Equity Capital as a Safety Cushion in the US Banking Sector

The incidence of US bank failures soared in the financial crisis and economic recession starting in 2008. Financial regulations promulgated by the Federal Reserve and issued through the Basel III Accord raised the minimum equity capital requirements of banks. The intent of the increase in equity capital was to serve as a greater safety cushion to reduce the probability of failure. The purpose of this study is to examine the financial statement variables that distinguish failed (zero equity capital) and nonfailed US banks. The methods employed to investigate our research question are: 1. univariate t-test, and 2. tobit regression analysis with equity capital as the dependent variable. Our results show that the factors explaining equity capital include real estate loans to assets, equity capital to total assets, log of total assets, return on equity, loan loss allowance to total loans, non-performing loans to total assets, total loans to total assets, mortgage-backed securities to total assets, total short-term debt securities to total assets, net gains on sales of loans to total non-interest income, and insured deposits to total deposits. Bank management and financial regulators need to focus on these financial characteristics to ensure adequate equity capital as a safety cushion.


Introduction
During the financial crisis and economic recession of 2008 to 2010 financial institution failures soared in the US especially in the banking sector.Investors, analysts and regulators scrutinized bank's financial statements in search of the underlying factors leading to bankruptcy.The financial characteristics examined included the asset mix (lending), earnings profile (interest and fees income, expense composition), liquidity, market risk susceptibility, and the capacity of equity capital to act as a safety cushion absorbing the operating loss shocks.
Governments and financial regulators are compelled to respond to the rise in bank failures and downturn in the economy.Actions taken to combat this financial storm, by the Federal Reserve, included lowering short-term interest rates, increasing loans to banks, expanding the list of collateral eligible to secure loans, and bailing out related financial institutions such as AIG who insured much of the credit default swap market.The federal government responded by reducing corporate income tax rates, adding refunds to individuals, increasing spending and changing legislation to make house foreclosures more difficult resulting in a greater likelihood of refinancing.
Further, financial regulation occurred at the international level, in particular, the Basel III Accord with respect to equity capital on the bank balance sheet.The minimum common equity tier 1 (CET1) to risk-weighted assets (RWA) ratio is 6 percent and 7 percent as of 2015 and 2019 respectively.A supplementary equity capital amount of as much as 2.5 percent can be required during periods of high growth.In conjunction with the international equity capital standards the Federal Reserve mandated a minimum financial leverage ratio (Tier 1 Capital to Total Assets) for US banks of 5 percent for holding companies and 8 percent for systemically important financial institutions (SIFI).In 2016 the eight US SIFIs are Bank of America, Bank of New York Mellon, Citigroup, Goldman Sachs, JP Morgan Chase, Morgan Stanley, State Street and Wells Fargo.banks, and by the Office of Thrift Supervision (dissolved in 2011) for savings associations having a federal government charter.The Federal Deposit Insurance Corporation (FDIC) can decide to close a state-chartered bank without the approval of the State Government Agency.The FDIC typically is appointed the receiver for the closed bank and can choose to conduct a deposit payoff or purchase and assumption.
In this study, first, we investigate the financial statement variables that distinguish failed and nonfailed US banks using a univariate t-test.Second, tobit regression analysis is shown to explain the financial characteristics associated with the amount of equity capital during the financial crisis of 2008 to 2010.Third, some suggestions are made for management on how to operate the bank to augment its equity capital and thereby strengthen its safety cushion to face economic and financial market downturns.
The paper is comprised as follows.Section 2 reviews the literature.Section 3 outlines the data, sample, and hypothesis.Section 4 presents the methodology.Section 5 details and discusses the empirical results.Finally, section 6 concludes the study.Sinkey (1974Sinkey ( , 1975) ) researched problem and non-problem banks finding that the growth in equity capital was not commensurate with the asset growth rate.Hutchison and Cox (2007) demonstrated a positive relation between financial leverage and the return on equity and return on assets.Brunnermeier and Pedersen (2009) as well as Shleifer and Vishny (2010) found that in economic downturns high financial leverage banks must liquidate their loans at a loss reducing their equity capital to the point of bank failure.James (1991) found the losses associated with the sale of closed bank assets to be 40 percent of book value.Acharya et al. (2010) showed that restricted debt capacity, partially caused by low equity capital, further increased the probability of bank failure.Wagner (2007) discovered banks that sell their loans also have a higher risk asset portfolio leading to instability.Moreover, Uzun and Web (2007) presented results that banks who securitize assets are larger and inversely related to the degree of equity capital.

Literature Review
Early warning systems of problem banks have been studied by Gonzalez-Hermosillo (1999), Cihak and Schaeck (2010), and Cole and White (2012), using the CAMELS approach, finding inadequate equity capital was a predictor of failure.CAMELS is the acronym for capital adequacy, asset quality, management quality, earnings, liquidity and sensitivity to the market.Cox and Wang (2014), utilizing discriminant analysis, discovered low equity capital as a factor in US bank failures in the 2008 to 2010 financial crisis.Mare (2015) discovered the contribution of macroeconomic factors to the forecasting of small Italian bank failures, leading to the notion that capital requirements should consider the stage of the business cycle in a countercyclical fashion.Ho et al. (2016) presented evidence that overconfident chief executive officers were more likely to increase the debt ratio prior to a crisis culminating in higher failure rates.

Data, Sample, and Hypothesis
Financial statement data for the variables in the models come from the Federal Deposit Insurance Corporation.House price index information (hpindexsa) comes from the Federal Housing Finance Agency and percentage change in personal income (pigrow) comes from the Bureau of Economic Analysis.We access the Bank Data and Statistics under Industry Analysis data assembled by the FDIC from the call reports of US banks for the 2005 to 2010 period.We gather information to calculate 29 independent variables.
The explanatory variables and the predicted relation between them and the dependent variable of bank equity capital is provided in Table 1.We examine five models explaining bank equity capital.The five models delineate different financial characteristic combinations explaining equity capital.The rationale for the different models revolve around the asset mix (loan type), growth of loans and quality of loans.
Book common equity is used as a proxy for market equity.When the common equity of a bank decreases to such an extent that it is negative or zero the bank is closed.There are other banks with very low equity capital that are closed by the respective regulator.In these cases the equity value is worthless.All surviving banks continue to have a positive equity capital balance.

Methodology
The first methodology is comparing banks that had a positive amount of capital (common equity>0) to the banks that had zero equity capital.A univariate t-test for mean differences for each of the 29 independent variables listed in Table 1 is conducted.
The second methodology is the use of tobit regression analysis.Tobit regression was created by Tobin (1958).
The suitability of tobit rests with the empirics of having a dependent variable with a limiting value typically zero.The limited value is the censored bound versus the upside of having an unlimited value called the uncensored value.In Tobit failed banks that are closed are censored.The efficacy of tobit, as opposed to ordinary least squares (OLS), regression, has been examined by McDonald and Moffitt (1980), Foster and Kalenkoski (2013), and Stewart (2013) among others.
There are five tobit regression equations representing five hypothesized models to explain the financial characteristics of banks with an equity capital amount.

Results
The results for the univariate t-tests are reported in Table 2 for 2007 Quarter 4 and Table 3 for 2008 Quarter 4. Clearly surviving banks have a significantly higher quantity of capital and tier 1 equity than banks that failed.The highly significant (alpha ≤ 0.01 for each of the 2 years) variables with failed banks having a higher value than surviving banks are realloan, cons_devlp, mul_family, chargeoff, lossallow, pastdue, foreclose, size, brokdep, interbank, and loan_ast.The highly significant variables with a lower value for failed banks versus surviving banks are sig_family, idloan, loangrowth, capital, tier 1, roa, sec_asset, debt_sec, non_income, and cash.This is in line with our a priori expectations with the exception of mul_family, loanast, interbank, realloan and size.Following previous research we believed that high exposure to residential real estate loans, higher percentage of assets in loans, higher percentage of interbank loans, and larger banks in terms of total assets would be associated with higher equity levels and increased odds of survival, but during the crisis which began in 2008 these associations were reversed.We reported the mean of explanatory variables for surviving and failed banks in the first two columns.The standard deviations are in the parenthesis.We also present the difference in mean and the t-statistic in the third column which tests the mean difference of both sample banks.*, ** and *** significant at the 10%, 5% and 1% level, variables are described in Table 1.
The results for each of the tobit models 1 through 5, excluding model 4, are in Appendix Tables A1 through A4 respectively.Results for model 4, discussed here, are given in Table 4. Model 4 appears to be the superior model as each and every variable is significant with an alpha level of at least five percent when using data from the first quarter of 2005.The likelihood ratio (LR) chi-square is very high peaking at 1102.55 in 2007 based on 2005 Quarter 1.The probability >Chi-square is significant at greater than 0.0000 across all time periods.The log likelihood is in the range of -129,197 to -132,924 during the entire period.The pseudo R-square is better than the other four models varying from 0.0035 to 0.0041.The set of factors in model 4 explaining equity capital includes real estate loans to assets, equity capital to total assets, log of total assets, return on equity, loan loss allowance to total loans, non-performing loans to total assets, total loans to total assets, mortgage-backed securities to total assets, total short-term debt securities to total assets, net gains on sales of loans to total non-interest income, and insured deposits to total deposits.The constant (in all of the models) is negative, very large ($3 million and up), and always significant.It is interesting to note that roe, lossallow, pastdue, and loansale all become less significant after 2005 even as the banks come closer to failure.
Similar to the univariate analysis (Tables 2 and 3), the tobit analysis (Tables A1 through A4 and Table 4) show an unexpected negative effect on equity with increased exposure to multi-family real estate loans, total loans to total assets, and real estate loans to total loans.However, tobit analysis shows a very strong and very large positive association between the size of the bank in terms of total assets and the expected equity value of the bank.

Conclusions
This paper studies US banks whose equity capital evaporated resulting in their demise during the financial crisis and economic recession of 2008 to 2010.The univariate t-test method is used to detect mean differences for 29 independent financial variables between censored banks (zero equity capital) and noncensored banks (positive equity capital).The tobit regression analysis indicates that realloan, capital, size, roe, lossallow, pastdue, loan_ast, MBS, debt_sec, loansale, and insureddep are the most significant in determining the amount of equity banks were able to maintain during the crisis.Comparing failed to surviving banks we discover a great disparity in performance.The operations of banks undergoing reductions in equity capital were far different in terms of riskiness of assets, capital structure, liquidity, and profitability.In particular, banks with plummeting equity capital had a loan portfolio tilted towards real estate and construction, higher levels of debt on the balance sheet, lower cash levels, and operating losses.
Managers as well as regulators need to take into consideration the danger that banks can pass into when taking on riskier loans and overexposing their loan portfolio to 1 or 2 industries.The result of such decisions leads to a low quality loan portfolio generating losses that ripple into overall operating losses reducing the amount of the equity capital safety cushion.This situation can lead to bank failure.-4,402,370 -5,181,992 -6,142,342 -6,492,635 (-27.54 -3,145,858 -3,832,873 -4,566,586 -4,334,350 (-12.43 -3,161,363 -3,726,046 -4,507,159 -4,309,906 (-12.60

Table 2 .
Descriptive statistics and univariate t-test for mean differences(2007Q4)

Table 3 .
Descriptive statistics and univariate t-test for mean differences (2008Q4) Note.We obtained the results by using the cross-sectional data of 2008Q4.Failure dummy variable defined as banks that failed in 2009-2010.

Table 4a .
Tobit regression results: Model 4 Panel A (zero equity in 2010)