I. INTRODUCTION

Did the financial support of distressed U.S. banks by the Capital Purchase Program (CPP) affect loan and funding rates, as measures of price competition? The CPP, the largest single element of the Troubled Asset Relief Program (TARP), dispersed around $204.9 billion to 707 U.S. banks between q4/08 and q4/09. As of July 31, 2014, the Treasury recovered $225.9 billion of this CPP support in the form of repayments, dividends, and interest, turning the program into a positive return for taxpayers. Timothy Masad, deputy Secretary of the Treasury in charge, accordingly called TARP a success in the final hearing of the Congressional Oversight Panel (COP) on March 4, 2011 (see http://cop.senate.gov), and Liu et al. (2013) agreed in their analysis of the substantial financial and return recovery of banks that received CPP funds. However, on an economic cost-benefit basis, it is not clear whether taxpayers had a net positive return (Calomiris and Khan 2015). Yet in its final assessment, the COP (2011) paints a more nuanced picture: Although the cost of TARP was much lower than anticipated, it might also have induced distortions of market mechanisms, in the form of increased risk taking and reduced competition. The former issue has received considerable attention in recent studies (Black and Hazelwood 2013; Dam and Koetter 2012; Duchin and Sosyura 2014; Gropp et al. 2011), whereas evidence about competitive distortions due to TARP is rare.

Bailout schemes can distort competition in two ways: directly, by subsidizing rescued banks, and indirectly, by inducing undesirable market conduct by unsupported banks. Specifically, government bailouts directly distort banking competition because insurance schemes treat banks differently depending on the size of the subsidy (Beck et al. 2010), which upsets any existing level playing field. Empirical evidence about the direct effect of bailouts on competition is mixed. Calderon and Schaeck (2015) show, with a sample of 46 banking crises in 138 countries, that government support of troubled banks led to more banking competition and lower interest margins after a crisis. The main benefits accrue to borrowers in already financially well-provided segments. In contrast, Berger and Roman (2015b) show that TARP-supported U.S. banks exhibited higher Lerner margins and market shares compared with unsupported banks in the period after q4/09, driven by banks that repaid early. While suppliers of funds require lower risk premiums, TARP capital infusions required a dividend yield of 5% in the first 5 years of support, increasing to 9% thereafter. In addition, TARP infusions were tied to executive compensation caps (Bayazitova and Shivdasani 2012). Berger and Roman (2015b) conclude that the safety net benefits of TARP outweighed the cost disadvantages. Even if bailouts are allocated on perfectly equal terms to all banks, the provided insurance creates socially undesirable, additional risk taking (Keeley 1990). Consistent with this view, the Congressional Oversight Panel (2011) voiced concerns that TARP equity provisions provided supported banks with a competitive advantage that could lead to consolidation and further concentration, to the detriment of small or local community banks in particular. In turn, these subsidized survivors, with their increased market power, could invoke additional welfare losses by charging higher interest rates to borrowers that represent poor credit risks.

Theoretically, Hakenes and Schnabel (2010) emphasize the importance of indirect effects of government bailouts on unsupported peers too. The increased protection of banks that anticipate bailouts reduces the margins and charter values of competing, unsupported banks. Prospective bailouts also induce depositors to require lower default premiums, such that the reduced funding costs imply more lending by protected banks, which translates into increased competitive pressure on unsupported incumbents. Depositors instead require higher risk premiums from unprotected banks, which reduces margins at given loan rates or could encourage higher risk taking by the banks in an attempt to increase expected returns and thus margins.

We focus on the latter effect and use political indicators in the banks' home markets and Congressional voting behavior on TARP to identify bailout expectations. In turn, we assess how unsupported banks responded, in terms of pricing power, to the bailout scheme. To our knowledge, this study is the first to analyze these indirect effects of prospective bailouts, or bailout expectations, among unsupported banks. They accounted, on average, for 40% of cumulative banking assets during the TARP disbursement period (q4/08-q4/09). We test whether the expectation of capital support affects unsupported banks' interest margins and loan and deposit growth. This approach complements the focus by Berger and Roman (2015b) and Li (2013) on differences between TARP and non-TARP banks in terms of markups and loan supply, respectively, such that we identify within the group of unsupported banks the presence and magnitude of competitive distortions.

The empirical challenge is that bailout expectations usually are not observable. The joint occurrence of bank support during the TARP disbursement period and bank failures is an important exception that enables us to estimate the likelihood that a distressed bank will be rescued, relative to the probability it will fail, according to banks' risk and size traits. In the spirit of Dam and Koetter (2012), we extrapolate bailout expectations for sound banks based on parameter estimates that can separate banks that received TARP support from those that exited the market due to failures with high accuracy. Proper identification of competition effects due to bailout expectations rather than other determinants requires factors that can discern between failing and supported banks but that are uncorrelated with the interest margins of unsupported banks. Similar to Duchin and Sosyura (2012, 2014) and Li (2013), we consider information of whether Congressional representatives of the banks' counties were on the subcommittee of financial services, their voting behavior in Congress about TARP, and their party membership. On the basis of these parameter estimates, we extrapolate the bailout expectations for sound banks. Controlling for risk taking, we regress the generated bailout expectations revealed during the TARP disbursement period on the loan rates charged, deposit rates incurred, and corresponding volume changes after the end of the subsidy program (q1/10-q4/13). These measures match the main channels Hakenes and Schnabel (2010) cite to describe how bank bailouts distort competition among unsupported peers.

Our results show that higher bailout expectations increase loan rates and reduce deposit rates in the post-TARP period q1/10-q4/13. This increase of interest rate margins is consistent with theory and robust to matched sampling tests that seek to ensure comparability across the TARP recipients we used to generate bailout expectations. These price effects are most pronounced in the immediate aftermath of the TARP disbursement, then turn insignificant after 2010. Any price distortions due to changed bailout expectations among unsupported banks thus appear to have been short lived. We find no evidence that banks that are perceived as particularly likely to receive a bailout exhibit significantly larger loan or deposit growth. This result mitigates concerns by the COP that small, unsupported banks were particularly at risk to lose market share. Overall, the increasing (decreasing) effect on loan (deposit) rates is amplified in states where competitive restrictions were more pronounced.

The remainder of this article is organized as follows: Section II outlines the empirical strategy, presents the data, and explains the identification methods we used to estimate bailout expectations due to government intervention via TARP. In Section III, we present the estimation results for the bailout expectation effects after 2009 before we conclude in Section IV.

II. EMPIRICAL STRATEGY AND DATA

A. Sampling

Following Hakenes and Schnabel (2010), we test the hypothesis that higher bailout expectations increase interest margins and possibly loan and deposit growth. But the likelihood of receiving a bailout, that is, bailout expectations are usually not observable. The simultaneous occurrence of both TARP support and bank failures between q4/08 and q4/09 is exceptional, because regulators revealed which banks they considered important enough to rescue. Selected banks received equity support, while many banks that did not receive TARP support failed. To test the indirect channel of competitive distortions due to bailout expectations, we use observed failures and TARP bailouts during q4/08 and q4/09 (t = 1) to generate bailout expectations for sound banks during q1/10 and q4/13 (t = 2). It is important to note that this approach does not assume implicitly that the TARP program as such would be extended, neither in terms of timing nor scope and volume. Instead, we assume that the regulator is not equally likely to rescue any given bank, but that there exists a latent implicit bailout propensity that varies in the cross section of banks and is inherently unobservable during normal times. All we exploit here is that the regulator had to (unexpectedly) reveal this latent variable in response to the threat of a system meltdown by admitting some banks to the TARP program while letting others fail. Figure 1 illustrates the empirical strategy and sampling.

In the upper part of Figure 1, we find that at the end of q3/08, banks were either distressed and in need of support or sound. The latter, sound banks should have no incentives to apply for TARP funds, for three reasons: the funds were expensive, receiving support meant limiting the compensation of managers, and TARP carried a potential stigma cost (Bayazitova and Shivdasani 2012; DeYoung et al. 2013; Wall Street Journal 2009). The regulator decides in period t = 1 which distressed banks to rescue. Sound banks are sampled as all other commercial banks that survived at least until q4/09, the end of the TARP disbursement period. Table 1 shows the frequency distribution of supported, failed, and sound banks per quarter during the crisis period q4/08-q4/09 and for the period q1/10-q4/12.

Corresponding with the columns in Table 1, we sampled 548 of the 707 banks that received TARP and observed 136 failures as reported by the Federal Deposit Insurance Corporation (FDIC). In nine cases, banks failed even after the holding company received TARP funds. We excluded these cases from our analysis, leaving 127 failures and a failure rate conditional on distress of around 22% during t = 1. Conditional on distress, as revealed by the observable outcomes of bailout versus failure, banks had to apply for TARP funds, though with only light formal requirements.

The indirect competitive distortions of Hakenes and Schnabel (2010) hinge on depositors' expectations that an unsupported bank they supply with funds will be protected by a prospective bailout. (1) We assume that agents form expectations about the likelihood of a bailout relative to failure during t = 1 and extrapolate expectations to nontreated banks after the TARP disbursement period ended, that is, to t = 2.

Table 2 shows that the relatively small number of rescued banks accounted for an average of 60% of aggregate (commercial) banking assets in the United States, relative to the approximately 5,800 sound banks in t = 1. More than half of the aggregate assets among TARP recipients accrued to what Li (2013) calls the eight mega banks (Citigroup, JP Morgan, Bank of America [including Merrill Lynch], Goldman Sachs, Morgan Stanley, State Street, Bank of New York Mellon, and Wells Fargo [including Wachovia]), which neither the government nor the Fed would let fail, such that they were forced to take TARP funds. The columns labeled "Forced" in Table 2 show that the mean size difference between supported and unsupported sound banks was driven by this group, such that the mean bank size of supported sound banks was $11 billion, whereas that for the unsupported sound banks was $0.5 billion. The COP's concern that smaller, unsupported banks would suffer from distortions thus seemed justified. Furthermore, the 40% share of total assets managed by sound banks warrants an analysis of potential competitive distortions within this group.

The lower panel in Figure 1 shows the four possible scenarios that banks faced in t = 2 (q1/10-q4/13). First, TARP recipients could fail or survive in t = 2. Only one TARP recipient failed. The remaining 547 TARP banks survived until q4/13, representing the distressed sample, as depicted by the branches inside the dashed box in Figure 1. Second, sound banks from t = 1 either failed or survived in t = 2, as noted in the solid box in Figure 1. Of the 5,900 sound banks in q4/09, 275 failed during t = 2, and 5,177 non-TARP recipients survived through q4/13.

A test of direct distortion effects (Berger and Roman, 2015b; Calderon and Schaeck 2015; Li 2013) would seek to identify the differential effect of bailout support in the full sample, as indicated in Figure 1 by the dotted box between TARP and non-TARP banks (dashed vs. solid boxes). We test the effect that heterogeneous bailout expectations have on unsupported banks only, sampled in the solid box in Figure 1. With this setup, we can determine whether government rescue schemes exert obvious effects on rescued banks relative to nonrescued ones but also affect the group of supposedly sound banks.

B. Specification

In the first stage, we approximate bailout expectations at t = 1 by using a probit model to estimate the probability of receiving TARP relative to failing, while controlling for bank traits X that gauge risk and importance, as well as regional economic conditions (Dam and Koetter 2012). The dependent variable TARP is an indicator variable equal to 1 if a bank i received equity in a quarter t between q4/08 and q4/09 or 0 if the bank failed:

(1) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

Control variables capturing the bank characteristics and regional control variables X are lagged by one-quarter and explained in detail in Section II.C and Table 3. However, the decision to bailout a bank is unlikely to be independent of the bank's market power, as reflected by its ability to set prices. Therefore, we need to deal with a potentially endogenous relationship between generated bailout expectations, loan rates [R.sup.L], and deposit rates [R.sup.D]. To identify the effect of bailout expectations on interest rates, we specify exclusion restrictions P that are uncorrelated with rates but that effectively distinguish between failing and rescued banks.

We follow Duchin and Sosyura (2012, 2014) and Li (2013) when specifying P and use four political variables, reflecting our allocation of each bank i to Congresspeople representing the region d where the bank resides. First, we define two dummy variables (SC0709 and SC0911) if a Congressperson was on the subcommittee of financial services during 2007-2009 and 2009-2011. As Li (2013) argues, members of this subcommittee should possess expert knowledge and qualifications that enhance their ability to judge the rescue program and the decision to provide funds to certain banks. Second, we use a dummy variable (2nd Vote) that shows whether a Congressperson voted yes (1) or no (0) for the second vote on TARP. The idea behind this variable is that representatives' opinions might have changed between the two votes on TARP if his or her region had been granted specific concessions, such as financial support in the form of government projects. We therefore regarded the second vote on TARP as more important for our analysis but find that our results do not change if we specify the first vote on TARP instead or use both simultaneously. Third, we identify the party of each Congressperson (Party0709 and Party0911) for the respective session, equal to 1 if the representative was a Democrat and 0 otherwise. This variable acknowledges that ideology differs systematically, such that conservative Republicans tend to oppose government interventions more categorically than Democrats Li (2013). We provide the descriptive statistics in the top panel of Table 4.

In the second stage, we assess the effect of bailout expectations [E[TARP = 1].sub.it] on price competition, as reflected by the interest rates that banks received on loans [R.sup.L] and paid for (deposit) funding [R.sup.D]. Note that we estimate parameters to predict bailouts in Equation (1) only for those banks that are distressed and applied (successfully) for TARP funding or failed during t = 1. To predict bailout expectations for sound banks in each quarter of t = 2, we use the estimated parameters of Equation (1), [??] and [??]. That is, we extrapolate bailout expectations to sound banks (Dam and Koetter 2012). These descriptive statistics appear in the third panel of Table 4.

We estimate a fixed effects regression for t = 2, q1/10 to q4/13. With our interest in the indirect effects of government bailouts, we estimate the relationship for sound banks only, that is, the sample indicated by the solid box in Figure 1. Formally,

(2) [R.sub.it] = [b.sub.0] + [b.sub.1] E [[TARP = 1].sub.it] + [R.summation over (r=2)] [b.sub.r] [X.sub.rit-1] + [[tau].sub.t] + [[mu].sub.i] + [[gamma].sub.year] x [[theta].sub.state] + [e.sub.it].

With this approach, we derive results for five dependent variables: interest rates on total loans ([R.sup.TL]), real estate loans ([R.sup.RE]), commercial and industrial lending ([R.sup.CI]), deposit rates ([R.sup.DI]), and total funding ([R.sup.TF]). All interest rates reflect the annualized quarterly yield a banks receives on loans or pays on funding (for the descriptive statistics, see the second panel of Table 4). Asset yields reflect interest and fee income for the respective asset class divided by the average of this asset class. The funding yields reflect all costs associated with the funding item divided by the average of the funding volume. (2) In addition to the identical vector of control variables X in Equation (1), we specify quarterly dummies [[tau].sup.t], bank-fixed effects [[mu].sub.i], and cluster standard errors at the bank level. The term [[gamma].sub.year] x [[theta].sub.state] reflects interacted year and state effects that capture additional time-varying differences on the state level.

The variable Bailout expectation in Table 4 describes E[[TARP = 1].sub.it] during and after TARP. During the TARP disbursement period, bailout expectations are significantly higher for banks that received TARP compared with the (extrapolated) bailout expectations of sound banks. This difference is statistically insignificant for the postdisbursement period. However, especially in the post-TARP period, the dispersion of bailout expectations is highest within the group of sound banks. This heterogeneity in agents' expectations about prospective bailouts should affect required risk premiums, and thus prices (Hakenes and Schnabel 2010), for both supported and especially unsupported banks.

C. Data Sources

We obtained data from five different sources. First, we used financial accounts and failed bank data from the FDIC. Second, we obtained TARP recipient identities from the Department of the U.S. Treasury. Third, we gathered data to measure the voting behavior of Congressional representatives and their party affiliations from the website of the U.S. House of Representatives. Fourth, county-level unemployment rates came from the Bureau of Labor Statistics, and the state-level Case-Shiller indices came from the FRED database provided by the Federal Reserve Bank of St. Louis. Fifth, we used data on loan and funding interest rates for U.S. banks obtained from the Uniform Bank Performance Reports of the Federal Financial Institution Examination office.

We started with 8,231 banks for the period q4/08-q4/13 but cleaned these data. First, we restricted the sample to commercial banks, leaving 7,191 banks. (3) Second, we dropped all banks with headquarters outside the U.S. mainland and the District of Columbia, resulting in a sample of 7,177 banks. Third, by requiring complete observations for all variables used in the analysis, we reduced the sample to 7,165 banks. (4) Fourth, to exclude mergers and voluntary exits, we followed Kashyap and Stein (2000) and required that all banks not recorded as failures by the FDIC survived until q4/13. This culling left 6,172 banks. Fifth, we required that the remaining banks have consecutive years, so the final sample included 6,135 banks.

We followed Wheelock and Wilson (2000) and Cole and White (2012) in our choice of control variables; the descriptive statistics for TARP, failed, and sound banks during and after the crisis period appear in Table 4.

To control for risk buffers, we used the equity-to-asset ratio (EQ). The variable Loans reflected the ratio of total loans to total assets, so as to control for the relative importance of credit business to the bank. The Cash variable indicated banks' cash, standardized by total assets. We control for profitability using the pre-tax return on assets, RoA. The share of nonperforming assets over assets (NPA) also controlled for asset risk. To address the differences between small and large banks, we used Size, the natural logarithm of total assets. In addition, to capture differences in funding structure, we specified Deposits as the share of total deposits to total assets. For the local economic conditions, we specified the county-level rate of unemployment UR. For each bank and quarter, this variable equaled the mean of county unemployment rates from the bank's business regions, as indicated by the summary of deposits weighted by the bank's deposits in each county. The variable CS index was the state-level Case-Shiller index. The bottom panel of Table 4 provides descriptive statistics.

III. RESULTS

A. Identification of Bailout Expectations

To assess the effect of prospective bank bailouts on price competition, we must identify bailout expectations accurately. Valid exclusion restrictions must explain bailout expectations as well as be weakly correlated with the endogenous variables, asset, and liability interest rates. Table 5 shows the estimated marginal effects of Equation (1) for each instrument, specified both individually and jointly.

The joint specification in column (1) shows that the instruments correlate significantly with bailout expectations and thereby confirms the relevance of political factors as means to discern between rescued and failed banks: Banks in districts with a Congressperson who voted "yes" on the second TARP vote were more likely to receive bailout funds. Banks in districts with a representative who also sat on the subcommittee of financial services were more likely to be bailed out during the 2007-2009 period. Banks in districts whose representatives were members of the subcommittee during 2009-2011 were less likely to receive TARP support. Party membership in both periods significantly predicted whether a bank would be bailed out or fail too. For example, banks operating in a region represented by a Democrat in the first session (2007-2009) were more likely to receive TARP funding, whereas this relation flips in the second session (2009-2011), indicating a shift in the assessment of Democratic Congresspersons about which banks were eligible for TARP. All instruments in column (1) differed individually significantly from 0. An E-test statistic larger than 15 corroborated the joint significance of all five variables, in support of the instruments' validity. The specifications in columns (2) through (6) show that most instruments also correlate significantly with bailout expectations on an individual basis. The coefficients for the subcommittee dummy of 2007-2009 and party membership of 2009-2011 change signs in columns (3) and (6) compared with column (1) but are insignificant in the individual specifications.

The average marginal effects of bank characteristics and the regional unemployment rate in Table 5 show that banks with larger equity buffers, banks acting in states with a higher Case-Shiller index, and more profitable banks were more likely to be bailed out. This result is broadly in line with the intention of the U.S. Treasury to rescue only those banks that had the potential to repay their TARP support (Duchin and Sosyura 2014; Li 2013). In contrast, banks with high loan ratios, lots of troubled assets, and with high cash ratios were less likely to receive TARP funds.

Regarding the orthogonality requirement between instruments and interest rates, we cannot use conventional tests, because we use extrapolated bailout expectations from t = 1 to explain the price-setting behavior of sound banks during t = 2. This instrumental variable setting is nonstandard, in that the first and second stages pertain to different samples at different time horizons. To test whether political indicators P are only weakly correlated with the interest rate outcome variables, we instead regressed the full set of controls X together with the instruments P on our main dependent variables, after the disbursement of TARP funds, for the sample of sound banks only (Table 6). Thus we can test if bailout expectations revealed during q4/08 and q4/09 affected asset and liability interest rates after the TARP disbursement period for the group of sound banks. Valid instruments should exhibit very weak correlations with the dependent variables during the disbursement period, but also after 2009.

Each column in Table 6 confirms that for each dependent variable, most of the instruments exhibited no correlation after the TARP disbursement period. Only party membership and sitting on the financial services subcommittee were significant and only in some cases. The F-tests in the bottom panel also indicate the joint insignificance for all interest rates except commercial and industrial lending after TARP stopped. In summary, these results supported the validity of the exclusion restrictions to identify bailout expectations.

B. Bailout Expectation Effects on Interest Rates

Table 7 reports the estimation results for the baseline specification of Equation (2), designed to explain the impact of bailout expectations on interest rates as a measure of banking market competition. The variable of interest is the contemporaneous bailout expectation calculated from the estimates of Equation (1). We also specify the same control variables as in Equation (1) to account for bank characteristics, and risk in particular, and regional economic factors. These variables are lagged by one-quarter, as indicated by the prefix L. Finally, we control for bank, quarter, and interacted year-state fixed effects.

The five columns in Table 7 reflect each of the five asset and liability interest rates. For column (1), which shows the results for interest rates on total loans, recall that the sample comprises banks that never received TARP support after the program stopped in q4/09, or the solid box in Figure 1. Sound banks that were considered more likely to receive a bailout, should they face distress, realized significantly higher yields. An increase of bailout expectations by 1 percentage point increased yields by 0.0014 basis points. This is minuscule and shows that despite being significant, the economic effect of higher bailout expectations absent for the sample of sound banks. Put into perspective, an increase of bailout expectations by one standard deviation (0.1899, see Table 3) would increase loan interest rates by 2.65 basis points. In light of an average loan yield of about 6% between q1/10 and q4/13, this reflects an increase of loan rates of about 4%.

Columns (2) and (3) confirm both the direction and the significance of these results for real estate and commercial and industrial lending, respectively. Banks with higher bailout expectations generated higher yields for real estate loans and commercial and industrial loans. In terms of economic magnitudes, the effects were comparable for real estate loans and total lending. The increase in commercial and industrial loan rates in response to an increase in bailout expectations was approximately around twice as large as the total loan rates.

These positive effects on loan rates, and thus markups as a measure of market power, are in line with the findings by Berger and Roman (2015b) and might indicate that loan customers consider a stable credit relationship important. While the failure of a bank for previously conducted credit disbursements to a company may not be disruptive, most companies rely on irrevocable credit commitments and credit lines from these banks as well. Therefore, they may be willing to incur somewhat higher loan interest rates with banks they consider more likely to be rescued in case of distress. However, our setting also differs in important ways from Berger and Roman (2015b), who study the contemporaneous effects of TARP support between recipients and nonrecipients on (generated) measures of market power. Because we consider solely the reactions of banks that were not directly rescued, we explain the within-group variation of interest rates among sound banks. If bailout policies do not alter competitive conditions, as reflected by loan prices, differences in bailout expectations should be uncorrelated. The reported positive significant effect therefore offers important evidence that the dominant safety net effect reported by Berger and Roman (2015b), that is, rescued banks are considered safer, also extends to banks for which suppliers of funds anticipate bailouts to be more likely.

Positive bank asset interest rates are not the primary channel by which prospective bailouts reduce markups, according to the theoretical model of Hakenes and Schnabel (2010) though. Instead, they propose bank market power, manifested as interest margins, increases because depositors are willing to accept a lower risk premium, which implies lower funding costs for implicitly protected banks. Columns (4) and (5) in Table 7 show that we cannot reject the null hypothesis of no relationship among bailout expectations, deposits, and total funding interest rates for the whole period, q1/10-q4/13.

In unreported results, we find very similar results when we use banks' Lerner indices as in Berger and Roman (2015b) and Li (2013) to approximate markups. Here, an increase of bailout expectations leads to significantly higher Lerner indices which are driven by a reduction of marginal cost. Average revenues seem to be unaffected by bailout expectations in our analysis. For four other loan categories (customer, credit card, agricultural, and foreign), we cannot estimate these effects with sufficient precision to obtain statistical significance, mostly because of much smaller sample sizes.

C. Extrapolation of Bailout Expectations Revisited

Both the absence of reduced required interest rates on the funding side and the positive correlation between lending rates and bailout expectations may be spurious results, due to the extrapolation of bailout expectations after q4/09 from observed bailout behavior between q4/08 and q4/09. We address these concerns with a series of robustness tests and report the results in Table 8. Out of space considerations, we only provide coefficients for the variable of interest, bailout expectations: we suppress the estimates for the other controls and fixed effects in Equation (2). To begin, in the first panel of Table 8, we replicate the baseline results from Table 7 for comparison. Next, in the second panel, we provide the results for a sample that excludes banks that were sound in t = 1 but failed in t = 2 (275, see Figure 1). According to Hakenes and Schnabel (2010), unsupported banks respond to competition from subsidized peers by taking riskier lending activities to increase their expected returns. We explicitly control for risk taking, but such formerly sound banks may be exactly those in the bold-outlined sample in Figure 1 that did not experience reduced funding costs and (over)compensated for the competitive pressure from their rescued peers by seeking high yield, high risk projects that eventually led to failure during t = 2. Because excluding these failing banks did not reduce the marginal effect significantly though, this test confirmed that our baseline results were not driven by (excessively) risky, unsupported banks.

The third panel of Table 8 features on the sample indicated by the dotted line in Figure 1, namely, both TARP and non-TARP banks considered jointly (Berger and Roman, 2015b; Li 2013). As Figure 1 shows, only one TARP bank failed after q4/09. With this robustness test, we still found a positive, significant effect for the interest rates of asset-side yields but no effect on the funding side. It remains unclear whether the lower risk premiums required by banks' financiers reflect a differential effect of TARP or variation in the within-sound bank group of formerly sound, unsupported banks' bailout expectations in the post-TARP period. This ambiguity motivated us to consider asset and liability interest rates in t = 2 among only those banks that were sound in t = 1. (5)

The specification of Equation (2) for the TARP-only sample in the fourth panel of Table 8, equivalent to the dashed box in Figure 1, illustrates that variation in the within-sound bank group drove the positive effect of bailout expectations on bank yields. During q1/10-q4/13, the 548 banks that received TARP funds and operated during period t = 2 did not exhibit any significant correlation with yields. The absence of this result affirms the theoretical prediction of Hakenes and Schnabel (2010) that the direct effect of support should be ambiguous and even potentially negative in terms of risk taking. In our study setting, we controlled for the level of risk taking using bank-specific covariates, which exhibited similar magnitudes, significance, and directions across the four samples. The variation of risk-controlled interest rates between TARP and non-TARP banks thus appeared to hinge on the relationship between bailout expectations and loan and funding rates.

Our approach also allows for the extrapolation of bailout expectations from the TARP disbursement period to sound banks to the subsequent period, with the crucial assumption that distressed banks during q4/08-q4/09 are comparable to sound banks as of q1/10. We challenge this assumption though by presenting, in the bottom panel of Table 8, results based on matched samples between bailed out and sound banks. To ensure that we calculated bailout expectations for sound banks that shared similar characteristics with banks that received TARP funds, we ran propensity score matching. The matching process relied on the vector of bank characteristics and regional control variables X from Equation (1). We specified a 1:1 matching, such that for each distressed bank, we linked one sound bank with the highest propensity score between q4/08 and q4/09. Formally, our propensity score matching method used a logit regression, E[[DIS = 1].sub.i] = [[lambda].sub.0] + [[summation].sup.C.sub.c-1] [[lambda].sub.c] [X.sub.ci] + [[phi].sub.it], to differentiate between TARP recipients from failing banks, whether distressed banks (DIS = 1) or sound ones (DIS = 0), during the crisis period of q4/08-q4/09. Using a nearest neighbor matching without replacement, we required that each pair was not different at a 1% level, according to the matrix of bank and regional variables X. We present the effect of the matching process and the resulting size of the treatment and control groups in Table 9, revealing both bias before and a significant reduction in bias after matching.

A comparison of distressed and sound banks that were matched (M) and unmatched (U) revealed the importance of an appropriate counterfactual sample when extrapolating bailout expectations. Specifically, unmatched sound banks were significantly better capitalized, more profitable, riskier, larger, more liquid, more retail funding oriented, and more loan-based in their asset composition. They tended to operate in regional markets with less unemployment and higher real estate prices. Thus, extrapolation of bailout expectations to any sound banks would appear overly optimistic.

Using only the sample of matched banks to assess the effect of bailout expectations on interest rates in the bottom panel of Table 8, we confirmed our baseline results for two of the three loan rates we considered. Specifically, the rows indicated by "M" in Table 9 show the comparability of these institutions with distressed banks. Higher bailout expectations generated higher yields on total loans and commercial and industrial loans, whereas the effect on real estate loan rates was insignificant for the matched sample. The magnitude of positive interest rate effects due to higher rescue probabilities reached twice as high for total and commercial and industrial loans. These results emphasized the importance of extrapolating bailout prospects only to sufficiently similar, sound banks.

Perhaps more important is the result showing that funding costs and deposit interest rates fell by approximately 1 basis point in response to a one standard deviation increase in bailout expectations. Thus, the reduction in required risk premiums predicted by Hakenes and Schnabel (2010) was statistically significant for this matched sample. Although lower than the effect on loan rates, the effect on deposit rates was economically more pronounced, given the average funding cost of 2% instead of 6% for the average loan rates (see Table 3).

The effects on interest margin components thus appear driven by within-sound bank differences in bailout prospects, rather than differences between TARP and non-TARP recipients. Generating bailout expectations also requires the careful construction of an appropriate counterfactual sample of sound banks that are sufficiently comparable to distressed banks during the TARP disbursement period. With this sample, we found statistically and economically significant effects of increased bailout expectations, in line with theory, including larger loan interest rates and reduced funding rates for banks.

D. Timing Differences

Most of the concerns about the potentially distortionary effects of bailouts on banking market competition were voiced shortly after TARP was terminated in q4/09 (Beck et al. 2010; Congressional Oversight Panel 2011). Beyond this focus, another critical question is whether emergency rescues affected interest rates only in the short run or if any potential distortions exhibited a longer duration.

In Table 10, we present the results of an interaction of generated bailout expectations with year dummies for the years 2010, 2011, 2012, and 2013 when estimating Equation (2). Given our preceding results in Tables 8 and 9, we consider only the matched sample. The baseline results did differ significantly across the 3 years after the TARP period, q4/08-q4/09. Regarding loan rates, we found significantly positive effects on total loan rates for the first 2 years after TARP stopped. Magnitudes declined from the 6 basis point hike in response to the one standard deviation increase in bailout expectations to 3.6 basis points in 2011. Thereafter, the estimated coefficients remained positive but no longer statistically significant. Contrary to the results across all post-TARP years in Table 8, both real estate and commercial and industrial loans exhibited significantly larger interest rate effects in 2010. After the initial increases in loan rates though, bailout expectations no longer had any impact on credit costs. The predicted reduction of funding rates similarly was significant only immediately after TARP stopped. After 2010, we found no significantly reduced deposit or total funding rates among the sample of matched, sound U.S. banks.

Our results thus suggest that the effect of TARP on nonrescued banks' loan rates was short lived. As such, we find no support for the concerns of the COP that competitive distortions, in the sense of more expensive credit, prevailed over a longer period of time.

E. Loan and Deposit Growth

In addition to the predicted effects on loan and deposit rates, Hakenes and Schnabel (2010) anticipate volume effects in response to differences in the likelihood of prospective bailouts. In their model, banks subjected to higher prospective bailouts can use their funding advantages to gain loan and deposit market shares from unprotected competitors. The intuition is that protected banks can afford to attract more deposits at given funding rates, because savers perceive those banks as safe havens. On the credit side, protected banks can offer more competitive interest rates on loans and thereby expand their lending faster than unprotected banks at a given risk level.

To test for possible volume effects, we detail quarterly changes in the level of loans and deposits during q1/10 and q4/13 in Table 11. The first columns show quarterly changes in total loans, real estate loans, and commercial and industrial loans as dependent variable. Increasing bailout expectations exerted no statistically significant effect on loan growth in our sample. Any competitive distortions to credit markets in response to TARP thus appear confined to markup pricing (Berger and Roman, 2015b) rather than creating an expansion of inefficient lending (Dell'Ariccia and Marquez 2004). For deposit growth, we again found no evidence that more protected banks enjoyed stronger inflows of deposits. (6)

One possible explanation for the absence of any volume effects could stem from the different timing of bailout expectation effects. In the immediate aftermath of the crisis, savers may have been eager to seek safe havens, but then they "forgot" about the real possibility of bank failures when determining their required deposit rates (regarding bounded rationality in the subprime crisis, see Gennaioli and Shleifer 2011). Table 12 shows the effects of bailout expectations on loan and deposit growth over time: We find no significant loan growth effects in any of the post-TARP periods and only a very weak immediate reduction in deposits in 2010.

Overall, these results indicated no crowding out of deposit taking or loan granting by banks that were more protected, in terms of higher bailout expectations. Thus, competitive distortions among U.S. banks due to TARP apparently were confined to markup pricing in the immediate aftermath of the support program.

F. Bailout Expectations Across Capitalization and Size Classes

Berger and Roman (2015a, 2015b) show that larger and better capitalized TARP recipients were able to reap competitive advantages and changed their lending patterns differently. Therefore, Tables 13 and 14 reproduce the results of the matched sample in Table 8 with interaction terms for well-capitalized banks and banks of different size classes, respectively. (7) We define well-capitalized banks according to Berger and Roman (2015b) as those that exhibit a total equity ratio of at least 7%. When we specify an according interaction term the results in Table 13 show that the direct positive effect of bailout expectations on loan rates reported in Table 8 is qualitatively confirmed for low-capitalized banks, but not statistically discernible from zero. The interaction terms for both total loans and real estate loans indicate, in turn, that especially well-capitalized banks that are considered more likely to receive support are also able to charge higher loan rates. The differential effect for commercial and industrial loans is also positive, but not statistically significant. Overall, the results are consistent with the finding of Berger and Roman (2015b) that better capitalized banks were able to realize market power due to the TARP program.

The result for deposit rates highlights that the coefficient estimate for bailout expectation in Table 8 is potentially driven by well-capitalized banks because the single term of bailout expectations is insignificant while the interaction effect is negative and significant. The single effect of being a well-capitalized bank, that is, with a bailout expectation of zero, is significantly positive. But with higher bailout expectations, well-capitalized banks can reduce the rates banks offer on deposits significantly in comparison to low-capitalized banks. In summary, higher perceived rescue outlooks seems to render depositors satisfied with lower risk premia for well-capitalized banks.

The stratification of different size classes is based on gross total assets. We distinguish three groups (small, medium, and large) as in Berger and Roman (2015b). Thereby, banks are categorized as small if they have less than $1 billion of total assets. Banks with more than $3 billion of total assets are classified as large banks. Banks with total assets between $1 billion and $3 billion are categorized as medium banks. Small banks are the reference group for the regressions in Table 14. The positive effect of bailout expectations on total loan yields remains intact and is not significantly different for medium or large banks. However, for commercial and industrial lending we find a differential effect of bank size on bailout expectations, namely a mitigating one for banks in the largest size class. Moreover, Table 14 shows that the specification of size class dummies implies that we can estimate the effect of bailout expectations on deposit rates and overall funding cost more precisely. We find that large banks have to pay significantly higher rates for deposits than small and medium banks. The coefficient on total funding yields is significantly negative as in Table 8 and thus in line with theoretical predictions of Hakenes and Schnabel (2010). The interaction terms further suggest that medium and large banks are not different to small banks and also enjoy benefits from lower bailout expectations in terms of having to offer reduced risk premia to suppliers of funding, a result also in line with Berger and Roman (2015b).

G. Branching Restrictions

As noted by Beck et al. (2010), a major challenge to any bailout scheme, even one with perfectly equal disbursement terms, is that banks already operate under distinct competitive conditions. For example, competitive conditions vary widely across U.S. states: Rice and Strahan (2010) even offer an index to gauge states' various implementations of the Riegle-Neal Act, permitting interstate and intrastate branching. Differences in the timing of states' regulation choices to ease entry by out-of-state banks affected lending to small and medium enterprises. Koetter et al. (2012) also show that these differences in branching restrictions after Riegle-Neal can explain differences in Lerner indices across U.S. banks from different states. Similarly, TARP interventions may have led to more pronounced price competition effects in regional banking markets that already were less competitive. By distinguishing three groups of regional banking markets by their values of state-specific branching restrictions, we derive a model of the interaction of bailout expectations with the three indicator variables for markets with low, medium, and high restriction levels.

Regarding the effects on total loan rates, the results in Table 15 indicate an increasing effect of larger bailout expectations. A one standard deviation increase in bailout expectations in a comparably competitive state (e.g., Michigan, with zero restrictions according to Rice and Strahan 2010) prompts a 6.5 basis point hike in mean total loan rates; this increase was 11 basis points in the least competitive states, such as Texas and Iowa. The significance of this pattern varies for real estate and commercial versus industrial lending, but it remains qualitatively intact. Banks that operated in more competitive environments prior to TARP, which presumably already faced thin economic margins, experienced the weakest hikes due to higher bailout expectations. In addition, higher bailout expectations reduced the funding costs in the regional banking markets that were least regulated. Banks operating in increasingly uncompetitive markets instead exhibited no significant reduction in deposit rates.

Overall, the concern that equity support for certain banks could aggravate existing differences in the level of market power seems justified for credit markets. Higher bailout expectations increased loan rates, especially in less competitive markets. With respect to deposit taking, only the least regulated states suffered the negative effect of bailout expectations on interest rates.

IV. CONCLUSION

We have investigated if bank bailouts between q4/08 and q4/09 affected the pricing and growth of loans and deposits among U.S. banks after the program stopped. Specifically, we used political indicators to identify the bailout expectations of U.S. banks through observed TARP equity support, relative to failure, between q4/08 and q4/09. From this revealed assessment of regulators about which types of banks warrant a bailout, we extrapolate bailout expectations among sound banks after TARP stopped.

This empirical test therefore addresses whether bank rescue schemes affected the competitive behavior of not only rescued but also sound banks. Political indicators of the voting behavior on TARP, party membership, and membership on the financial subcommittee are appropriate exclusion restrictions for explaining the probability that a bank will receive a bailout. After controlling for risk differences across banks and local macro conditions, these covariates effectively explain TARP support, but they remain uncorrelated with key measures of pricing power, namely interest rates on loans and deposits.

Using our model parameters to explain TARP support, we generate bailout expectations for the group of sound banks after q4/09. The differences in loan and deposit rates can be explained by these expectations, though doing so requires an adequate counterfactual sample of sound banks that is sufficiently similar to distressed banks until q4/09. After matching distressed banks with sound banks, we demonstrate that an increase in bailout expectations by one standard deviation has a statistically significant effect on loan rates. However, the economic effect on total loan rates is small on the order of 4.5 basis points. Deposit rates fall by around I basis point, which may reflect lower risk premiums required by savers for protected banks. The small economic effects indicate that TARP, despite being statistically significantly related to loan and deposit yields after 2009, did not distort loan and deposit rates of sound banks economically.

Further tests indicate that the interest rate effects of bailout expectations pertain primarily to the immediate aftermath of TARP but become insignificant after 2010. Likewise, we find little indication that protected banks expanded either their lending or deposit taking at the expense of less protected banks. The concerns of the Congressional Oversight Panel (2011), about creating sustained differences in regional banking market competition, to the detriment of smaller banks, thus appear unfounded. However, loan rate increases were largest for well-capitalized banks and banks in states that had been most restrictive in the implementation of interstate branching. Thus, TARP might have aggravated differences in banking competition that existed prior to the rescue period.

doi: 10.1111/ecin.12281

Online Early publication October 30, 2015

ABBREVIATIONS

COP: Congressional Oversight Panel

CPP: Capital Purchase Program

TARP: Troubled Asset Relief Program

REFERENCES

Bayazitova, D., and A. Shivdasani. "Assessing Tarp." Review of Financial Studies, 25, 2012, 377-407.

Beck, T., D. Coyle, M. Dewatripont, X. Freixas, and P. Seabright. "Bailing Out the Banks: Reconciling Stability and Competition." CEPR Report, Centre for Economic Policy Research, London, 2010.

Berger, A., and Roman, R. "Did Saving Wall Street Really Save Main Street? The Real Effects of TARP on Local Economic Conditions." University of South Carolina Working Paper, 2015a.

--. Forthcoming. "Did TARP Banks Get Competitive Advantages?" Journal of Financial and Quantitative Analysis, 2015b.

Black, L., and L. Hazelwood. "The Effect of TARP on Bank Risk-Taking." Journal of Financial Stability, 9, 2013, 790-803.

Calderon, C., and K. Schaeck. Forthcoming. "The Effects of Government Interventions in the Financial Sector on Banking Competition and the Evolution of Zombie Banks." Journal of Financial and Quantitative Analysis, 2015

Calomiris, C., and U. Khan. "An Assessment of TARP Assistance to Financial Institutions." Journal of Economic Perspectives, 29, 2015, 53-80.

Cole, R., and L. White. "Deja Vu All Over Again: The Causes of U.S. Commercial Bank Failures This Time Around." Journal of Financial Services Research, 42, 2012, 5-29.

Congressional Oversight Panel. Final Report, March. Washington, DC: Congressional Oversight Panel, 2011.

Dam, L., and M. Koetter. "Bank Bailouts and Moral Hazard: Evidence from Germany." Review of Financial Studies, 25, 2012, 2343-80.

Dell'Ariccia, G., and R. Marquez. "Information and Bank Credit Allocation." Journal of Financial Economics, 72, 2004, 185-214.

DeYoung, R., E. Peng, and M. Yan. "Executive Compensation and Business Policy Choices at U.S. Commercial Banks." Journal of Financial and Quantitative Analysis, 48, 2013, 165-96.

Duchin, R., and D. Sosyura. "The Politics of Government Investment." Journal of Financial Economics, 106, 2012, 24-48.

--. "Safer Ratios, Riskier Portfolios: Banks' Response to Government Aid." Journal of Financial Economics, 113, 2014, 1-28.

Gennaioli, N., and A. Shleifer. "What Comes to Mind." Quarterly Journal of Economics, 125, 2011, 1399-433.

Gropp, R., H. Hakenes, and I. Schnabel. "Competition, Risk-Shifting, and Public Bail-Out Policies." Review of Financial Studies, 24, 2011, 2084-120.

Hakenes, H., and I. Schnabel. "Banks without Parachutes: Competitive Effects of Government Bail-Out Policies." Journal of Financial Stability, 6, 2010, 156-68.

Huang, R., and L. Ratnovski. "The Dark Side of Bank Wholesale Funding." Journal of Financial Intermediation, 20, 2011,248-63.

Kashyap, A., and J. Stein. "What Do a Million Observations on Banks Say about the Transmission of Monetary Policy?" American Economic Review, 90, 2000, 407-28.

Keeley, M. "Deposit Insurance, Risk and Market Power in Banking." American Economic Review, 80, 1990, 1183-200.

Koetter, M., J. Kolari, and L. Spierdijk. "Enjoying the Quiet Life under Deregulation? Evidence from Adjusted Lerner Indices for U.S. Banks." Review of Economics and Statistics, 94, 2012, 462-80.

Lambert, C., F. Noth, and U. Schiiwer. Forthcoming. "How Do Insured Deposits Affect Bank Stability? Evidence from the 2008 Emergency Economic Stabilization Act." Journal of Financial Intermediation, 2015.

Li, L. "TARP Funds Distribution and Bank Loan Supply." Journal of Banking and Finance, 37, 2013, 4777-92.

Liu, W., J. Kolari, T. Kyle Tippens, and D. Fraser. "Did Capital Infusions Enhance Bank Recovery from the Great Recession?" Journal of Banking and Finance, 37, 2013, 5048-61.

Rice, T., and P. Strahan. "Does Credit Competition Affect Small-Firm Finance?" Journal of Finance, 65, 2010, 861-89.

Wall Street Journal. "U.S. Eyes Bank Pay Overhaul." 2009. Accessed September 1, 2015. http://online.wsj.com/ article/SB 124215896684211987.html.

Wheelock, D., and P. Wilson. "Why Do Banks Disappear? The Determinants of U.S. Bank Failures and Acquisitions." Review of Economics and Statistics, 82, 2000, 127-38.

MICHAEL KOETTER and FELIX NOTH *

* We thank two anonymous referees and the editor Martin Gervais for their valuable input. We are grateful for the feedback received at the 2013 meetings of the European Economic Association, the German Finance Association, the Financial Management Association, and the 2014 SUERF/FinLawMetrics conference. We also received valuable feedback at the seminar series of the Universities of Cologne and Diisseldorf as well as the European Central Bank. Without implicating them, we thank Allen Berger, Philipp Hartmann, Marie Hoerova, Gunter Loffler, Stephen Karolyi, Andrea Tiseno, and Michael Wedow for questions, discussions, and input. This paper has been prepared by the authors under the Wim Duisenberg Research Fellowship Programme sponsored by the ECB. Any views expressed are only those of the authors and do not necessarily represent the views of the ECB or the Eurosystem. All errors are our own.

Koetter: Frankfurt School of Finance & Management, 60314, Frankfurt am Main, Germany; Halle Institute for Economic Research (IWH), 06017, Halle (Saale), Germany. Phone +49 69153008446, Fax +49 691530084446, E-mail m.koetter@fs.de

Noth: Halle Institute for Economic Research (IWH), 06017, Halle (Saale), Germany; Otto-von-Guericke University, 39114, Magdeburg, Germany. Phone +49 345 7753 702, Fax +49 345 7753 820, E-mail felix.noth@iwh-halle.de

(1.) Note that this mechanism also holds in the presence of deposit insurance, given insurance caps of $250,000 for deposits that apply to all banks equally (Lambert et al. 2015) since October 2008 and $100,000 before that date. During the period when we extract bailout expectations, only around 60% of deposits are insured in our sample, thus leaving a substantial uninsured portion of retail funding. Moreover, Huang and Ratnovski (2011) show that the share of generally uninsured wholesale funding dominated retail borrowing in recent years.

(2.) Details on the exact data items from UBPR and the calculation are in Table 3.

(3.) We consider a bank a commercial bank if the FDIC's data item "charter class" is either "N," "NM," or "SM." That excludes all state chartered savings banks and thrifts and OCC (OTS) supervised federally chartered thrifts.

(4.) We winsorize all bank variables at the 0.5% and 99.5% levels.

(5.) In unreported tests, we confirmed that our results did not reflect only those banks regarded as "too big to fail," by estimating Equation (2) without the very big banks (Duchin and Sosyura 2014).

(6.) In unreported robustness checks, we leave out lagged loan and deposit shares as explanatory variables to account for the fact that both variables are very likely to be correlated with growth rates of loans and deposits. As for the baseline, we find no significant effects of bailout expectations on credit growth. But the negative effect on deposit growth is significant.

(7.) Note that we leave out the continuous control variable for equity in Table 13 and size in Table 14. Both variables are highly correlated with the introduced dummy variables. However, the results are not qualitatively different when we leave the continuous terms in the regressions.

TABLE 1
Distressed and Sound Banks

        TARP   Fail   Entry   Sound   Survivor

q4/08   171     12      0     5,837    6,008
q1/09   239     24      1     5,745    5,985
q2/09    81     21     33     5,883    5,997
q3/09    30     42      9     5,925    5,964
q4/09    27     37      0     5,900    5,927
q1/10     0     35      0     5,892    5,892
q2/10     0     36      6     5,856    5,862
q3/10     0     31      1     5,831    5,832
q4/10     0     27      0     5,805    5.805
q1/11     0     24      2     5,781    5,783
q2/11     0     19     10     5,764    5,774
q3/11     0     24      7     5,750    5,757
q4/11     0     17      4     5,740    5.744
q1/12     0     13      8     5,731    5,739
q2/12     0     11     14     5,728    5,742
q3/12     0     10      2     5.732    5,734
q4/12     0      6      6     5,728    5,734
q1/13     0      3      7     5,731    5,738
q2/13     0     11      2     5,727    5,729
q3/13     0      6      3     5,723    5,726
q4/13     0      2      0     5,724    5,724

Notes: The columns of Table 1 list the number of banks that received
assistance from the trouble asset relief program (TARP), failed
(Fail), entered the sample (Entry), or were sound, not in distress at
a particular point in time. The last column shows the number of banks
that survived at the end of a quarter between q4/08 and q4/13. In
this table, we include the nine banks that failed during the crisis
period while their holding company received TARP, though we exclude
these cases in our regression analysis.

TABLE 2
Size of TARP Banks

             Sum of Total Assets ($billion)

                     TARP

          All       Forced     Other      Sound

q4/08   4,023.59   2,532.19   1,491.39   2,940.18
q1/09   4,118.00   2,466.66   1,651.35   2,727.04
q2/09   4,122.16   2,443.76   1,678.40   2,687.32
q3/09   4,253.08   2,468.62   1,784.46   2,687.33
q4/09   4,425.83   2,479.84   1,946.00   2,687.34
q1/10   4,913.99   2,987.70   1,926.29   2,687.35
q2/10   4,823.35   2,890.65   1,932.70   2,687.36
q3/10   4,933.49   2,973.26   1,960.23   2,687.37
q4/10   4,937.41   2,980.60   1,956.81   2,687.38
q1/11   5,063.04   3,075.87   1,987.17   2,687.39
q2/11   5,218.12   3,176.24   2,041.87   2,687.40
q3/11   5,356.53   3,264.77   2,091.76   2,687.41
q4/11   5,425.53   3,210.25   2,215.28   2,687.42
q1/12   5,451.84   3,233.67   2,218.17   2,687.43
q2/12   5,508.04   3,216.00   2,292.04   2,687.44
q3/12   5,614.10   3,370.17   2,243.92   2,687.45
q4/12   5,880.63   3,491.13   2,389.50   2,687.46
q1/13   5,904.50   3.540.57   2,363.92   2,687.47
q2/13   5,961.98   3,567.85   2,394.13   2,687.48
q3/13   6.084.01   3,659.43   2,424.58   2,687.49
q4/13   6,185.00   3,689.88   2,495.12   2,687.50

           Average of Total Assets ($billion)

                     TARP

          All       Forced     Other      Sound

q4/08    23.53      422.03      9.04       0.54
q1/09    10.04      411.11      4.09       0.52
q2/09     8.40      407.29      3.46       0.52
q3/09     8.16      411.44      3.46       0.53
q4/09     8.08      413.31      3.59       0.55
q1/10     8.97      497.95      3.55       0.56
q2/10     8.80      481.78      3.57       0.57
q3/10     9.00      495.54      3.62       0.58
q4/10     9.01      496.77      3.61       0.59
q1/11     9.24      512.64      3.67       0.60
q2/11     9.52      529.37      3.77       0.62
q3/11     9.77      544.13      3.86       0.64
q4/11     9.90      642.05      4.08       0.65
q1/12     9.95      646.73      4.09       0.66
q2/12    10.05      643.20      4.22       0.69
q3/12    10.24      561.70      4.14       0.70
q4/12    10.73      581.85      4.41       0.71
q1/13    10.77      590.10      4.36       0.71
q2/13    10.90      594.64      4.43       0.71
q3/13    11.12      609.90      4.48       0.72
q4/13    11.31      614.98      4.61       0.74

Notes: The columns of Table 2 show the sum (average) of total assets
in $billion per quarter between q4/08 and q4/13 for the groups of
TARP and sound banks. We further split the sample of TARP banks
according to those that were forced to accept TARP: Citigroup, JP
Morgan, Bank of America (including Merrill Lynch), Goldman Sachs,
Morgan Stanley, State Street, Bank of New York Mellon, and Wells
Fargo (including Wachovia).

TABLE 3
Variable Description

FDIC Variables     Calculation by
Variable Name      FDIC Codes           Description

Size               ln(asset)            Total assets: Log of total
                                          assets. Total assets
                                          comprise the sum of all
                                          assets owned by the
                                          institution including cash,
                                          loans, securities, bank
                                          premises, and other assets.
                                          This total does not include
                                          off-balance-sheet accounts.
EQ                 eqtot/asset          Total equity over assets:
                                          Banks' total equity
                                          capital.
NPA                (p3asset + p9asset   Nonperforming assets over
                     + naasset)/asset     total assets: Total assets
                                          past due 30-90 days and
                                          still accruing interest
                                          (p3asset). Total assets
                                          past due 90 or more days
                                          and still accruing interest
                                          (p9asset). Total assets,
                                          which are no longer
                                          accruing interest
                                          (naasset). Total assets
                                          include real estate loans,
                                          installment loans, credit
                                          cards and related plan
                                          loans, commercial loans and
                                          all other loans, lease
                                          financing receivables, debt
                                          securities, and other
                                          assets.
Loans              lnlsgr/asset         Total loans over assets:
                                          Total loans and lease
                                          financing receivables, net
                                          of unearned income.
RoA                roaptx               Return on assets: Profits
                                          before taxes over total
                                          assets.
Deposits           dep/asset            Total deposits over assets:
                                          The sum of all deposits
                                          over total assets.
Cash               cbal/asset           Total cash balances over
                                          assets: The sum of all
                                          cash balances over total
                                          assets.
[DELTA]total       ln(lnlsgr)--ln(L.    Growth rate of total loans:
  loans              lnlnlsgr)            One-quarter growth rate of
                                          loans secured primarily by
                                          real estate, whether
                                          originated by the bank or
                                          purchased.
[DELTA]real        ln(lnre)--ln(L.      Growth rate of real estate
  estate loans       Lire)                loans: One-quarter growth
                                          rate of total loans and
                                          lease financing
                                          receivables, net of
                                          unearned income.
[DELTA]C&I loans   ln(lnci)--ln(L.      Growth rate of C & l loans:
                     lnci)                One-quarter growth rate of
                                          loans other than loans
                                          secured by real estate,
                                          loans to individuals, loans
                                          to depository institutions
                                          and foreign governments,
                                          and loans to states and
                                          political subdivisions and
                                          lease financing
                                          receivables.
[DELTA]deposits    ln(dep)--ln(L.       Growth rate of deposits:
                     dep)                 One-quarter growth rate of
                                          total deposits.
CAPH               eqtot/asset          Capitalization dummy: This
                                          dummy is one if a bank has
                                          a larger equity-to-total
                                          assets ratio than 7% and
                                          zero otherwise.
Size ()            asset                Size dummy: The three dummies
                                          (Size (small), Size
                                          (medium), Size (large))
                                          indicate banks with total
                                          assets of less than $1
                                          billion (small), more than
                                          $3 billion (large), and
                                          medium if total assets are
                                          between $1 billion and $3
                                          billion.

Further
variables
Variable Name      UBPR Data Item       Description

[R.sup.TL]         UBPRE686             Interest rate on total loans:
                                          Quarterly (annualized)
                                          yield on total loans for
                                          each bank, which reflects
                                          the ratio of interest and
                                          fees on loans and income on
                                          direct lease financing
                                          receivables (including tax
                                          benefit on tax exempt on
                                          loan and lease income) to
                                          average total loans and
                                          lease financing
                                          receivables.
[R.sup.RF]         UBPRE688             Interest rate on real estate
                                          loans: Quarterly
                                          (annualized) yield on real
                                          estate loans for each bank,
                                          which reflects the ratio of
                                          interest and fees on
                                          domestic office loans
                                          secured primarily by real
                                          estate to average domestic
                                          real estate loans.
[R.sup.CI]         UBPRE689             Interest rate on commercial
                                          and industrial loans:
                                          Quarterly (annualized)
                                          yield on commercial and
                                          industrial loans for each
                                          bank, which reflects the
                                          ratio of interest and fees
                                          on domestic office
                                          commercial and industrial
                                          loans to average domestic
                                          commercial and industrial
                                          loans.
[R.sup.DEP]        UBPRE701             Interest rate on total
                                          deposits: Quarterly
                                          (annualized) cost of total
                                          interest-bearing deposits
                                          for each bank, which
                                          reflects the ratio of
                                          interest on all
                                          interest-bearing time and
                                          savings deposits in
                                          domestic and foreign
                                          offices to average
                                          interest-bearing time and
                                          savings deposits in
                                          domestic and foreign
                                          offices.
[R.sup.TF]         UBPRE710             Interest rate on total
                                          funding: Quarterly
                                          (annualized) cost of all
                                          interest-bearing funds for
                                          each bank, which reflects
                                          the ratio of interest on
                                          all interest-bearing
                                          deposits in domestic
                                          offices, interest-bearing
                                          foreign office deposits,
                                          demand notes issued to the
                                          U.S. Treasury, other
                                          borrowed money,
                                          subordinated notes and
                                          debentures, and expense on
                                          federal funds purchased and
                                          securities sold under
                                          agreements to repurchase,
                                          interest expense on
                                          mortgage and capitalized
                                          leases to the average of
                                          the liabilities or funds
                                          that generated those
                                          expenses.
UR                                      Quarterly rate of
                                          unemployment per county:
                                          Using county-level
                                          information provided by the
                                          Bureau of Labor Economics,
                                          we weighted the
                                          unemployment rates for each
                                          bank by its county
                                          presence, according to the
                                          summary of deposits.
Bailout                                 Bailout expectation:
  expectation                             Predicted probability from
                                          regression coefficients
                                          that result from probit
                                          regression of Equation (1).
Branching index                         Branching restriction index:
                                          According to Rice and
                                          Strahan (2010), an index
                                          that separates states
                                          according to their
                                          branching restrictions. A
                                          higher value indicates more
                                          restrictions.
Case-Shiller                            CS index: The Case-Shiller
  index                                   house price index per state
                                          provided by the economic
                                          research center of the Fed
                                          of St. Louis.
SubC                                    Member of subcommittee: A
                                          dummy variable that
                                          indicates whether the
                                          Congressperson is part of
                                          the financial services
                                          subcommittee. The ending
                                          0709 indicates membership
                                          for the period between 2007
                                          and 2009 and the ending
                                          0911 indicates membership
                                          between 2009 and 2011.
2nd Vote                                Second vote on TARP: A dummy
                                          variable indicating the
                                          Congressperson's vote in
                                          the second Congressional
                                          TARP vote.
Party                                   Party of member: A dummy
                                          variable that indicates the
                                          party membership of each
                                          Congressperson. The ending
                                          0709 indicates membership
                                          for the period between 2007
                                          and 2009, and the ending
                                          0911 indicates membership
                                          between 2009 and 2011.

Notes: The source for all FDIC variables and their descriptions is
the FDIC Statistics on Depository Institutions website. For more
details, refer to http://www2.fdic.gov/SDI/main.asp.

TABLE 4
Descriptive Statistics

                                      q4/08-q4/09

                               TARP                 Fail

                          Mean        SD       Mean        SD

Exclusions restrictions first stage: political variables

  2nd Vote                0.5124    0.5000     0.3971    0.4898
  SC0709                  0.1200    0.3251     0.1408    0.3481
  SC0911                  0.1191    0.3240     0.1492    0.3566
  Party0709               0.4717    0.4993     0.3571    0.4797
  Party0911               0.5096    0.5000     0.4118    0.4927

Dependent variables second stage: loan and funding interest rates

  [R.sup.TL]              0.0601    0.0089     0.0552    0.0119
  [R.sup.RE]              0.0593    0.0088     0.0546    0.0097
  [R.sup.CI]              0.0616    0.0183     0.0626    0.0228
  [R.sup.DEP]             0.0205    0.0068     0.0312    0.0077
  [R.sup.TF]              0.0213    0.0067     0.0307    0.0080

Main explanatory variable second stage: bailout expectations

  Bailout expectation     0.9842    0.0813     0.2492    0.3882

Independent variables first/second stage: bank and regional
characteristics

  EQ                      0.1028    0.0323     0.0594    0.0295
  RoA                    -0.0005    0.0098    -0.0266    0.0194
  NPA                     0.0340    0.0243     0.1246    0.0594
  Size                   13.3297    1.5316    12.7946    1.5477
  Cash                    0.0495    0.0576     0.0635    0.0641
  Deposits                0.7862    0.0779     0.8361    0.1255
  Loans                   0.7210    0.1250     0.7279    0.1250
  CS index              346.4701   86.0662   345.3248   68.8263
  UR                      0.0860    0.0251     0.0822    0.0257

                            q4/08-q4/09          q1/10-q4/13

                               Sound                TARP

                          Mean        SD       Mean        SD

Exclusions restrictions first stage: political variables

  2nd Vote                0.4769    0.4995
  SC0709                  0.0765    0.2658
  SC0911                  0.0692    0.2539
  Party0709               0.4455    0.4970
  Party0911               0.4621    0.4986

Dependent variables second stage: loan and funding interest rates

  [R.sup.TL]              0.0654    0.0099     0.0567    0.0090
  [R.sup.RE]              0.0642    0.0097     0.0557    0.0085
  [R.sup.CI]              0.0664    0.0200     0.0599    0.0190
  [R.sup.DEP]             0.0213    0.0068     0.0094    0.0050
  [R.sup.TF]              0.0219    0.0067     0.0105    0.0053

Main explanatory variable second stage: bailout expectations

  Bailout expectation     0.9082    0.2663     0.9350    0.2171

Independent variables first/second stage: bank and regional
characteristics

  EQ                      0.1082    0.0497     0.1034    0.0255
  RoA                     0.0015    0.0120     0.0021    0.0086
  NPA                     0.0374    0.0430     0.0414    0.0342
  Size                   11.8519    1.1376    13.2748    1.4703
  Cash                    0.0630    0.0637     0.0753    0.0688
  Deposits                0.8262    0.0765     0.8253    0.0631
  Loans                   0.6633    0.1512     0.6688    0.1295
  CS index              304.9202   76.1083   318.3926   76.3886
  UR                      0.0782    0.0293     0.0848    0.0233

                                      q1/10-q4/13

                               Fail                 Sound

                          Mean        SD       Mean        SD

Exclusions restrictions first stage: political variables

  2nd Vote
  SC0709
  SC0911
  Party0709
  Party0911

Dependent variables second stage: loan and funding interest rates

  [R.sup.TL]              0.0532    0.0081     0.0608    0.0095
  [R.sup.RE]              0.0529    0.0093     0.0595    0.0090
  [R.sup.CI]              0.0611    0.0230     0.0634    0.0203
  [R.sup.DEP]             0.0162    0.0057     0.0098    0.0049
  [R.sup.TF]              0.0171    0.0057     0.0104    0.0051

Main explanatory variable second stage: bailout expectations

  Bailout expectation     0.1012    0.2729     0.9501    0.1899

Independent variables first/second stage: bank and regional
characteristics

  EQ                      0.0498    0.0210     0.1095    0.0348
  RoA                    -0.0217    0.0179     0.0043    0.0076
  NPA                     0.1513    0.0473     0.0303    0.0323
  Size                   12.2225    1.0087    11.9338    1.1373
  Cash                    0.1018    0.0674     0.0951    0.0825
  Deposits                0.8912    0.0542     0.8469    0.0588
  Loans                   0.7040    0.0940     0.6051    0.1533
  CS index              299.5813   53.4818   286.0212   68.6227
  UR                      0.0983    0.0218     0.0783    0.0270

Notes: Table 4 shows descriptive statistics for banks that received
assistance from TARP, failed (Failed), or did not have severe
troubles (Sound), and thus received no money from TARP and did not
fail. The table presents descriptive statistics (mean and standard
deviation) for the crisis period between the last quarter of 2008 and
the last quarter of 2009 and for the subsequent period until the last
quarter of 2013. Variable definitions are in Table 3.

TABLE 5
Bailout Regression Results

                     Dependent Variable: Tarp/Fail-Dummy

                       (1)           (2)           (3)

2nd Vote            0.0222 **     0.0226 **
                    (0.0104)      (0.0099)
SC0709              0.0294 *                     -0.0097
                    (0.0163)                    (0.0075)
SC0911             -0.0372 ***
                    (0.0139)
Party0709           0.0230 **
                    (0.0092)
Party0911          -0.0164 **
                    (0.0078)
L.EQ               1.5688 ***    1.4085 ***    1.3041 ***
                    (0.4347)      (0.3580)      (0.3558)
L.RoA              0.8323 ***    0.9762 ***    0.8671 ***
                    (0.3190)      (0.3006)      (0.2606)
L.NPA              -0.5744 ***   -0.5675 ***   -0.6072 ***
                    (0.1156)      (0.1188)      (0.1281)
L.Size               0.0042        0.0032        0.0029
                    (0.0030)      (0.0027)      (0.0030)
L.Cash             -0.1893 ***   -0.1998 ***   -0.1852 ***
                    (0.0722)      (0.0674)      (0.0708)
L.Deposits           -0.0372       -0.0305       -0.0573
                    (0.0653)      (0.0619)      (0.0702)
L.Loans            -0.1121 **    -0.1042 **    -0.0745 **
                    (0.0513)      (0.0420)      (0.0366)
L.CS index         0.0001 ***    0.0001 ***    0.0002 ***
                    (0.0001)      (0.0001)      (0.0001)
L.UR                 0.0963        0.0484        0.0775
                    (0.1443)      (0.1291)      (0.1591)
Observations           675           675           675
Pseudo [R.sup.2]     0.9330        0.9258        0.9179
Log likelihood       -21.87        -24.20        -26.78
F-value               15.53
p Value              0.0083

                     Dependent Variable: Tarp/Fail-Dumim

                       (4)           (5)           (6)

2nd Vote

SC0709

SC0911             -0.0162 **
                    (0.0074)
Party0709                         0.0160 *
                                  (0.0084)
Party0911                                        0.0108
                                                (0.0074)
L.EQ               1.3120 ***    1.3750 ***    1.3426 ***
                    (0.3537)      (0.3483)      (0.3366)
L.RoA              0.8670 ***    0.8425 ***    0.9317 ***
                    (0.2545)      (0.2708)      (0.2914)
L.NPA              -0.5998 ***   -0.6488 ***   -0.6375 ***
                    (0.1254)      (0.1276)      (0.1276)
L.Size               0.0031        0.0034        0.0032
                    (0.0030)      (0.0030)      (0.0029)
L.Cash             -0.1787 ***   -0.2108 ***   -0.2121 ***
                    (0.0693)      (0.0729)      (0.0754)
L.Deposits           -0.0625       -0.0524       -0.0538
                    (0.0737)      (0.0641)      (0.0664)
L.Loans            -0.0700 **    -0.0956 **    -0.0900 **
                    (0.0356)      (0.0402)      (0.0397)
L.CS index         0.0002 ***     0.0001 **    0.0002 ***
                    (0.0001)      (0.0001)      (0.0001)
L.UR                 0.0831        0.0869        0.0677
                    (0.1625)      (0.1549)      (0.1565)
Observations           675           675           675
Pseudo [R.sup.2]     0.9201        0.9211        0.9188
Log likelihood       -26.08        -25.77        -26.50
F-value
p Value

Notes: Table 5 contains the results for regressions explaining
whether a bank received assistance from TARP between q4/08 and q4/09
(1) or failed (0), as outlined in Equation (1). Only banks that
received TARP or failed during this period are considered. The prefix
"L" indicates that a variable was lagged by one-quarter. Coefficients
are marginal effects. The first column shows results with bank
characteristics, all political variables, and the regional
unemployment rate and Case-Shiller index on U.S. state level as
explanatory variables. The remaining columns show results for each
political variable separately. The F-value and reported p value
denote whether all political variables are jointly significant in
explaining whether a bank receives government support or fails.
Variable definitions are in Table 3. Clustered (bank level) standard
errors are in parentheses.

***, **, and * indicate significant coefficients at the 1%, 5%, and
10% levels, respectively.

TABLE 6 Weak Correlation Test of Political Instruments and Yields

                                  Interest Rates R

                     Total Loans   Real Estate Loans    C&I Loans
Dependent Variable       (1)              (2)              (3)

2nd Vote               0.0007           -0.0031          0.0034
                      (0.0014)         (0.0026)         (0.0037)
SC0709                 -0.0008         -0.0040 *         0.0019
                      (0.0015)         (0.0023)         (0.0062)
SC0911                 0.0011          0.0040 *          -0.0053
                      (0.0019)         (0.0023)         (0.0066)
Party0709             0.0020 *          0.0011         0.0117 ***
                      (0.0012)         (0.0012)         (0.0040)
Party0911            -0.0032 **         -0.0012        -0.0161 ***
                      (0.0016)         (0.0016)         (0.0045)
L.EQ                  0.0099 **        0.0083 *          0.0142
                      (0.0048)         (0.0049)         (0.0109)
L.RoA                0.0385 ***       0.0279 ***       0.0453 ***
                      (0.0056)         (0.0066)         (0.0134)
L.NPA                -0.0253 ***      -0.0283 ***        -0.0026
                      (0.0033)         (0.0033)         (0.0078)
L.Size                 -0.0012        -0.0016 ***        -0.0002
                      (0.0007)         (0.0006)         (0.0016)
L.Cash                0.0018 *         0.0021 **         0.0035
                      (0.0010)         (0.0010)         (0.0025)
L.Deposits            0.0043 **        0.0036 **         -0.0066
                      (0.0018)         (0.0017)         (0.0053)
L.Loans              -0.0109 ***      -0.0082 ***      -0.0107 ***
                      (0.0010)         (0.0009)         (0.0028)
E CS index             0.0000          0.0000 *          0.0000
                      (0.0000)         (0.0000)         (0.0000)
L.UR                   0.0004           -0.0029          0.0002
                      (0.0018)         (0.0024)         (0.0065)
Constant             0.0765 ***       0.0822 ***       0.0718 ***
                      (0.0091)         (0.0082)         (0.0193)
TE                       Yes              Yes              Yes
FE                       Yes              Yes              Yes
YE x SE                  Yes              Yes              Yes
No. of Banks            5,416            5,416            5,416
Observations           83,550           83,550           83,550
Adj. [R.sup.2]          0.84             0.70             0.55
E-value                 1.12             1.10             2.80
p Value                0.3492           0.3591           0.0158

                           Interest Rates R

                      Deposits     Total Funding
Dependent Variable       (4)            (5)

2nd Vote               -0.0003        -0.0003
                      (0.0004)       (0.0004)
SC0709                 -0.0002        -0.0001
                      (0.0004)       (0.0005)
SC0911                 -0.0004        -0.0004
                      (0.0005)       (0.0005)
Party0709              0.0003         0.0004
                      (0.0007)       (0.0008)
Party0911              0.0002         -0.0000
                      (0.0008)       (0.0009)
L.EQ                 -0.0038 ***    -0.0102 ***
                      (0.0013)       (0.0015)
L.RoA                -0.0142 ***    -0.0155 ***
                      (0.0017)       (0.0017)
L.NPA                -0.0046 ***    -0.0036 ***
                      (0.0010)       (0.0010)
L.Size               0.0010 ***     0.0009 ***
                      (0.0002)       (0.0002)
L.Cash                 -0.0005        -0.0003
                      (0.0003)       (0.0003)
L.Deposits             0.0008       -0.0053 ***
                      (0.0007)       (0.0010)
L.Loans              0.0011 ***      0.0007 **
                      (0.0003)       (0.0004)
E CS index           0.0000 ***     0.0000 ***
                      (0.0000)       (0.0000)
L.UR                 0.0020 ***     0.0021 ***
                      (0.0007)       (0.0008)
Constant               -0.0010      0.0065 ***
                      (0.0022)       (0.0023)
TE                       Yes            Yes
FE                       Yes            Yes
YE x SE                  Yes            Yes
No. of Banks            5,416          5,416
Observations           83,550         83,550
Adj. [R.sup.2]          0.93           0.93
E-value                 0.68           0.52
p Value                0.6404         0.7582

Notes: Table 6 represents a panel regression with bank (FE), quarter
(TE), and interacted year-state (YE x SE) fixed effects for the
period q1/10-q4/13. The F-values and the reported p values indicate
whether all political variables are jointly significant in explaining
yields on loans and funding in the period after q4/09. The prefix "L"
indicates that a variable is lagged by one-quarter. Variable
definitions are in Table 3. Clustered (bank level) standard errors
are in parentheses.

***, **, and * indicate significant coefficients at the 1%, 5%, and
10% levels, respectively.

TABLE 7
Bailout Expectation Effects on Lending and Funding Rates

                                    Interest Rates R

                      Ttotal Loans   Real Estate Loans    C&I Loans
Dependent Variable        (1)               (2)              (3)

Bailout expectation    0.0014 ***        0.0012 **        0.0025 **
                        (0.0004)         (0.0005)         (0.0011)
L.EQ                     0.0073           0.0058           0.0096
                        (0.0049)         (0.0049)         (0.0113)
L.RoA                  0.0321 ***       0.0220 ***        0.0335 **
                        (0.0050)         (0.0063)         (0.0136)
L.NPA                 -0.0223 ***       -0.0256 ***        0.0031
                        (0.0032)         (0.0034)         (0.0080)
L.Size                 -0.0012 *        -0.0017 ***        -0.0003
                        (0.0007)         (0.0006)         (0.0016)
L.Cash                 0.0022 **         0.0025 **        0.0043 *
                        (0.0010)         (0.0010)         (0.0025)
L.Deposits             0.0043 **         0.0036 **         -0.0065
                        (0.0018)         (0.0017)         (0.0053)
L. Loans              -0.0107 ***       -0.0081 ***      -0.0105 ***
                        (0.0010)         (0.0009)         (0.0028)
L.CS index               0.0000           0.0000           0.0000
                        (0.0000)         (0.0000)         (0.0000)
EUR                      0.0001           -0.0031          -0.0005
                        (0.0018)         (0.0024)         (0.0065)
Constant               0.0756 ***       0.0811 ***       0.0691 ***
                        (0.0091)         (0.0082)         (0.0191)
TE                        Yes               Yes              Yes
FE                        Yes               Yes              Yes
YE x SE                   Yes               Yes              Yes
No. of Banks             5,416             5,416            5,416
Observations             83,550           83,550           83,550
Adj. [R.sup.2]            0.84             0.70             0.55

                            Interest Rates R

                       Deposits     Total Funding
Dependent Variable        (4)            (5)

Bailout expectation     -0.0001        0.0000
                       (0.0001)       (0.0001)
L.EQ                  -0.0037 ***    -0.0102 ***
                       (0.0013)       (0.0015)
L.RoA                 -0.0139 ***    -0.0155 ***
                       (0.0017)       (0.0018)
L.NPA                 -0.0047 ***    -0.0036 ***
                       (0.0010)       (0.0011)
L.Size                0.0010 ***     0.0009 ***
                       (0.0002)       (0.0002)
L.Cash                  -0.0005        -0.0003
                       (0.0003)       (0.0004)
L.Deposits              0.0008       -0.0053 ***
                       (0.0007)       (0.0010)
L. Loans              0.0011 ***      0.0007 **
                       (0.0003)       (0.0004)
L.CS index            0.0000 ***     0.0000 ***
                       (0.0000)       (0.0000)
EUR                   0.0020 ***     0.0022 ***
                       (0.0007)       (0.0008)
Constant                -0.0009      0.0066 ***
                       (0.0022)       (0.0023)
TE                        Yes            Yes
FE                        Yes            Yes
YE x SE                   Yes            Yes
No. of Banks             5,416          5,416
Observations            83,550         83,550
Adj. [R.sup.2]           0.93           0.93

Notes: Table 7 shows regression results for Equation (2). Each
regression includes bank (FE), quarter (TE), and interacted
year-state (YE x SE) fixed effects for the period q1/10-q4/l2. The
prefix "L" indicates that a variable is lagged by one-quarter.
Variable definitions are in Table 3. Clustered (bank level) standard
errors are in parentheses.

***, **, and * indicate significant coefficients at the 1%, 5%, and
10% levels, respectively.

TABLE 8
Validity of Extrapolated Bailout Expectations

                                     Interest Rates R

                        Total Loans   Real Estate Loans   C&I Loans
Dependent Variable          (1)              (2)             (3)

Baseline
  Bailout expectation   0.0014 ***        0.0012 **       0.0025 **
  Adj. [R.sup.2]         (0.0004)         (0.0005)         (0.0011)
                          0.8408           0.7022           0.5494
  No. of banks             5,416            5,416           5,416
  Observations            83,550           83,550           83,550
Without failures
  Bailout expectation   0.0013 ***        0.0011 **       0.0023 **
                         (0.0005)         (0.0005)         (0.0011)
  Adj. [R.sup.2]          0.8413           0.7026           0.5504
  No. of banks             5,177            5,177           5,177
  Observations            82,268           82,268           82,268
With TARP
  Bailout expectation   0.0013 ***        0.0011 **       0.0027 ***
                         (0.0004)         (0.0004)         (0.0010)
  Adj. [R.sup.2]          0.8370           0.7000           0.5515
  No. of banks             5,964            5,964           5,964
  Observations            92,315           92,315           92,315
TARP only
  Bailout expectation     0.0005           0.0005           0.0039
  Adj. [R.sup.2]         (0.0009)         (0.0009)         (0.0025)
                          0.7777           0.6415           0.5737
  No. of banks              548              548             548
  Observations             8,765            8,765           8,765
Matched sample
  Bailout expectation    0.0024 **         0.0024          0.0043 *
                         (0.0011)         (0.0014)         (0.0025)
  Adj. [R.sup.2]          0.8454           0.6934           0.6226
  No. of banks              597              597             597
  Observations             8,654            8,654           8,654
  TE                        Yes              Yes             Yes
  FE                        Yes              Yes             Yes
  YE x SE                   Yes              Yes             Yes
  Controls                  Yes              Yes             Yes

                            Interest Rates R

                        Deposits    Total Funding
Dependent Variable         (4)           (5)

Baseline
  Bailout expectation    -0.0001       0.0000
  Adj. [R.sup.2]        (0.0001)      (0.0001)
                         0.9262        0.9256
  No. of banks            5,416         5,416
  Observations           83.550        83.550
Without failures
  Bailout expectation    -0.0001       -0.0000
                        (0.0001)      (0.0001)
  Adj. [R.sup.2]         0.9245        0.9238
  No. of banks            5,177         5,177
  Observations           82.268        82.268
With TARP
  Bailout expectation    -0.0001       -0.0000
                        (0.0001)      (0.0001)
  Adj. [R.sup.2]         0.9246        0.9234
  No. of banks            5,964         5.964
  Observations           92.315        92.315
TARP only
  Bailout expectation    -0.0002       0.0000
  Adj. [R.sup.2]        (0.0003)      (0.0004)
                         0.9170        0.9106
  No. of banks             548           548
  Observations            8.765         8.765
Matched sample
  Bailout expectation   -0.0006 *    -0.0007 **
                        (0.0003)      (0.0003)
  Adj. [R.sup.2]         0.9332        0.9332
  No. of banks             597           597
  Observations            8,654         8,654
  TE                       Yes           Yes
  FE                       Yes           Yes
  YE x SE                  Yes           Yes
  Controls                 Yes           Yes

Notes: Table 8 shows regression results for Equation (2) for
different samples. The first block shows results for the baseline
sample that resembles the solid box on the right side in t = 2 of
Figure 1. The second block excludes failed banks from this sample.
The third block comprises all banks after q4/09, as appear in the
dotted box. The fourth block includes TARP banks only. The fifth
block includes only banks from the baseline sample that are 1:1
matches with the distressed banks in the crisis period, according to
propensity score matching. Each regression includes bank (FE),
quarter (TE), and interacted year-state (YE x SE) fixed effects, as
well as all other control variables from the baseline regression for
the period q1/10-q4/13. Variable definitions are in Table 3.
Clustered (bank level) standard errors are in parentheses.

***, **, and * indicate significant coefficients at the 1%, 5%, and
10% levels, respectively.

TABLE 9 Covariate Differences Matched versus Unmatched Samples

                                 Mean
             Unmatched/
Variable      Matched     Distressed     Sound    Bias (%)

L.EQ             U           0.0942      0.1126    -44.10
                 M           0.0959      0.0944      3.70
L.RoA            U          -0.0050      0.0028    -68.70
                 M          -0.0037     -0.0033     -3.40
L.NPA            U           0.0493      0.0309     46.70
                 M           0.0460      0.0458      0.50
L.Size           U          13.06       11.82       93.90
                 M          12.97       12.96        0.50
L.Cash           U           0.0500      0.0617    -22.40
                 M           0.0500      0.0509     -1.60
L.Deposits       U           0.7986      0.8219    -30.70
                 M           0.7998      0.7978      2.70
L.Loans          U           0.7268      0.6582     50.50
                 M           0.7257      0.7328     -5.20
L.CS index       U         344.33      304.85       49.90
                 M         341.31      345.72       -5.60
L.UR             U           0.0816      0.0773     19.90
                 M           0.0815      0.0815     -0.10

                                                      T-test
             Unmatched/
Variable      Matched     Reduction of Bias (%)   T-test    p Value

L.EQ             U                                 -9.40    0.0000
                 M                91.50             0.81    0.4200
L.RoA            U                                -20.46    0.0000
                 M                95.00            -0.59    0.5580
L.NPA            U                                 13.78    0.0000
                 M                98.90             0.08    0.9360
L.Size           U                                 25.32    0.0000
                 M                99.50             0.08    0.9370
L.Cash           U                                 -5.05    0.0000
                 M                93.00            -0.33    0.7380
L.Deposits       U                                 -7.51    0.0000
                 M                91.40             0.39    0.6940
L.Loans          U                                 11.27    0.0000
                 M                89.60            -1.08    0.2810
L.CS index       U                                 12.43    0.0000
                 M                88.80            -0.89    0.3750
L.UR             U                                  4.55    0.0000
                 M                99.40            -0.02    0.9810

Notes: Table 9 shows the outcome of a 1:1 propensity score matching
between sound and distressed banks in the crisis period, including
the mean for each variable for the treated and control group and for
the sample of matched (M) and unmatched (U) banks. It further shows
the reduction in bias for each variable between the groups and
significant differences in means before and after matching.

TABLE 10
Bailout Expectation Effects on Interest Rates Over Time

                                       Interest Rates R

                             Total Loans   Real Estate   C&I Loans
Dependent Variable               (1)        Loans (2)       (3)

Bailout expectation (2010)    0.0037 **     0.0039 **    0.0058 **
                              (0.0016)      (0.0019)     (0.0028)
Bailout expectation (2011)    0.0019 **      0.0016       0.0046
                              (0.0009)      (0.0013)     (0.0029)
Bailout expectation (2012)     0.0012        0.0017       0.0016
                              (0.0010)      (0.0013)     (0.0031)
Bailout expectation (2013)     -0.0004       -0.0009      0.0008
                              (0.0012)      (0.0014)     (0.0035)
L.EQ                          0.0159 **      0.0140       0.0357
                              (0.0071)      (0.0097)     (0.0227)
L.RoA                         0.0253 **      0.0030       0.0366
                              (0.0126)      (0.0196)     (0.0338)
L.NPA                        -0.0302 ***   -0.0364 ***    0.0022
                              (0.0076)      (0.0076)     (0.0226)
L.Size                         0.0032        0.0037       0.0045
                              (0.0027)      (0.0027)     (0.0039)
L.Cash                         0.0087        0.0042       0.0065
                              (0.0063)      (0.0061)     (0.0110)
L.Deposits                   0.0120 ***     0.0101 **     0.0019
                              (0.0045)      (0.0049)     (0.0132)
L.Loans                      -0.0074 ***   -0.0069 ***    0.0000
                              (0.0023)      (0.0024)     (0.0069)
L.CS index                     -0.0000       -0.0000      0.0000
                              (0.0000)      (0.0000)     (0.0001)
L.UR                           0.0052        0.0066       -0.0026
                              (0.0053)      (0.0068)     (0.0170)
Constant                       0.0118        0.0006       -0.0004
                              (0.0384)      (0.0384)     (0.0568)
TE                               Yes           Yes          Yes
FE                               Yes           Yes          Yes
YE x SE                          Yes           Yes          Yes
No. of Banks                     597           597          597
Observations                    8,654         8,654        8,654
Adj. [R.sup.2]                  0.85          0.70         0.62

                                   Interest Rates R

                              Deposits     Total Funding
Dependent Variable               (4)            (5)

Bailout expectation (2010)   -0.0011 ***    -0.0012 ***
                              (0.0003)       (0.0003)
Bailout expectation (2011)     -0.0004        -0.0005
                              (0.0003)       (0.0003)
Bailout expectation (2012)     0.0001         -0.0000
                              (0.0004)       (0.0004)
Bailout expectation (2013)     0.0001         -0.0001
                              (0.0005)       (0.0005)
L.EQ                           0.0010         -0.0060
                              (0.0032)       (0.0045)
L.RoA                        -0.0085 **     -0.0074 **
                              (0.0037)       (0.0034)
L.NPA                         -0.0059 *     -0.0072 **
                              (0.0031)       (0.0031)
L.Size                        0.0012 **      0.0011 **
                              (0.0005)       (0.0005)
L.Cash                        -0.0023 *       -0.0020
                              (0.0012)       (0.0012)
L.Deposits                    0.0047 **       -0.0025
                              (0.0021)       (0.0037)
L.Loans                        0.0004         -0.0001
                              (0.0011)       (0.0012)
L.CS index                    0.0000 **      0.0000 **
                              (0.0000)       (0.0000)
L.UR                           0.0031         0.0027
                              (0.0022)       (0.0023)
Constant                       -0.0060        0.0033
                              (0.0070)       (0.0077)
TE                               Yes            Yes
FE                               Yes            Yes
YE x SE                          Yes            Yes
No. of Banks                     597            597
Observations                    8,654          8,654
Adj. [R.sup.2]                  0.93           0.93

NOTES: Table 10 shows regression results for Equation (2), in which
bailout expectations are interacted with year dummies for 2010, 2011,
2012, and 2013. Each regression includes bank (FE), quarter (TE), and
interacted year-state (YE x SE) fixed effects for the period
q1/10-q4/12. The prefix "L" indicates that a variable is lagged by
one-quarter. Variable definitions are in Table 3. Clustered (bank
level) standard errors are in parentheses.

***, **, and * indicate significant coefficients at the 1%, 5%, and
10% levels, respectively.

TABLE 11
Loan and Deposit Growth

                                 Loan and Deposit Growth

                      [DELTA] Total Loans   [DELTA] Real Estate Loans
Dependent Variable            (1)                      (2)

Bailout expectation         -0.0010                  -0.0089
                           (0.0075)                 (0.0086)
L.EQ                      0.7346 ***               0.7026 ***
                           (0.1521)                 (0.1644)
L.RoA                     -0.5042 ***              -0.3610 **
                           (0.1701)                 (0.1642)
L.NPA                     -0.4375 ***              -0.4754 ***
                           (0.0570)                 (0.0626)
L.Size                    -0.0444 ***              -0.0409 ***
                           (0.0148)                 (0.0145)
L.Cash                      0.0150                   -0.0119
                           (0.0363)                 (0.0412)
L.Deposits                 -0.0557 *                 -0.0378
                           (0.0332)                 (0.0383)
L.Loans                   -0.1599 ***              -0.1323 ***
                           (0.0300)                 (0.0313)
L.CS index                  0.0002                  0.0003 *
                           (0.0001)                 (0.0002)
L.UR                        -0.0286                  0.1375
                           (0.0733)                 (0.1179)
Constant                  0.5783 ***                0.4693 **
                           (0.2003)                 (0.2089)
TE                            Yes                      Yes
FE                            Yes                      Yes
YE x SE                       Yes                      Yes
No. of Banks                  597                      597
Observations                 8,654                    8,654
Adj. [R.sup.2]               0.27                     0.26

                            Loan and Deposit Growth

                      [DELTA] C&I Loans   [DELTA] Deposits
Dependent Variable           (3)                (4)

Bailout expectation        0.0051             -0.0114
                          (0.0168)            (0.0078)
L.EQ                     0.8178 ***          0.6364 ***
                          (0.1478)            (0.1440)
L.RoA                      -0.2675           -0.3460 *
                          (0.3074)            (0.1897)
L.NPA                    -0.3885 ***        -0.1941 ***
                          (0.1352)            (0.0649)
L.Size                   -0.0590 **         -0.0972 ***
                          (0.0251)            (0.0106)
L.Cash                     0.0539            -0.1245 **
                          (0.0712)            (0.0574)
L.Deposits                 -0.0738          -0.4455 ***
                          (0.0757)            (0.0548)
L.Loans                  -0.1914 ***         0.2264 ***
                          (0.0551)            (0.0423)
L.CS index                 -0.0001             0.0000
                          (0.0004)            (0.0001)
L.UR                      -0.3583 *           -0.1055
                          (0.2046)            (0.0768)
Constant                 1.0096 ***          1.4506 ***
                          (0.3803)            (0.1675)
TE                           Yes                Yes
FE                           Yes                Yes
YE x SE                      Yes                Yes
No. of Banks                 597                597
Observations                8,653              8,654
Adj. [R.sup.2]              0.07                0.22

Notes: Table 11 shows regression results for Equation (2) and uses
quarterly growth rates of total loans, real estate loans, commercial
and industrial loans, and deposits as dependent variables. Each
regression includes bank (FE), quarter (TE), and interacted
year-state (YE x SE) fixed effects for the period q1/10-q4/12. The
prefix "L" indicates that a variable is lagged by one-quarter.
Variable definitions are in Table 3. Clustered (bank level) standard
errors are in parentheses.

***, **, and * indicate significant coefficients at the 1%, 5%, and
10% levels, respectively.

TABLE 12
Loan and Deposit Growth Over Time

                                   Loan and Deposit Growth

                             [DELTA] Total   [DELTA] Real Estate
Dependent Variable             Loans (1)          Loans (2)

Bailout expectation (2010)      -0.0033            -0.0092
                               (0.0078)           (0.0089)
Bailout expectation (2011)      -0.0052            -0.0133
                               (0.0079)           (0.0091)
Bailout expectation (2012)      0.0028             -0.0116
                               (0.0106)           (0.0112)
Bailout expectation (2013)      0.0132             0.0071
                               (0.0135)           (0.0150)
L.EQ                          0.7159 ***         0.6875 ***
                               (0.1640)           (0.1758)
L.RoA                         -0.4853 ***        -0.3448 **
                               (0.1704)           (0.1657)
L.NPA                         -0.4317 ***        -0.4723 ***
                               (0.0574)           (0.0639)
L.Size                        -0.0486 ***        -0.0443 ***
                               (0.0168)           (0.0163)
L.Cash                          0.0195             -0.0082
                               (0.0361)           (0.0409)
1..Deposits                    -0.0593 *           -0.0412
                               (0.0334)           (0.0383)
L.Loans                       -0.1620 ***        -0.1343 ***
                               (0.0300)           (0.0311)
L.CS index                      0.0002            0.0003 *
                               (0.0001)           (0.0002)
L.UR                            -0.0302            0.1383
                               (0.0732)           (0.1178)
Constant                      0.6412 ***          0.5206 **
                               (0.2279)           (0.2318)
TE                                Yes                Yes
FE                                Yes                Yes
YE x SE                           Yes                Yes
No. of Banks                      597                597
Observations                     8,654              8,654
Adj. [R.sup.2]                   0.27               0.26

                                   Loan and Deposit Growth

                             [DELTA] C&I Loans   [DELTA] Deposits
Dependent Variable                  (3)                (4)

Bailout expectation (2010)        -0.0001           -0.0167 *
                                 (0.0185)            (0.0089)
Bailout expectation (2011)        -0.0050            -0.0090
                                 (0.0181)            (0.0092)
Bailout expectation (2012)        0.0177             -0.0086
                                 (0.0232)            (0.0083)
Bailout expectation (2013)        0.0328             -0.0000
                                 (0.0242)            (0.0117)
L.EQ                            0.7773 ***          0.6202 ***
                                 (0.1602)            (0.1493)
L.RoA                             -0.2270           -0.3311 *
                                 (0.3042)            (0.1898)
L.NPA                           -0.3762 ***        -0.1843 ***
                                 (0.1358)            (0.0666)
L.Size                          -0.0680 **         -0.1009 ***
                                 (0.0278)            (0.0109)
L.Cash                            0.0637            -0.1210 **
                                 (0.0715)            (0.0578)
1..Deposits                       -0.0814          -0.4488 ***
                                 (0.0758)            (0.0548)
L.Loans                         -0.1955 ***         0.2246 ***
                                 (0.0548)            (0.0425)
L.CS index                        -0.0001             0.0000
                                 (0.0004)            (0.0001)
L.UR                             -0.3632 *           -0.1079
                                 (0.2045)            (0.0767)
Constant                        1.1586 ***          1.5097 ***
                                 (0.4104)            (0.1745)
TE                                  Yes                Yes
FE                                  Yes                Yes
YE x SE                             Yes                Yes
No. of Banks                        597                597
Observations                       8,653              8,654
Adj. [R.sup.2]                     0.07                0.22

Notes: Table 12 shows regression results for Equation (2) and uses
quarterly growth rates of total loans, real estate loans, commercial
and industrial loans, and deposits as dependent variables. Bailout
expectations are interacted with year dummies for 2010, 2011, 2012,
and 2013. Each regression includes bank (FE), quarter (TE), and
interacted year--state (YE x SE) fixed effects for the period
q1/10-q4/12. The prefix "L" indicates that a variable is lagged by
one-quarter. Variable definitions are in Table 3. Clustered (bank
level) standard errors are in parentheses.

***, **, and * indicate significant coefficients at the 1%, 5%, and
10% levels, respectively.

TABLE 13
Bailout Expectation Effects on Lending and Funding Rates
Stratified by Equity

                                    Interest Rates R

                             Total Loans   Real Estate Loans
Dependent Variable               (1)              (2)

Bailout expectation            0.0015           0.0013
                              (0.0012)         (0.0015)
CAPH                           -0.0002          -0.0008
                              (0.0006)         (0.0008)
CAPH x Bailout expectation    0.0016 **        0.0021 *
                              (0.0008)         (0.0011)
L.RoA                         0.0286 **         0.0072
                              (0.0135)         (0.0206)
L.NPA                        -0.0276 ***      -0.0330 ***
                              (0.0072)         (0.0073)
L.Size                         0.0020           0.0025
                              (0.0024)         (0.0025)
L.Cash                         0.0101           0.0056
                              (0.0068)         (0.0064)
L.Deposits                    0.0077 **         0.0063
                              (0.0030)         (0.0049)
L.Loans                      -0.0081 ***      -0.0077 ***
                              (0.0022)         (0.0024)
L.CS index                     -0.0000          -0.0000
                              (0.0000)         (0.0000)
L.UR                           0.0043           0.0056
                              (0.0053)         (0.0069)
Constant                       0.0328           0.0220
                              (0.0320)         (0.0368)
TE                               Yes              Yes
FE                               Yes              Yes
YE x SE                          Yes              Yes
No. of Banks                     597              597
Observations                    8,654            8,654
Adj. [R.sup.2]                  0.85             0.69

                                        Interest Rates R

                             C&I Loans    Deposits     Total Funding
Dependent Variable              (3)          (4)            (5)

Bailout expectation           0.0044       -0.0001        -0.0004
                             (0.0037)     (0.0004)       (0.0004)
CAPH                          0.0008      0.0009 **      0.0007 *
                             (0.0029)     (0.0004)       (0.0004)
CAPH x Bailout expectation    0.0006     -0.0010 **     -0.0011 **
                             (0.0036)     (0.0004)       (0.0005)
L.RoA                         0.0395     -0.0106 ***    -0.0087 ***
                             (0.0330)     (0.0037)       (0.0033)
L.NPA                         0.0050     -0.0072 **     -0.0086 ***
                             (0.0213)     (0.0030)       (0.0030)
L.Size                        0.0024     0.0014 ***     0.0014 ***
                             (0.0032)     (0.0005)       (0.0005)
L.Cash                        0.0091     -0.0025 **     -0.0024 **
                             (0.0117)     (0.0012)       (0.0012)
L.Deposits                    -0.0079     0.0044 **       -0.0007
                             (0.0100)     (0.0019)       (0.0027)
L.Loans                       -0.0010      0.0004         0.0001
                             (0.0069)     (0.0011)       (0.0012)
L.CS index                    0.0000      0.0000 **      0.0000 **
                             (0.0001)     (0.0000)       (0.0000)
L.UR                          -0.0041      0.0035         0.0033
                             (0.0169)     (0.0022)       (0.0023)
Constant                      0.0355       -0.0087        -0.0032
                             (0.0434)     (0.0064)       (0.0065)
TE                              Yes          Yes            Yes
FE                              Yes          Yes            Yes
YE x SE                         Yes          Yes
No. of Banks                    597          597            597
Observations                   8,654        8,654          8,654
Adj. [R.sup.2]                 0.62         0.93           0.93

Notes: Table 13 shows regression results for our baseline
regression. In each regression, we interact Bailout expectation with
CAPH. CAPH indicates banks with a total equity ratio of more than 7%.
Each regression includes bank (FE), quarter (TE), and interacted
year-state (YE x SE) fixed effects for the period q1/10-q4/12. The
prefix "L" indicates that a variable is lagged by one-quarter.
Clustered (bank level) standard errors are in parentheses.

***, **, and * indicate significant coefficients at the 1%, 5%, and
10% levels, respectively.

TABLE 14
Bailout Expectation Effects on Lending and Funding Rates Stratified
by Size

                                   Interest Rates R

                                        Real Estate
                          Total Loans      Loans       C&I Loans
Dependent Variable            (1)           (2)           (3)

Bailout expectation        0.0025 *       0.0023       0.0059 **
                           (0.0014)      (0.0017)      (0.0028)
Size (medium)               0.0001        -0.0012       0.0042
                           (0.0024)      (0.0022)      (0.0054)
Size (large)               0.0051 *       0.0032       0.0131 *
                           (0.0031)      (0.0029)      (0.0070)
Size (medium) x Bailout     0.0013        0.0034        -0.0057
  expectation              (0.0024)      (0.0022)      (0.0055)
Size (large) X Bailout      -0.0016       -0.0005     -0.0137 ***
  expectation              (0.0021)      (0.0019)      (0.0049)
L.EQ                        0.0084        0.0054        0.0244
                           (0.0076)      (0.0131)      (0.0206)
L.RoA                      0.0297 **      0.0076        0.0440
                           (0.0146)      (0.0209)      (0.0349)
L.NPA                     -0.0282 ***   -0.0336 ***     0.0030
                           (0.0073)      (0.0074)      (0.0219)
L.Cash                      0.0105        0.0063        0.0075
                           (0.0074)      (0.0072)      (0.0117)
L.Deposits                0.0106 ***     0.0082 *       0.0003
                           (0.0038)      (0.0044)      (0.0132)
L.Loans                   -0.0073 ***   -0.0066 ***     -0.0002
                           (0.0025)      (0.0025)      (0.0071)
L.CS index                  -0.0000       -0.0000       0.0000
                           (0.0000)      (0.0000)      (0.0001)
L.UR                        0.0051        0.0066        -0.0022
                           (0.0054)      (0.0069)      (0.0169)
Constant                  0.0547 ***    0.0501 ***     0.0597 **
                           (0.0075)      (0.0080)      (0.0283)
TE                            Yes           Yes           Yes
FE                            Yes           Yes           Yes
YE x SE                       Yes           Yes           Yes
No. of Banks                  597           597           597
Observations                 8,654         8,654         8,654
Adj. [R.sup.2]               0.84          0.69          0.62

                               Interest Rates R

                           Deposits     Total Funding
Dependent Variable            (4)            (5)

Bailout expectation         -0.0005       -0.0006 *
                           (0.0003)       (0.0003)
Size (medium)               -0.0003        -0.0005
                           (0.0005)       (0.0005)
Size (large)                0.0003         0.0004
                           (0.0010)       (0.0011)
Size (medium) x Bailout     -0.0001        0.0001
  expectation              (0.0005)       (0.0005)
Size (large) X Bailout     0.0009 **       0.0009
  expectation              (0.0004)       (0.0007)
L.EQ                        -0.0001        -0.0069
                           (0.0031)       (0.0044)
L.RoA                     -0.0098 ***    -0.0087 **
                           (0.0038)       (0.0034)
L.NPA                     -0.0066 **     -0.0080 **
                           (0.0031)       (0.0031)
L.Cash                      -0.0020        -0.0017
                           (0.0012)       (0.0012)
L.Deposits                 0.0044 **       -0.0027
                           (0.0022)       (0.0038)
L.Loans                     0.0009         0.0005
                           (0.0011)       (0.0012)
L.CS index                 0.0000 **      0.0000 **
                           (0.0000)       (0.0000)
L.UR                       0.0037 *        0.0034
                           (0.0022)       (0.0023)
Constant                  0.0084 ***     0.0163 ***
                           (0.0025)       (0.0039)
TE                            Yes            Yes
FE                            Yes            Yes
YE x SE                       Yes            Yes
No. of Banks                  597            597
Observations                 8,654          8,654
Adj. [R.sup.2]               0.93           0.93

Notes: Table 14 shows regression results for our baseline regression.
In each regression, we interact Bailout expectation with dummy
variables for size indicating small, medium, and large banks. Small
banks have total asset of less than 1$ billion while large banks have
more than $3 billion of total assets. Medium banks have more than $1
billion but less than $3 billion of total assets. Each regression
includes bank (FE), quarter (TE), and interacted year--state (YE x
SE) fixed effects for the period q1/10-q4/12. The prefix "L"
indicates that a variable is lagged by one-quarter. Clustered (bank
level) standard errors are in parentheses.

***, **, and * indicate significant coefficients at the 1%, 5%, and
10% levels, respectively.

TABLE 15
Branching Restrictions

                                  Interest Rates R

                                       Real Estate
                         Total Loans      Loans      C&I Loans
Dependent Variable           (1)           (2)          (3)

Bailout expectation       0.0034 *       0.0037        0.0022
  (restriction low)       (0.0020)      (0.0026)      (0.0049)

Bailout expectation       0.0040 **    0.0051 ***    0.0112 ***
  (restriction medium)    (0.0017)      (0.0019)      (0.0038)
Bailout expectation       0.0059 **     0.0060 **      0.0064
  (restriction high)      (0.0029)      (0.0029)      (0.0041)
L.EQ                      0.0328 **    0.0362 ***    0.0658 **
                          (0.0134)      (0.0096)      (0.0271)
L.RoA                     0.0360 *       0.0025       -0.0014
                          (0.0184)      (0.0228)      (0.0367)
L.NPA                     -0.0154 *     -0.0176 *      0.0034
                          (0.0092)      (0.0106)      (0.0253)
L.Size                     0.0040        0.0036       0.0098 *
                          (0.0028)      (0.0026)      (0.0054)
L.Cash                    0.0106 *       0.0080        0.0087
                          (0.0061)      (0.0062)      (0.0115)
L.Deposits                 -0.0018       0.0074       -0.0051
                          (0.0103)      (0.0056)      (0.0126)
L.Loans                  -0.0080 **      -0.0052       0.0022
                          (0.0032)      (0.0033)      (0.0069)
L.CS index                 -0.0000       -0.0000      -0.0000
                          (0.0000)      (0.0000)      (0.0001)
L.UR                       0.0002        -0.0065      -0.0167
                          (0.0079)      (0.0095)      (0.0233)
Constant                   0.0127        0.0124       -0.0643
                          (0.0398)      (0.0358)      (0.0697)
TE                           Yes           Yes          Yes
FE                           Yes           Yes          Yes
YE x SE                      Yes           Yes          Yes
No. of Banks                 597           597          597
Observations               24,097        24,097        24,097
Adj. [R.sup.2]              0.91          0.84          0.80

                               Interest Rates R

                          Deposits     Total Funding
Dependent Variable           (4)            (5)

Bailout expectation      -0.0010 **     -0.0013 ***
  (restriction low)       (0.0004)       (0.0005)

Bailout expectation        -0.0003        -0.0004
  (restriction medium)    (0.0005)       (0.0005)
Bailout expectation        0.0001         -0.0001
  (restriction high)      (0.0009)       (0.0009)
L.EQ                       0.0039         -0.0030
                          (0.0033)       (0.0044)
L.RoA                      -0.0061        -0.0040
                          (0.0062)       (0.0060)
L.NPA                    -0.0099 ***    -0.0106 ***
                          (0.0036)       (0.0036)
L.Size                   0.0022 ***     0.0021 ***
                          (0.0005)       (0.0005)
L.Cash                   -0.0033 **     -0.0033 **
                          (0.0014)       (0.0015)
L.Deposits                0.0052 **       -0.0017
                          (0.0020)       (0.0035)
L.Loans                    0.0014         0.0008
                          (0.0012)       (0.0013)
L.CS index                0.0000 **      0.0000 *
                          (0.0000)       (0.0000)
L.UR                       0.0008         0.0014
                          (0.0034)       (0.0038)
Constant                 -0.0241 ***     -0.0150 *
                          (0.0073)       (0.0078)
TE                           Yes            Yes
FE                           Yes            Yes
YE x SE                      Yes            Yes
No. of Banks                 597            597
Observations               24.097         24,097
Adj. [R.sup.2]              0.96           0.96

Notes: Table 15 shows regression results for Equation (2) while
interacting bailout expectation with dummies that reflect whether a
banks resides in a state with low, medium, or high restrictions on
banking and branching, according to (Rice and Strahan 2010). Each
regression includes bank (FE), quarter (TE), and interacted
year--state (YE x SE) fixed effects for the period q1/10-q4/12. The
prefix "L" indicates that a variable is lagged by one-quarter.
Variable definitions are in Table 3. Clustered (bank level) standard
errors are in parentheses.

***, **, and * indicate significant coefficients at the 1%, 5%, and
10% levels, respectively.