deregulation ** (6.535) Observations R-squared Table V: Relative Effect of Bank Deregulation on Bank Failures in Unit Banking States versus States that Allowed Limited Branching The table presents estimates of the impact of bank deregulation on the measures of bank failures as a function of initial branching laws. Bank deregulation equals one during all years in which a state permits in-state branching and equals zero otherwise. All models control for state and year fixed effects. Since we control for state fixed effects the initial state characteristics are dropped from the regressions. Unit banking states include: CO, AR, FL, IL, IA, KS, MN, MO, MT, NE, ND, OK, TX, WI, WV, and WY. Standard errors are adjusted for state level clustering and appear in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels respectively. (1) (2) (3) Log of proportion of three year lead of deposits owned by Type 2 Type 3 All banks failing banks of (Intra-state banking deregulation post indicator)*(unit banking) ** (0.876) *** (0.443) ** (0.891) Intra-state banking deregulation post indicator (0.766) (0.224) (0.820) Per capita GDP growth, one year after deregulation *** *** (6.694) (2.722) (6.124) Per capita GDP growth, two year after deregulation *** *** (4.700) (1.549) (5.094) Per capita GDP growth, three year after deregulation (5.508) (3.508) (5.779) State, Year FE Yes Yes Yes Observations R-squared

36 Table VI: Relative Impact of Deregulation on Failures of Banks of Different Sizes The table shows estimates of the impact of bank branch deregulation on the proportion of deposits owned by all failing banks by various size categories. The bank deregulation indicator equals one during all years in which a state permits in-state branching and equals zero otherwise. The number of observations in each regression equals 49 states (we exclude Delaware and South Dakota) multiplied by 19 years from 1976 to All regressions control for state and year fixed effects. Up to three leads of growth rate of per capita GDP at 2000 prices is also used as control in all the regressions. Standard errors are clustered at the state level and appear in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels respectively. (1) (2) (3) (4) First quartile (largest) Second quartile Third quartile Fourth quartile (smallest) Intra-state banking deregulation post indicator ** * ** (0.459) (0.506) (0.436) (0.667) Inter-state banking deregulation post indicator (0.593) (0.633) (0.404) (0.522) Per capita GDP, one year after deregulation *** *** *** (3.821) (6.198) (4.626) (3.605) Per capita GDP, two year after deregulation ** ** * (5.819) (3.814) (4.481) (4.550) Per capita GDP, three year after deregulation *** (3.344) (3.692) (4.108) (4.430) State, Year FE Yes Yes Yes Yes Observations R-squared Table VII: Relative Effect of Deregulation on Bank Failures in Unit Banking States versus States that Allowed Limited branching on Banks of Different Sizes The table presents estimates of the impact of bank deregulation on the proportion of deposits owned by failing banks of various size categories as a function of the initial branching laws. We look at the impact of deregulation three years after the deregulation. Bank deregulation dummy equals one during all years in which a state permits in-state branching and equals zero otherwise. All models control for state and year fixed effects. Since we control for state fixed effects the initial state characteristics are dropped from the regressions. Unit banking states include CO, AR, FL, IL, IA, KS, MN, MO, MT, NE, ND, OK, TX, WI, WV, and WY. The data excludes 12 states that deregulated before Standard errors are adjusted for state level clustering and appear in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels respectively. (1) (2) (3) (4) First quartile (largest) Second quartile Third quartile Fourth quartile (smallest) (Intra-state banking deregulation post indicator)*(unit banking) ** (0.679) (0.800) ** (0.629) * (0.873) Intra-state banking deregulation post indicator (0.531) (0.577) (0.522) (0.758) Per capita GDP growth, one year after deregulation *** (3.775) (6.210) *** (4.507) *** (3.530) Per capita GDP growth, two year after deregulation (5.847) ** (3.767) *** (4.323) * (4.451) Per capita GDP growth, three year after deregulation ** (3.407) (3.671) (4.010) (4.263) State, Year FE Yes Yes Yes Yes Observations R-squared

37 Internet Appendix for Deregulation of Bank Entry and Bank Failures Krishnamurthy Subramanian Ajay Yadav 35

38 Figure A-2: The dynamic impact of deregulation on proportion of assets owned by failing banks The figure plots the impact of intrastate bank deregulation on the natural logarithm of the proportion of assets owned by failing banks. Figure a, b, c, and d respectively show the impact of deregulation on failing banks of type 2, failing banks of type 3, failing commercial banks and all failing banks. We consider a 25-year window, spanning from 10 years before deregulation until 15 years after deregulation. The dashed lines represent 95% confidence intervals, adjusted for state-level clustering. Specifically, we report estimated coefficients from the following regression: Log ( ) = α εst The D's equal zero, except as follows: D j equals one for states in the jth year before deregulation, while D +j equals one for states in the jth year after deregulation. We exclude the year of deregulation, thus estimating the dynamic effect of deregulation on the different categories relative to the year of deregulation. A s and B t are vectors of state and year dummy variables that account for state and year fixed effects, respectively. 36

39 Figure A-1: The impact of deregulation on different deciles of banks The figure plots the impact of bank deregulation on proportion of assets owned by failing banks by different deciles. Dark bars represent estimates that are significant at 10% level. Light bars represent statistically insignificant estimates. 37