The role of managerial prudence in bank loan loss provisioning.

The recent market crisis has prompted greater awareness of the fragility of managerial estimates in bank financial statements. In the past, financial statement preparers were directed to employ conservatism in reported figures (Financial Accounting Standards Board, 2004). This practice could be perceived as favoring financial system stability because conservative accruals during good times would build capital buffers and preserve solvency and maintain stable market confidence during times of financial system volatility. Through a joint project with international accounting standard setters, the U.S. Financial Accounting Standards Board (FASB) has recently eliminated the characteristic of conservatism from its accounting framework and adopted an unbiased estimate or neutral approach of accounting (FASB, 2010; Watts, 2003). Watts (2003b) cautioned accounting standard setters to make this change only after obtaining a clear understanding of potential ramifications of removing the element of conservatism. This study investigates the relationship between management prudence (a management characteristic or trait characterized by a focus on the long-term viability of the enterprise) and accounting conservatism (a systematic approach to accounting that leads to a faster recognition of bad vs. good news as reflected in management's published accounting figures). While earnings management and loan loss provisioning has been extensively studied, prior studies have not evaluated the relationship between loss provisioning and characteristics of prudent management.

The characteristic of accounting conservatism, which Basu (1997) defines as early recognition of losses and delayed recognition of gains, has been extensively evaluated with respect to its relationship with negative attributes, such as hiding information from potential investors or regulators, smoothing earnings, or managing capital levels (e.g., LaFond and Watts, 2008; Liu and Ryan, 2006). To distinguish managerial prudence from accounting conservatism, this paper defines prudence as management's focus on the longterm viability of the organization and imprudence as management decisions that have been found to be associated with bank insolvency. To summarize, prudence/imprudence is a measure of how management's current decisions affect the long-term viability of the bank, whereas conservatism is a measure of how quickly the financial statements reflect the impact of management's imprudent actions. This study makes no claims about earnings management and does not investigate any deviation from a "natural" or "normal" level of accounting accruals. Rather, this study investigates if the loan loss provisions of prudent banking organizations appear to be different (either more or less conservative) than those of less prudent banking organizations.

This study attempts to isolate a managerial attribute and evaluate whether that attribute has a relationship with accounting accruals, specifically the loan loss provision. Finding an association would be consistent with the notion that managers use accruals to convey information about the nature of their organization's strategic view or other managerial traits. This study uses the bank setting because the loan loss provision is a well-documented managerial estimate and has been extensively studied (e.g., Kanagaretnam et al., 2004; Liu and Ryan, 2006). While other studies have evaluated accounting conservatism and accounting accruals in banks (e.g., Handorf and Zhu, 2006, Kanagaretnam et al., 2003; Kanagaretnam et al., 2004; Liu and Ryan, 2006), this study adds to the research on bank-specific accruals by investigating the link between accounting conservatism and management decisions.

Due to the lack of prior research defining quantitative attributes of management prudence in banks, this study develops a managerial prudence measure by isolating characteristics that have been associated with risky bank management behaviors and ultimately led to bank failures, according to the Office of the Comptroller of the Currency (OCC), a banking regulator (OCC, 1988), and other studies (e.g., Ahmed et al., 1999; Wheelock and Wilson, 2000). Therefore, an assumption is made that avoiding behaviors associated with bank failure can be used as a proxy for managerial prudence.

An analysis of the sample of commercial banks in this study shows that banks with less prudent management have higher subsequent ratios of quarterly provision expense to average loans than banks with more prudent leadership. This finding is consistent with less prudent management being less conservative in their accounting. Conservatism would require imprudent banks to reflect the impact of their current imprudent actions in the current period's provision expense. The finding that imprudent banks have higher subsequent provision expense is consistent with imprudent banks being less conservative in their accounting and therefore being required to make a larger provision expense in the subsequent period to "catch up." Furthermore, banks with less prudent management have less robust coverage of estimated losses within their loan loss allowance balances, which is consistent with the notion that banks with less prudent management insufficiently provide for identified nondiscretionary and other probable loan losses. These results lead to the conclusion that banks with less prudent management are more likely to need higher provisions to address loan loss estimate shortfalls, which could occur through insufficient provisions or through delays in recognizing loan quality declines.

This paper is organized as follows. The next section describes a bank manager's motivation for recognizing discretionary accounting accruals. Section 3 outlines the characteristics associated with prudent managerial leadership in banks. The research design is described in Section 4. Section 5 provides the study's results, and Section 6 outlines conclusions and suggestions for future work.

ACCOUNTING ACCRUALS IN BANKS

Generally Accepted Accounting Principles require that loans (receivables) be reported at their net realizable value (NRV). NRV is the actual amount of cash that a bank expects to collect on its loan portfolio. NRV is equal to the total amount of loans outstanding less an estimate for the amount of the loans that will not be collected. The estimate of loans that will not be collected is recorded in the allowance for loan and lease losses (ALLL), a balance sheet (contra-asset) account. Each year, amounts are added to the ALLL based on management's estimate of future loan losses. Amounts are added to the ALLL by recording a charge against the bank's earnings, called the provision for loan losses (hereafter provision) which is an expense. When a bank confirms that a loan will not be collected, the uncollectible loans are first charged off against the ALLL (this charge off does not affect income or capital). If the ALLL is depleted, additional credit losses would be absorbed by the bank's capital accounts (Wall and Koch, 2000). In a perfect world the yearly provision would represent expected credit losses due to loan activity during the current year. However, due to management's inability to estimate with perfect foresight, future provisions are often increased or decreased as a "catch up" for past under/over provisioning.

When making accounting accrual decisions, which includes provisioning, rational corporate leadership would balance personal benefits for selecting the most favorable treatment against the risk of admonishment, possibly from declines in the company's stock price, or more directly from reduced remuneration tied to profits (LaFond and Watts, 2008). Earnings management in non-banking organizations is the subject of much research and debate (DuCharme et al., 2004; Givoly et al., 2007; Louis and Robinson, 2005). Studies have attempted to understand whether earnings have been managed and what incentives prompted earnings management (Kothari et al., 2005).

There have been some initial steps to identify and measure the strength of corporate governance attributes and how these characteristics influence accounting accruals, information quality, and future earnings prospects (Larcker et al., 2007). Accruals evaluation techniques, which are similar to those used in Larcker et al. (2007), are used in this study. However, this study specifically targets behaviors that could be associated with imprudent behavior, while the Larcker et al. (2007) study covers a broad spectrum of fourteen characteristics of members from boards of directors.

Typically, a bank's loan portfolio represents about 60% of its total assets and commonly represents a bank's riskiest investment (Basu, 2003). Excessive loan losses or bank leadership's ineffectiveness in managing the risks associated with the loan portfolio can cause bank failure (Beattie et al., 1995; OCC, 1988). When a bank confirms that loans are worthless, the losses are first absorbed by the bank's allowance for loan and lease losses (ALLL) (2) (Wall and Koch, 2000). If the ALLL is depleted, losses would be absorbed by a bank's capital accounts (Wall and Koch, 2000). If a bank is unable to generate sufficient earnings to replenish the loss allowance account and capital accounts, the government may intervene to protect the bank's depositors through the Federal Deposit Insurance Corporation's insurance fund (Wall and Koch, 2000).

A bank's manager must balance conflicting pressures concerning ALLL levels, as various users of reporting information have divergent desires to seek either higher or lower levels of loss estimates (Gray and Clarke, 2004). According to accounting guidance and bank supervisory guidance, bank management is encouraged to use judgment when estimating loan loss (Securities and Exchange Commission, 2001). Bank management's responses to these standards and guidance have prompted some to believe that the level of the ALLL is managed by bankers to smooth earnings, manipulate capital, or signal the stock market (Beattie et al., 1995; Gray and Clarke, 2004; Kanagaretnam et al., 2004). Research has been conducted to determine whether bankers use the ALLL to achieve desired financial outcomes. The results of extant research are inconclusive or divergent. Furthermore, much of the research that evaluates discretionary expenses is designed to understand whether management intentionally deceives the market or manipulates accounting data for personal gain. Research designed to determine whether the extent of management prudence is a predictor of accrual levels is limited, with no recently published studies evaluating this relationship in banks.

MANAGEMENT PRUDENCE IN BANKS

Adam Smith defined prudence as a virtue relating to the proper care of an individual's health, fortune, rank and reputation. Smith noted it is the virtue of the sound banker that commanded "a certain cold esteem" (Smith, 1969/1853: 6). Based on this definition from Smith, this paper defines management prudence as management's focus on the long-term viability of the organization. Therefore, management attributes associated with bank failures are classified as imprudent, and the reciprocal of those attributes is used as a proxy for management's focus on the long-term viability of the bank, prudent. A search of the academic literature has not identified any prior research that has used a managerial prudence construct similar to what is used in this paper. (3) In accounting terms, conservatism is defined as "a prudent reaction to uncertainty to try to ensure that uncertainty and risks inherent in business situations are adequately considered" (FASB, 2007: 38). In practice, accounting conservatism is implemented by using estimates that would tend to understate current levels of income and assets when there is uncertainty regarding future outcomes. Using these definitions, prudence is distinct from the accounting concept of conservatism. Prudence involves the investment and credit decisions that management make, while conservatism refers to how uncertainty relating to the future outcomes of the investment and credit decisions made by the manager are reflected in the current financial statements. While the concepts are distinct, it is expected that managers, who make decisions that are less prudent, are also likely to be less conservative in their application of accounting principles.

Accounting conservatism research has traditionally focused on measuring the timing of accruals (Givoly et al., 2007). In 1997, Basu advanced an interpretation of accounting conservatism that has been validated in more recent studies (Roychowdhury and Watts, 2007). Basu (1997) interpreted accounting conservatism as an expedient response to negative information and a slow response to positive information (e.g., expenses and losses are recognized sooner than are revenues and gains). A similar definition of accounting conservatism is the tendency to make accounting decisions that will result in the lowest book value of equity (Roychowdhury and Watts, 2007). (4)

Results of studies evaluating management attributes and accounting conservatism have come to conflicting conclusions. For example, Larcker et al., (2007) concluded that none of the governance characteristics, which they evaluated, have significant relationships with an accounting conservatism indicator, while Ahmed and Duellman (2007) concluded that board of director members rely upon accounting conservatism to modify decision-making.

Bank-specific studies that have evaluated motivation and accounting conservatism have drawn conflicting conclusions (Kanagaretnam et al., 2004), with inconsistent results. While some studies conclude that earnings management is evident (Handorf and Zhu, 2006; Liu and Ryan, 2006), others conclude that management does not manage earnings through accrual manipulation (Ahmed et al., 1999; Beatty et al., 1995). These prior studies do not evaluate characteristics of managerial prudence and accruals. To address this apparent research gap, this study isolates attributes of prudent bank management, and evaluates the relationship between the prudence indicator and bank accounting accruals (specifically, the provision).

Design of the Prudence Indicator

Much of the research on bank failures is conducted on data from the period immediately following the late 1980s and early 1990s, because many banks failed during that period. Bank failures increased during this time, ill part, because of weakened economic conditions. In particular, geographic areas were affected by declines in regionally influential sectors, such as the agriculture or oil and gas industries (Wheelock and Wilson, 2000). However, many banks survived these influences, which supports a presumption that attributes besides a bad economy are the ultimate cause of bank failures. While it is likely that poor management decisions are ultimately the reason for bank failures (OCC, 1988), management characteristics are difficult to quantify (Wheelock and Wilson, 2000). Therefore, studies that evaluate bank failures include variables that are more readily quantifiable, such as lending to risky borrowers, proportions of types of investment holdings, loss provisioning tendencies, and capital levels (Wheelock and Wilson, 2000). Five quantifiable attributes related to how a bank is managed are used to construct a measure of prudence. The five characteristics selected as components for this study's Prudence Indicator (PI) are discussed in the following sections.

Loan Growth Factor

A significant and common characteristic of banks that failed during the 1980s is an aggressive lending philosophy (OCC, 1988). This factor is also identified by Hanc (1997) as a feature that differentiated troubled banks that failed from those that did not fail during the 1980s and early 1990s. Additionally, Ahmed et al. (1999) conclude that banks with high loan growth faced greater pressure to address lower capital ratios and managed their capital positions through discretionary, loan loss provisions. Basu (2003) suggests that banks that aggressively compete with other banks for growth may have a tendency to underprice loans and may have a higher risk of failure. Bank supervisors utilize risk measurement models to predict bank insolvency that have traditionally focused on excessive loan growth (King et al., 2006).

Interest and Fees on Loans Factor

Another characteristic that is common to banks that failed in the 1980s and 1990s is a high ratio of interest and fees to loans (Hanc, 1997; Wheelock and Wilson, 2000). High ratios of interest and fees to loans could indicate that a bank was assuming excessively high credit risk. Banks with high ratios of interests and fees on loans may be taking greater credit risks than those banks with low ratios of interest and fees on loans. Oshinsky and Olin (2005) conclude that troubled banks that failed invested in riskier assets than troubled banks that did not fail, as evidenced by significantly higher interest and fee income in troubled banks that failed.

Salary to Employee Ratio Factor

Hanc (1997) concludes that banks with highly-compensated management have a greater tendency to fail. Additionally, Kanagaretnam et al. (2003) conclude that bank managers are influenced by job security when making accounting choices. Therefore, individuals with a greater aversion to loss of compensation may have greater incentive to behave in a more prudent manner. Based on these findings, it is expected that a higher salary to employee ratio indicates that bank management is more willing to assume risks and is therefore less prudent.

High Risk Loans Factor

Banks were more susceptible to failure in the 1980s and 1990s if they had a high proportion of loans to total assets (Hanc, 1997; Thomson, 1991; Wheelock and Wilson, 2000). Banks with high proportion of loans to total assets increased their risk of failure by making high levels of loans to risky business sectors (OCC, 1988). The ratio of commercial and industrial (C&I) loans to total loans is a proxy for credit risk in the loan portfolio. C&I loans are typically less marketable than credit card and mortgage loans, and a high proportion of C&I loans is a characteristic that differentiated failed from surviving banking organizations (Greenawalt and Sinkey, 1988; Wheelock and Wilson, 2000). Additionally, Beatty et al. (2002) conclude that banks record provisions at higher levels when they have greater proportions of commercial loans, which may indicate that commercial loans carry greater credit risk. Furthermore, these types of loans have had higher historic loss rates over the period of this study than other significant types of bank loans, such as real estate loans ("Charge-off and Delinquency Rates," 2005), and have been attributed to increasing the risk of bank failures (Oshinsky and Olin, 2005; Wheelock and Wilson, 2000). Therefore, banks with lower than average C&I loans to total loans ratios are expected to have more prudent management. Additionally, the proportion of C&I loans is one of the variables considered in the Federal Reserve System's bank failure prediction model (King et al., 2006).

Recoveries to Prior Period Loan Losses Factor

Bankers are directed to charge off loans before they are considered totally uncollectible or worthless (Board of Governors of the Federal Reserve System, 2006). When a loan is repaid after it has been charged off, a banker reports a recovery for the amount paid (Federal Financial Institutions Examinations Council, 2006). A higher than average ratio of recoveries to the prior period's losses (R/PPLL ratio) could indicate that management was willing to charge off loans expeditiously and recognize a recovery if the borrower decided to repay the loan. Alternatively, a high ratio could indicate that bank management was expeditiously charging-off loans in response to directives made by banking supervisors. In either case, banks with a high R/PPLL ratio are expected to have more prudent management.

Calculating the Prudence Index

Each bank within the sample receives a prudence index calculated from the five prudence factors described above. For each firm quarter observation, the factors are calculated and each bank's percentile ranking is determined. Each bank's percentile ranking for each factor is aggregated and the mean of the factors determines the prudence indicator (Pi). The equation to determine the Pi is as follows:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.]

where:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.]

A summary of the order of influence for the characteristics within the Pi is in Table 1. An evaluation of the descriptive statistics of the Pi factors indicates that no single factor appears to overwhelmingly influence the Pi in the aggregate nor when grouping the prudent or imprudent observations into subsamples. Additionally, no one factor was inconsequential in the aggregate or in the subsamples.

Once the Pi is determined, a dummy variable is assigned for managerial prudence for each firm quarter observation. The dummy variable, called the Prudence Indicator (PI), is 1 if Pi [greater than or equal to] 0.50 and 0 if Pi < 0.50 (5). The scores are interpreted as follows: banks with prudent management are identified as 0 and banks with imprudent management are identified as 1.

The Pi and PI are calculated for a sample of all public and nonpublic commercial banks with total assets greater than $10 million (except for banks chartered less than five years) that filed Reports of Condition and Income (aka "Call Report") with United States federal banking supervisors. The time frame encompassed quarterly data reported from December 31, 1984 through December 31, 2004. The number of commercial banks evaluated ranged from nearly 8,000 at the end of 1984 to about 4,800 at the end of 2004 for the 81 quarters identified. The number of firm-quarter observations in the final sample was 544,072.

RESEARCH DESIGN

Hypothesis

The PI developed in the previous section measures the extent to which the bank's management is taking actions in the current period that will increase future credit losses. This paper hypothesizes that managers that are imprudent are less conservative in recognizing the increased credit risk in the loan portfolio in the bank's current-period provision. Conservative accounting would recognize an increase in the provision in the same period that the imprudent decisions are made. If a manager is less conservative in applying accounting principles, the increase in the provision will be delayed to future period(s) and will be recorded as "catch up" adjustment. This leads to the hypothesis tested in this paper:

HI: Banks with managers who are classified as imprudent in the current period will have higher future provisions for loan and lease losses than banks with managers who are not classified as imprudent.

Accounting for Loan Losses

This study relies upon the work of Wahlen (1994) as advanced by Kanagaretnam et al. (2003) and Kanagaretnam et al. (2004). In Wahlen (1994), the current year's provision is a function of three elements: (1) the beginning balance of the ALLL, (2) nonperforming loans at the beginning of the period and the change in nonperforming loans during the period, and (3) total loans outstanding of the six major categories of loans at the beginning of the period. Subsequent studies have relied upon this initial definition of nondiscretionary provisions (Beatty et al., 1995; Beatty et al., 2002; Kanagaretnam et al., 2003; Kanagaretnam et al., 2004; Kim and Kross, 1998), with some modification. This paper adds to Wahlen (1994) by adding a variable to examine the effect of managerial prudence on the provision. A discussion of the three variables used in Wahlen (1994), which are used to control for the nondiscretionary portion of the provision in this study, follows.

The Beginning Balance of the ALLL

The beginning of the period ALLL represents management's expectations of estimated loss at the beginning of the period, and historic loss estimates generally predict future provisioning behaviors (Wahlen, 1994). Additionally, this variable should explain whether the ALLL is either over or under funded in the prior year (Beatty et al., 1995). This variable should be negatively associated with provisions because the bigger the balance of the combined historical discretionary and nondiscretionary ALLL is at the beginning of a period, the lower the nondiscretionary provision expense needs to be in the following period (Kanagaretnam et al., 2004).

Non-performing Loans

Non-performing loans are loans with payment in arrears by at least 90 days. The total of non-performing loans can be used as a proxy for the level of loan quality within the portfolio (Beatty et al., 2002). Expected behaviors for provisioning include an understanding of provisioning for both historical non-performing loans and the change in non-performing loans that occurs within the period (Wahlen, 1994). In other words, as the level of non-performing loans within a loan portfolio increases, the provision for loan losses should also increase. The opposite should also be true. As the level of nonperforming loans within the portfolio decreases, the provision should decline.

Total Loans Outstanding

Wahlen (1994) assumes that the aggregate number of loans outstanding in six predominant loan categories would explain the migration of loans into riskier categories over the period. Kanagaretnam et al. (2004) include a variable for the change in total loans outstanding, and hypothesize that the relationship between this variable and provisioning behavior would depend upon the quality of new loan growth. Others have included this variable to measure overall loan growth and have assumed that loan growth would require a larger ALLL, which would result in an increase in the provision (Beaver and Engel, 1996; Kanagaretnam et al., 2003). By this same logic, shrinkage in the loan portfolio would necessitate a smaller ALLL and a lower provision (Beaver and Engel, 1996; Kanagaretnam et al., 2003). In this study, a loan growth ratio (ATL) is included to explain that growth should be accompanied by directionally consistent changes in the provision. Movements toward riskier loans within the loan portfolio should be explained by the PI. Table 2 presents descriptive statistics for the sample. The mean of scaled, annualized, loan loss provision expense is consistent with other studies of loan loss provisioning behaviors (Ahmed et al., 1999; Kanagaretnam et al., 2004). While the level of negative provisions for this sample is larger than other studies, a review of the data reveals that negative provisions have been more prevalent in the last five years of data used in this study, and earlier studies do not include these more recent years. For example, Ahmed et al. (1999) evaluate the time frame between 1987 and 1995, and Kanagaretnam et al. (2004) between 1980 and 1997. Kanagaretnam et al. (2003) eliminate observations with low or negative provisions. Since this study evaluates prudent behavior characteristics, potentially aggressive behaviors may be important to include in the data set; therefore, low and negative provisions are retained in the data set.

Regression Equation 

Based on the above considerations, the following regression equation is used to determine whether a relationship exists between the loan loss provision and managerial prudence:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.]

i = bank

t = time

ALLL = Allowance for loan and lease losses

LLP = Loan loss provision expense

NPL = Nonperforming loans

ANPL = Change in the value of nonperforming loans since the prior period

ATL = Change in total loans since the prior period

PI = Prudence indicator (prudent management = 0 and imprudent management = 1)

E = Error term

[a.sub.0] = Intercept for prudent management

Consistent with Kanagaretnam et al. (2003), this study deflates all independent variables, except for the PI, by the balance of outstanding loans at the beginning of the period.

RESULTS

Coefficients for the Nondiseretionary Independent Variables

As Table 3 shows, the coefficient for the beginning of the period balance of the ALLL (-0.033) is negative and statistically different from zero (p-value = 0.0237). This variable measures the amount of the ALLL that had been previously recorded with a provision through earnings but had not yet been eliminated through a charge-off of a problem loan. The ALLL balance would be available for any loan charge-off; therefore, if the ALLL balance were high, there would be less need for additional future provisions.

The coefficients for the beginning of the period nonperforming loans (0.086) and the change in nonperforming loans (0.035) are both positive and statistically significant. These variables measure the level of poor quality loans within the portfolio at the beginning of the period and the changes in the level of poor quality loans during the period. The positive relationships of these variables with the dependent variable LLP are as expected, as higher levels of non-performing loans necessitate greater future loss provisions (Kanagaretnam et al., 2003). After controlling for heteroscedasticity, the coefficient for loan growth (ATL) is positive (0.001) but not significant (p-value < 0.2108).

While the commercial loans to total loans ratio measures prudence, it could also measure credit risk and be correlated with the other independent variables. In order to evaluate whether this prudence factor is correlated with the other variables and potentially biases the results, the Pi factor and regression results are calculated without the commercial loans to total loans ratio factor. The outcome of this test indicates that the revised regression results do not change when the C&I factor is excluded. Furthermore, to test for correlation between the PI and the other independent variables in the final regression equation, the Pearson Correlation Coefficient test is used. There is no evidence of correlation, which further supports the conclusion that the results are not biased by including the commercial loans to total loans ratio as a factor in the Pi.

Coefficient for the PI

The coefficient on PI is positive (0.04269) and statistically significant (p-value < 0.0001). This relationship may seem contrary to expectations, because one might expect prudent management to over-provide for losses. However, LLP is measured at time (t) while PI is measured at time (t-1), indicating that imprudent management is recognizing the losses later than more prudent management. Conversely, prudent managers recognize losses sooner. This finding is consistent with management becoming less prudent and less conservative in accounting by delaying the recognition of the increased risk in the provision. Imprudent management may lack diligence in recognizing problem loans and may increase provisions to compensate for problems that are not identified in a timely manner and are not included as part of the nondiscretionary provisions. Alternatively, imprudent bankers may not sufficiently provide for identified nondiscretionary loss estimates. These findings support the view that imprudent bankers may have been compelled to provide at higher levels in order to address past provisioning shortfalls. All bank managers are eventually required to provide a reasonable ALLL to maintain a satisfactory bank rating from the regulator and to satisfy the auditor (Douthett et al., 2001).

Additional evidence is obtained by examining the ratio of non-performing loans to ALLL and the ALLL as a percentage of total loans, for both prudent bank observations and imprudent bank observations. This analysis reveals, on average, the banks with less prudent management have lower coverage ratios of both non-performing loans and of total loans. These results indicate that while imprudent managers may periodically recognize larger annual provisions, the increase in the annual provision lags behind management's more aggressive (less prudent) actions. Additionally, the lower coverage ratios pertaining to the ALLL as a percentage of total loans and non-performing loans indicate that banks with imprudent management face a greater likelihood of having an inadequate ALLL to provide for known losses.

CONCLUSIONS AND IMPLICATIONS

Prudence, defined as the extent of bank management's focus on the long-term viability of the organization, is associated with loan loss provisioning behaviors. Banks with less prudent leadership have subsequent higher ratios of quarterly provision expense to average loans than banks with prudent leadership. Furthermore, banks with imprudent leadership have less robust coverage of estimated losses within their loan loss allowance balances, which suggests that banks with imprudent leadership did not sufficiently provide for identified nondiscretionary and other probable loan losses. In other words, imprudent managers delay the recognition of loan loss provisions that are needed due to changes in loan quality.

This study indicates that managerial prudence influences accounting decisions with respect to loan loss estimates. Now that this relationship is established, studies designed to evaluate this association over a time-series or when loan quality is stressed appear warranted. Furthermore, future studies that evaluate the relationship between bank loan loss provision behaviors and earnings projections, stock price movements, or compensation contracts may consider including the effects of managerial prudence within their results.

Future research should be conducted to more fully understand the concept of prudence and its relationship with loan loss provisioning. Research on the relationship between management culture and accounting conservatism is particularly important considering the shift in accounting's conceptual framework from conservative principles to an unbiased estimate approach without a full understanding of the potential information that could be lost as a result of this change (Watts, 2003a). The results of this study indicate that the extent of managerial prudence influences provisioning behaviors within the current conservative accounting framework. Furthermore, the results suggest that imprudent leadership may provide insufficient amounts for identified problem loans.

References

Ahmed, A. S. and S. Duellman. 2007. "Accounting Conservatism and Board of Director Characteristics: An Empirical Analysis." Journal of Accounting and Economics 43(2-3): 411-437.

--, C. Takeda, and S. Thomas. 1999. "Bank Loan Loss Provisions: A Reexamination of Capital Management, Earnings Management, and Signaling Effects." Journal of Accounting and Economics 28(1): 1-25.

Basu, S. 2003. "Why Do Banks Fail?" International Review of Applied Economics 17(3): 231-248.

--. 1997. "The Conservatism Principle and the Asymmetric Timeliness of Earnings." Journal of Accounting and Economics 24: 3-37.

Beattie, V. A., P. D. Casson, R. S. Dale, G. W. McKenzie, C. M. S. Sutcliffe, and M. J. Turner. 1995. Banks and Bad Debts: Accounting for Loan Losses in International Banking. Hoboken, NJ: John Wiley & Sons.

Beatty, A., S. L. Chamberlain, and J. Magliolo. 1995. "Managing Financial Reports of Commercial Banks: Influence of Taxes, Regulatory Capital, and Earnings." Journal of Accounting Research 33(2): 231-261.

--, B. Ke, and K. R. Petroni. 2002. "Earnings Management to Avoid Earnings Declines across Publicly and Privately Held Banks." The Accounting Review 77(3): 547-570.

Beaver, W. H. and E. E. Englel. 1996. "Discretionary Behavior with Respect to Allowances for Loan Losses and the Behavior of Security Prices." Journal of Accounting and Economics 22(1-3): 177-206)

Board of Governors of the Federal Reserve System. 2006. Commercial Bank Examination Manual. Washington, DC: Board of Governors of the Federal Reserve System.

"Charge-off and Delinquency, Rates on Loans and Leases at Commercial Banks." 2005. Retrieved November 25, 2005. http://www.federalreserve.gov/releases/chargeoff.

Douthett, E. B., J. E. Duchac, and D. L. Warren. 2001. "The Association between Auditor Size and Bank Regulator Ratings." Journal of Managerial Issues 13(1): 13-27.

DuCharme, L. L., P. H. Malatesta, and S. E. Sefcik. 2004. "Earnings Management, Stock Issues, and Shareholder Lawsuits." Journal of Financial Economics 71(1): 27-49.

Federal Financial Institutions Examinations Council. 2006. Instructions for Preparation of Consolidated Reports of Condition and Income. Washington, DC: Federal Financial Institutions Examinations Council.

Financial Accounting Standards Board. 2010. Statements of Financial Accounting Concepts. Norwalk, CT.

Financial Accounting Standards Board. 2007. Statements of Financial Accounting Concepts. Norwalk, CT.

Financial Accounting Standards Board. 2006. "Preliminary View: Conceptual Framework for Financial Reporting: Objective of Financial Reporting and Qualitative Characteristics of Decision-useful Financial Reporting Information." Financial Accounting Series No. 1260-001. Norwalk, CT.

Financial Accounting Standards Board. 2004. Original Pronouncements, as Amended. FASB Concepts Statements. Norwalk, CT.

Givoly, D., C. K. Hayn, and A. Natarajan. 2007. "Measuring Reporting Conservatism." The Accounting Review 82(1): 65-106.

Gray, R. P. and F. L. Clarke. 2004. "A Methodology for Calculating the Allowance for Loan Losses in Commercial Banks." Abacus 40(3): 321-341.

Greenawalt, M. B. and J. F. J. Sinkey. 1988. "Bank Loan-Loss Provisions and the Income Smoothing Hypothesis: An Empirical Analysis, 1976-1984." Journal of Financial Services Research 1(4): 301-318.

Hanc, G. 1997. "The Banking Crisis of the 1980s and Early 1990s: Summary and Implications." Chapter in History of the Eighties: Lessons for the Future, Volumel: An Examination of the Banking Crisis of the 1980s and Early 1990s. Eds. D. Voesar, J. Lewin, and C. Blair. Washington, DC: Federal Deposit Insurance Corporation. pp. 4-85.

Handorf, W. C. and L. Zhu. 2006. "US Bank Loan-Loss Provisions, Economic Conditions, and Regulatory Guidance." Journal of Applied Finance 16(1): 97-114.

Kanagaremam, K., G. J. Lobo, and R. Mathieu. 2003. "Managerial Incentives for Income Smoothing Through Bank Loan Loss Provisions." Review of Quantitative Finance and Accounting 20(1): 63-80.

--, and D. H. Yang. 2004. "Joint Tests of Signaling and Income Smoothing Through Bank Loan Loss Provisions." Contemporary Accounting Research 21 (4): 843-884.

Kim, M. S. and W. Kross. 1998. "The Impact of the 1989 Change in Bank Capital Standards on Loan Loss Provisions and Loan Write-offs." Journal of Accounting and Economics 25: 69-99.

King, T. B., D. A. Nuxoll, and T. J. Yeager. 2006. "Are the Causes of Bank Distress Changing? Can Researchers Keep Up?" Federal Reserve Bank of St. Louis Review 88(1): 57-80.

Kothari, S., A. Leone, and C. Wasley. 2005. "Performance Matched Discretionary Accrual Measures." Journal of Accounting and Economics 39(1): 163-197.

LaFond, R. and R. L. Watts. 2008. "The Information Role of Conservatism." The Accounting Review 83(2): 447-478.

Larcker, D. F., S. A. Richardson, and I. Tuna. 2007. "Corporate Governance, Accounting Outcomes, and Organizational Performance." The Accounting Review 82(4): 963-1008.

Liu, C. C. and S. G. Ryan. 2006. "Income Smoothing over the Business Cycle: Changes in Banks' Coordinated Management of Provisions for Loan Losses and Loan Charge-offs From the pre-1990 Bust to the 1990s Boom." The Accounting Review 81(2): 421-441.

Louis, H. and D. Robinson. 2005. "Do Managers Credibly use Accruals to Signal Private Information? Evidence from Pricing of Discretionary Accruals around Stock Splits." Journal of Accounting and Economics 39(1): 361-380.

Office of the Comptroller of the Currency. 1988. Bank Failure: An Evaluation of the Factors Contributing to the Failure of National Banks. Washington, DC: Office of the Comptroller of the Currency.

Oshinsky, R. and V. Olin. 2005. "Troubled Banks: Why Don't They Fail?" Federal Deposit Insurance Corporation Working Paper No. 2005-03, pp. 1-50. Retrieved November 26, 2005. http://www.fdic.gov/bank/analytical/working/wp2005/WP2005_03.pdf.

Roychowdhury, S. and R. L. Watts. 2007. "Asymmetric Timeliness of Earnings, Market-to-Book and Conservatism in Financial Reporting." Journal of Accounting and Economics 44(1-2): 2-31.

Securities and Exchange Commission. 2001. SEC Staff Accounting Bulletin: No. 102--Selected Loan Loss Methodology and Documentation Issues. Available at: http://www.sec.gov/interps/account/sab102.htm

Smith, A. 1969. Theory of Moral Sentiments. New Rochelle, NY: Arlington House, 1969. (Originally published 1853)

Thomson, J. B. 1991. "Predicting Bank Failures in the 1980s." Federal Reserve Bank of Cleveland Economic Review 27(1): 9-20.

Wahlen, J. M. 1994. "The Nature of Information in Commercial Bank Loan Loss Disclosures." The Accounting Review 69(3): 455-478.

Wall, L. D. and T. W. Koch. 2000. "Bank Loan-Loss Accounting: A Review of Theoretical and Empirical Evidence." Federal Reserve Bank of Atlanta Economic Review 85(2): 1-19.

Watts, R. L. 2003a. "Conservatism in Accounting Part I: Explanations and Implications." Accounting Horizons 17(3): 207-221.

Watts, R. L. 2003b. "Conservatism in Accounting Part II: Evidence and Research Opportunities." Accounting Horizons 17(4): 287-301.

Wheelock, C. D. and P. W. Wilson. 2000. "Why Do Banks Disappear? The Determinants of the U. S. Bank Failures and Acquisitions." The Review of Economics and Statistics 82(1): 127-138.

(1) The authors appreciate the helpful comments of the Editor and two anonymous reviewers on prior drafts of this paper.

(2) This allowance is also commonly referred to as the reserve for loan losses and, internationally, as the provision account. It refers to the amount on a bank's balance sheet that represents the estimated incurred losses inherent within the bank's loan portfolio. The balance decreases through charging off bad loans and increases through provision expenses taken through a bank's income statement or through recoveries on prior period loan losses.

(3) The concept of prudence has been studied previously in banks. However, typically the concept of prudence is studied in the context of "prudent regulation" or prudent investment choices relying on the legal concept of the "prudent man."

(4) Watts (2003a) suggests there are four primary reasons to employ conservative accounting behaviors. The first three apply to management and their information environment and the last applies to the accounting standard setters and securities regulators. First, there are contracting benefits associated with conservative accounting practices. Second, managers may be conservative in response to the threat of shareholder suits and claims of aggressive accounting practices. Third, conservative accounting techniques could lower tax liabilities and conform to tax strategies. Fourth, accounting standard setters and securities regulators may support conservative accounting standards in response to a public preference for conservatism. Watts (2003b) puts forth these potential explanations for conservative accounting behaviors and suggests that conservative accounting practices may be motivated by reasons that are both self-serving and efficient given the costliness of information.

(5) The 0.50 cutoffs used for the PI index are arbitrary. Other possible classification schemes could focus on the more extreme measures of Pi. Using cutoffs that focus on observations in the ends of the distributions increases the probability of finding results. The 0.5 cutoff is a reasonable starting point given that the PI index is new to the literature.

Linda V. Ditchkus *

Faculty Member

University of Phoenix

Retired, Federal Reserve Board of Governors

Washington, DC

Gregory E. Sierra

Assistant Professor of Accounting

Southern Illinois University Edwardsville

Brad J. Reed

Professor of Accounting

Southern Illinois University Edwardsville

* The opinions expressed here do not necessarily represent those of the Federal Reserve Board of Governors or the Federal Reserve System.

Table 1

Prudence Indicator Factors Ranked by Order of Influence

Prudence Factor

Proportion of commercial and industrial loans (COMPROP)

Salary expense ratio (SAL)

Growth (GROWTH)

Interest and fees on loans to average loans (INT)

Recoveries to prior period loan losses (RECOV)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.]

Data are taken from "Call Reports" (reporting forms FFIEC
031 and FFIEC 041), which are regulatory filing of U. S.
commercial banks. All data are publicly available. For more
information on the Call Report see the Federal Financial
Institutions Examination Council's (FFIEC) web site at ffiec.gov.
Specific Call Report Variable Identification Numbers are as
follows: Commercial and industrial loans (RCFDI600), total
loans (RCFD2122), salary expense (RIAD4135), full-time staff
equivalent (FIAD4150), total assets (RCFD2170), interest and
fees on loans (RIAD4010), average loans (RCFD3360), charge-off
recoveries (RIAD4605), and loan losses (RIAD4230).

Table 2
Descriptive Statistics

Descriptive statistics for the sample of commercial banks, with total
assets of greater than $10 million, except for newly chartered banks,
that filed Reports of Condition and Income with United States
Federal banking supervisors

Variable       Mean       SD      Median      Min        Max

LLP           0.2095    0.5838    0.0849   -63.0037   114.8985
ALLL          1.6338    1.0256    1.3639     0.0000    94.3112
N PL          1.9746    2.4919    1.1652     0.0000    83.1202
[DELTA]NPL   -0.0079    1.2091   -0.0091   -37.1701   159.9798
Growth        0.0205    0.1338    0.0125    -0.9345    21.2932
PI Index     50.0479   13.2760   49.8000     4.4000    98.0000

Data are taken from "Call Reports" (reporting forms FFIEC 031 and
FFIEC 041), which are regulatory filing of U. S. commercial banks.
All data are publicly available. For more information on the Call
Report see the Federal Financial Institutions Examination Council's
(FFIEC) web site at ffiec.gov. All data are publicly available except
those items related to the aging schedule of loans receivable
(items RCFD 1407 and RCFD 1403).

LLP: Loan loss provision expense (RIAD4230) scaled by beginning of
the period total loans (RCFD2122).

ALLL: Allowance for loan and lease losses (RCFD3123) scaled by
beginning of the period loans (RCFD2122).

NPL: Non-performing loans, which includes loans delinquent greater
than 90 days (RCFD1407 + RCFD1403), scaled by beginning of the period
loans (RCFD2122).

[DELTA]NPL: Change in the value of non-performing loan since the
prior period scaled by beginning of the period loans (see definition
of NPL Ratio).

Growth: Total assets at the end of the period (RCFD2170)--total assets
at the beginning of the period scaled by total assets at the
beginning of the period.

PI Index: Prudence Indicator (see Table 1).

Table 3
Regression Results for the Entire Sample

Variable     Parameter   T statistics   P values    Expected
             Estimate                              Coefficient

Intercept     0.7363         3.10        0.0010         +
ALLL         -0.03322       -1.98        0.0237         -
NPL           0.08561       32.94       <0.0001         +
[DELTA]NPL    0.03471        4.63       <0.0001         +
[DELTA]TL     0.00064        0.80        0.2108         +
PI            0.04269       29.36       <0.0001         +

Variable      Resulting
             Coefficient

Intercept         +
ALLL              -
NPL               +
[DELTA]NPL        +
[DELTA]TL         +
PI                +

[LLP.sub.it] = [[alpha].sub.0] + [a.sub.1][ALLL.sub.it-1] +
[[alpha].sub.2][NPL.sub.it-1] + [[alpha].sub.3][DELTA][NPL.sub.it] +
[[alpha].sub.4][DELTA][TL.sub.it] + [[alpha].sub.5][Pl.sub.it] +
[[epsilon].sub.it]

Note: Adjusted R-Squared is 0.1133 and the t and p values are adjusted
for heteroscedasticity. Data are taken from "Gall Reports" (reporting
forms FFIEC: 031 and FFIEC 041), which are regulatory filing of U. S.
conunercial banks. All (lata are publicly available. For more
information on the Call Report see the Federal Financial Institutions
Examination Council's (FFIEC) web site at fec.gov. All data are
publicly available except those items related to the aging schedule of
loans receivable (items RCFD1407 and RCFD1403).

LLP: Loan loss provision expense (RIAD4230) scaled by beginning of the
period total loans (RCFD2122).

ALLL: Allowance for loan and lease losses (RCFD3123) scaled by
beginning of the period loans (RCFD2122).

NPL: Non-performing loans, which includes loans delinquent greater than
90 days (RCFD1407 + RCFD1403), scaled by beginning of the period loans
(RCFD2122).

[DELTA]NPL: Change in the value of non-performing loan since the prior
period scaled by beginning of the period loans (see definition of NPI.
Ratio).

[DELTA]TL: Total loans at the end of the period (RCFD2122)--total
loans at the beginning of the period scaled by total loans at the
beginning of the period.

PI Index: Prudence Indicator (see Table 1).