Loan Loss Provisioning Accounting Features

Journal of Economics and Business 78 (2015) 92–117
Contents lists available at ScienceDirect
Journal of Economics and Business
Loan loss provisions, accounting constraints,

and bank ownership structure
Eliana Balla a, Morgan J. Rose b,∗
a
Federal Reserve Bank of Richmond, 502 S. Sharp Street, Baltimore, MD 21201, United States
b
UMBC and OCC, 1000 Hilltop Circle, Baltimore, MD 21250, United States
a r t i c l e i n f o a b s t r a c t
Article history: We examine bank-level changes in the relationship between earn-

Received 3 December 2013 ings and loan loss provisioning, a measure of earnings management,
Received in revised form 13 October 2014 following the tightening of accounting constraints associated with
Accepted 24 December 2014
the SEC’s 1998 SunTrust Bank decision. By exploiting both tem-
Available online 31 December 2014
poral variation in the regulatory environment and cross-sectional
variation in bank ownership structure, we find evidence that
JEL classification:
shortly after the SEC action, the relationship between earnings and
G21
provisions weakened for publicly-held banks but not for privately-
G28
G32 held banks, consistent with reduced earnings management among
E65 publicly-held banks only. This difference does not persist over time,
with evidence indicating a weakening of the relationship for both
Keywords: ownership types.
Loan loss provisioning
Published by Elsevier Inc.
Earnings management
Ownership structure
Privately-held banks
Financial institutions
Banking regulation
1. Introduction
Understanding the trade-offs between the supervisory and accounting standards for loan loss pro-
visions and the economic significance of such trade-offs is critical for ongoing bank regulation. In the
∗ Corresponding author. Tel.: +1 410 455 8485.

E-mail address: mrose@umbc.edu (M.J. Rose).
http://dx.doi.org/10.1016/j.jeconbus.2014.12.003
0148-6195/Published by Elsevier Inc.
E. Balla, M.J. Rose / Journal of Economics and Business 78 (2015) 92–117 93
wake of the 2007–2009 financial crisis and the ensuing reforms in the Dodd–Frank Act and Basel III,
policymakers, academics, and industry experts have debated the nature of embedded losses in banks’
loan portfolios. Some argue that prior to the crisis bank managers underestimated the credit risk in
their loan portfolios; others argue that bank managers may have been constrained by accounting rules
from recognizing estimated loan losses early.
We present new evidence on bank-level changes in loan loss provisioning in response to a shift
in the regulatory environment, evidence that is consistent with the second view. We utilize a natural
experiment that occurred a decade before the financial crisis to obtain a cleaner examination of the
accounting constraint. Specifically, the 1998 SunTrust decision (described below) by the Securities
and Exchange Commission (SEC) indicated stricter enforcement of accounting priorities relative to
supervisory priorities, but at first directly affected only publicly-held banks that fall under the SEC’s
purview. By exploiting both temporal variation in regulatory emphasis and cross-sectional variation
in bank ownership structure, we identify the effects of the strengthening of the accounting constraint
on loan loss provisioning.
We find that in the two years following the SEC action, the positive relationship between earnings
and provisions weakened for publicly-held banks but not for privately-held banks, consistent with
a tightening of accounting constraints affecting only publicly-held banks directly under the SEC’s
purview. This is true for both small and large banks, although large privately-held banks exhibit a
stronger relationship between earnings and provisions after the SEC action. Measured over a longer
time period, including several years following the joint interagency guidance of 2001, we find that
both publicly-held and privately-held banks show a weakening of the relationship between earnings
and provisions, consistent with the uniform application of provisioning standards for all banks by
supervisory agencies. The only exception is large publicly-held banks, which do not show a significant
weakening in earnings management over the long term. We also examine changes in loan loss reserves
after the SEC action. We find that the reserves fell for publicly-held banks relative to privately-held
banks, particularly among small banks, and that this difference persisted in the years leading up to the
crisis. This too is consistent with a period following the SEC action during which publicly-held banks
were constrained against provisioning as much as they may have desired while privately-held banks
were not similarly constrained.
This paper makes several contributions to the literature on loan loss provisions and earnings man-
agement in banks. First, ours is the first paper to the best of our knowledge to examine the short term
and long term implications of the 1998 SEC action on provisioning policies of U.S. banks, and does
so for both publicly-held and privately-held banks, and banks stratified by size. Second, our paper
speaks to the efficacy of financial regulation, as the evidence indicates that the SEC was successful
in its intention of constraining earnings management among publicly-held banks. Third, and most
relevant for the ongoing debate over bank regulation in the wake of the financial crisis, our evidence
that the relative strengthening of accounting priorities reduced earnings management, and that the
resulting difference in loan loss reserves persisted, supports the argument that binding accounting
constraints prevented banks from recognizing loan losses early in the years prior to the 2007–2009
crisis. This implies that the subsequent need under the incurred-loss model for banks to procyclically
increase provisions during the crisis potentially amplified the economic downturn.
2. Background
A bank’s loan loss reserve account, also known as the allowance for loans and leases losses, is a
contra-asset account used to reduce the value of total loans and leases on the bank’s balance sheet
by the amount of losses that bank managers anticipate in the most likely future state of the world.1
Provisioning is the act of building the loan loss reserve account through a provision expense item on
1
Economists generally view loan loss reserves as intended to capture expected future losses that will occur if a borrower does
not repay in accordance with the loan contract, a view most helpful for the pricing of loans in the secondary market. Benston
and Wall (2005) point out that if loans could be reported reliably at fair value, where fair value is value in use, there would
be no need for a loan loss provision or reserves. A market for the full transfer of credit risk does not exist and loans cannot be
reported reliably at fair value.
94 E. Balla, M.J. Rose / Journal of Economics and Business 78 (2015) 92–117
the income statement. As a relatively large accrual for commercial banks,2 loan loss provisions have a
significant effect on earnings and regulatory capital.3
With respect to loan loss provisioning policy, there is a tension between what might be termed
accounting priorities and supervisory priorities.4 Accounting priorities emphasize the objectivity and
comparability of financial statements to facilitate bank monitoring. As reflected by the “incurred-loss
model” under standards set by the Financial Accounting Standards Board (FASB),5 an inherent credit
loss should be recognized only upon the occurrence of an event indicating that a loss is probable and
if the amount of the loss can be reasonably estimated.6 Supervisory priorities emphasize the ability of
banks to maintain solvency through changing business environments. The evaluation of the adequacy
of loan loss reserves is one of the most important functions of bank examinations.7 From the perspec-
tive of bank supervisors, adequate provisioning is a safety and soundness issue because a deficit in loan
loss reserves implies that the bank’s capital ratios overstate its ability to absorb unexpected losses.8
For the supervisory priorities to hold in practice, bank managers would incorporate into their loan loss
provisioning expectations about future losses due to changes in economic conditions that affect credit
defaults, even if no event has yet occurred to indicate specific estimable losses.9 The tension between
objective but backward-looking historical data on the one hand and subjective but forward-looking
expectations on the other reflects a trade-off between two laudable goals, transparency of financial
statements and safety and soundness.
That tension is heightened because the incurred-loss model has been criticized as contributing
to procyclicality in the banking sector. Stylized facts and a substantial literature suggest that bank
lending behavior is highly procyclical. The classical principal-agent problem between shareholders
and managers may lead to procyclical bank lending if managers’ objectives are related to credit growth.
Rajan (1994) suggests that credit mistakes are judged more leniently if they are common to the whole
industry. Berger and Udell (2003) suggest that, as the time between the current period and the last
crisis increases, experienced loan officers retire or genuinely forget about the lending errors of the last
crisis and become more likely to make “bad” loans. Asea and Blomberg (1998) and Lown and Morgan
(2006) show that changing credit standards over the business cycle contribute to the procyclicality of
bank lending and exacerbate the business cycle.
The concern with the incurred-loss model, as with bank regulatory capital more broadly, is that a
provisioning approach in which banks have to rapidly raise provisions during bad times could prolong
the bad times. Laeven and Majnoni (2003) and Bouvatier and Lepetit (2008) discuss the procyclicality
of loan loss provisions with cross-country data. They argue that banks delay provisioning for bad
loans until economic downturns have already begun, amplifying the impact of the economic cycle
on banks’ income and capital. If loan losses are not recognized until the occurrence of specific events,
then during good economic times when fewer such events occur, provisions will be relatively low. The
incurred-loss approach to provisioning also does not reflect the relaxation of underwriting standards
and greater risk-taking that often occurs in banks during a booming economy, given that most of the
resulting bad loans will only reveal themselves in a recession. Requiring banks to build up reserves
during an economic downturn, when bank funds may already be strained, can compel banks to reduce
2
During 1992–2013, the median and mean ratios of loan loss provisions to earnings before provisions and taxes for our
dataset (winsorized at the 1st and 99th percentiles) were 6.7 and 14.6 percent, respectively.
3
See Ahmed et al. (1999).
4
See Wall and Koch (2000) for an extensive summary of the theoretical and empirical evidence on bank loan loss accounting
and provisioning philosophies.
5
Financial Accounting Standards Board and the International Accounting Standards Board. Information for Observers. March
2009 Meeting. Project: Loan Loss Provisioning.
6
The Appendix of this paper discusses regulatory and accounting standards concerning provisioning policy in greater detail.
7
See Gunther and Moore (2003a, 2003b) for a discussion of the role of bank exams for bank loan loss reserves.
8
The view that loan loss reserves serve to cover expected credit losses and capital unexpected losses is reflected in the Basel
II (2006) and Basel III (2011) capital frameworks. See Laeven and Majnoni (2003), Appendix A, for a detailed description of the
conceptual relationship between loan loss reserves, provisions, capital, and earnings.
9
One way to separate the incurred-loss perspective from an expected-loss perspective is by stating that no expected economic
impacts are taken into account in incurred-loss methodology. A bank manager cannot, for example, consider the increases in
default risk due to future increases in unemployment.
lending activities, potentially magnifying the downturn by exacerbating a credit crunch. Bouvatier
and Lepetit (2012) develop a theoretical model analyzing how provisioning rules affect bank lending.
They find that a backward-looking provisioning approach magnifies the procyclicality of bank lending,
while in a forward-looking approach in which provisions are smoothed over the business cycle the
procyclicality of bank lending does not arise.
Provisioning more when earnings are high could reflect bank managers building up reserves as a
precaution against economic downturns as supervisory priorities suggest. This early recognition of
loan losses is observationally equivalent to earnings management (or “smoothing income”) across
the business cycle to portray greater earnings stability than is really the case, which is viewed as
undesirable by the accounting profession.10 Lambert (1984) and Dye (1988) illustrate how risk-averse
managers without access to capital markets have incentives to manage reported earnings. Trueman
and Titman (1988) show that managers may have an incentive to manage earnings even in the absence
of risk aversion and restricted access to capital markets in order to reduce perceptions of firm earnings
volatility, lowering assessments of the probability of financial distress and potentially increasing firm
market value. Managers may also manage earnings to increase compensation tied to meeting specific
earnings thresholds, deliberately reducing earnings in good periods so they can raise earnings in bad
periods when they would otherwise miss their thresholds. Wall and Koch (2000) offer a review of
the theoretical and empirical evidence on earnings management in banks via loan loss accounting.
The evidence they summarize suggests that banks have an incentive to manage reported earnings
and that, while the empirical evidence is not conclusive, several papers find that banks use loan loss
accounting to manage reported earnings.11
The existing literature on bank earnings management has typically focused on publicly-held
banks.12 It is unclear a priori whether privately-held firms manage earnings more or less than publicly-
held firms. Because outside investors have less information on privately-held firms, reported earnings
may have relatively more importance in terms of signaling for privately-held firms, possibly giving
them greater incentives to manage earnings. On the other hand, assuming that equity-based compen-
sation is less relevant for privately-held firms, managers at privately-held firms will have less incentive
to manage earnings in order to influence equity value. Beatty, Ke, & Petroni (2002) find that both public
and private banks manage earnings but public banks manage earnings more.13 Fonseca and González
(2008) provide a panel study of 40 countries (excluding the United States) and find that neither the
amount of income smoothing using loan loss provisions nor the difference in income smoothing using
loan loss provisions between public and private banks is stable across countries.14
In 1997, the SEC expressed concern that publicly-held U.S. banks were systematically overstating
their loan loss reserves in order to manage earnings. The following year, the SEC required SunTrust
Bank to restate its earnings for 1994–1996 based on a stricter interpretation of accounting standards
than had previously been applied by SunTrust, lowering the bank’s loan loss reserve by $100 million.
While directed toward a single bank, the SEC action signaled to publicly-held banks the SEC’s lack
of tolerance for over-provisioning in good economic times. As described by Wall and Koch (2000),
the SEC action sparked lengthy debate among banking analysts, accounting standard-setters, and
bank regulators over the appropriate balance of accounting and supervisory priorities with respect to
loan loss provisioning. Shortly thereafter, the SEC and bank supervisory agencies (the Federal Deposit
Insurance Corporation, the Federal Reserve, the Office of the Comptroller of the Currency, and the Office
of Thrift Supervision) entered a period of dialog to jointly examine loan loss reserve issues in order
to provide further guidance to all banks on appropriate provisioning methodologies. This process was
10
In empirical analysis both are measurable as a positive relationship between pre-provision earnings and loan loss provisions.
11
The evidence Wall and Koch cite includes Greenwald and Sinkey (1988), Wahlen (1994), Ahmed et al. (1999). For more
recent treatments see Beatty et al. (2002), Laeven and Majnoni (2003), Bikker and Metzemakers (2005), Fonseca and González
(2008), Adams et al. (2009), Nichols et al. (2009), Gebhardt and Novotny-Farkas (2011), and Bouvatier et al. (2014).
12
See Kwan (2004) for a general discussion of differences in performance and risk-taking behavior of publicly-traded vs.
privately-held U.S. bank holding companies.
13
Beatty et al. (2002) define “managed earnings” as more frequent announcements of small increases in earnings than small
decreases, reflecting managers’ incentives to avoid the reporting of negative earnings when possible. For another example of
differences between publicly-held and privately-owned banks, see Nichols et al. (2009).
14
Neither of these two studies uses samples that extend past 2002, while our sample extends beyond the crisis to 2013.
1.4
1.2
0.8
0.6
0.4
0.2
All Banks Public Banks Private Banks
Fig. 1. Loan loss provision as a percentage of average assets from the previous quarter. This figure indicates the sample median
loan loss provision as a percentage of average assets from the previous quarter for the fourth quarter of each year from 1990
to 2013. Data are from Call Reports. The vertical line indicates the fourth quarter of 1998, when the SEC action associated with
the SunTrust Bank earnings restatement decision occurred.
reflected in supervisory interagency letters to banks in November 1998, and March and July 1999, and
culminated in the final interagency guidance of July 2001.15 The stance of the 2001 guidance was one
of “prudent, conservative, but not excessive” loan loss reserves, and asserted to the banking industry
the relative importance of accounting priorities as part of the supervisory standards applicable to all
banks, both publicly-held and privately-held.
Following this guidance, many banks entered the financial crisis of 2007–2009 with low loan loss
reserves, and then had to sharply increase provisions in recognition of pending losses (see Fig. 1). In a
speech in March 2009, Ben Bernanke, then Chairman of the Board of Governors of the Federal Reserve,
stated that there is “considerable uncertainty regarding the appropriate levels of loan loss reserves over
the cycle. As a result, further review of accounting standards governing. . .loan loss provisioning would
be useful, and might result in modifications to the accounting rules that reduce their procyclical effects
without compromising the goals of disclosure and transparency.”16 In the aftermath of the 2007–2009
experience, as part of the Basel III communications, the Basel Committee on Banking Supervision
formally encouraged accounting regulatory bodies to pursue a forward-looking loan loss provisioning
regime.17 In July 2014, IASB announced a new standard that will go into effect on January 1, 2018,
which requires banks to provision for potential loan losses based on a forward-looking, expected
losses model.18 As of this writing, FASB is expected to take a similar action in the months ahead.19
15
The full text of these interagency letters can be found at http://www.federalreserve.gov/bankinforeg/srletters/srletters.htm.
The SEC issued parallel guidance for publicly-held banks in SEC Staff Accounting Bulletin 102 (2001).
16
Bernanke (2009).
17
While Basel III (2011) explicitly builds in countercyclical capital requirements, no specific provisioning framework is
provided.
18
IASB Completes Reform of Financial Instruments Accounting. July 24, 2014.
19
Project Update on Accounting for Financial Instruments – Joint Project of the FASB and IASB. February 6, 2014.
3. Hypotheses
Based on the preceding discussion, we empirically test the following hypotheses concerning the
effects of the strengthening of accounting priorities represented by the SEC action:
Hypothesis 1a. In the short term after the SEC action, the relationship between pre-provision earn-
ings and loan loss provisions for publicly-held banks weakened relative to that for privately-held
banks.
Hypothesis 1b. In the short term after the SEC action, the relationship between pre-provision earn-
ings and loan loss provisions was unaffected for privately-held banks.
By requiring a stricter adherence to accounting rules on the part of banks subject to SEC oversight,
the SEC SunTrust action constrained the ability of publicly-held banks to use loan loss management
countercyclically during times of positive earnings to either manage earnings or prudentially increase
loan loss reserves as a precaution against future downturns. Privately-held banks are not subject to SEC
oversight, so their loan loss management need not have been affected. However, if bank supervisors
rapidly incorporated the requirements of the SEC action into the rules as applied to all banks, then
privately-held banks may also have weakened the relationship between earnings and provisions.
Hypothesis 1a implies that the SEC action placed more binding constraints on publicly-held banks than
on privately-held banks. Hypothesis 1b implies that the SEC action did not impose binding constraints
on privately-held banks in the short term.
Hypothesis 2. In the long term after the SEC action, the relationship between pre-provision earnings
and loan loss provisions weakened for both publicly-held banks and privately-held banks.
Hypothesis 2 is consistent with bank supervisors incorporating the requirements of the SEC action
into the rules as applied to all banks, such that the relationship between earnings and provisions
weakened for both publicly-held and privately-held banks. Note that Hypothesis 2 does not require
that Hypotheses 1a and 1b be supported. If bank supervisors incorporated the SEC action requirements
with some delay, then only publicly-held banks would have reacted in the short term but over time the
reaction of privately-held banks would be similar to the reaction of publicly-held banks. If, however,
bank supervisors incorporated the SEC action requirements rapidly after the SEC action, then publicly-
held and privately-held banks would have reacted similarly in the short term.20
Hypothesis 3. In both the short term and the long term after the SEC action, the level of loan loss
reserves fell for publicly-held banks relative to privately-held banks.
Hypothesis 3 is consistent with there being a period following the SEC action in which publicly-
held banks faced more binding constraints against provisioning than privately-held banks, resulting
in the levels of loan loss reserves falling for publicly-held banks relative to privately-held banks. Even
if bank supervisors later incorporated the requirements of the SEC action into the rules as applied to all
banks such that both publicly-held and privately-held banks subsequently faced similar provisioning
constraints, the difference in the level of reserves would persist (i.e., a temporary difference in the
flow variable causes a permanent difference in the stock variable).
4. Data and methodology
The data for this paper come primarily from banking regulatory databases. We use CRSP to identify
U.S. banking institutions that are publicly-held, and link those firms with Federal Reserve data using a
20
Note that there is reason to think that the relationship between earnings and provisions could weaken more for privately-
held banks than for publicly-held banks, in contradiction with our hypotheses. If, prior to the SEC action, SEC oversight and
market analyst coverage already constrained publicly-held banks from earnings management, then the additional constraints
as applied by bank supervisors would primarily have affected the behavior of privately-held banks.
mapping provided by the Federal Reserve Banks of New York.21 We use SNL Financial and SDC Platinum
to validate the IPO dates from CRSP.22 We also use SNL Financial to identify publicly-held institutions
not traded on the three largest exchanges (NYSE, AMEX, and NASDAQ), and exclude those banks from
our sample because the market scrutiny they face may be substantially less than that faced by banks
on the major exchanges.23 While equity offerings are made at the bank holding company level, loan
loss provision policies are generally set at the bank level. For that reason and for data completeness
(small bank holding companies report less frequently and not on a consolidated basis), we use bank-
level financial data from regulatory filings (Call Reports). We obtain structure data on relationships
between holding companies and banks, as well as data on firm age, bank mergers and acquisitions,
and failures, from Federal Reserve databases. If the bank holding company or the top holding company
(one layer removed from the bank) was publicly-held, we consider the bank publicly-held.
Our empirical analysis uses two sample periods. Our short term sample period includes eight quar-
ters before and eight quarters after the fourth quarter of 1998 (the quarter in which the SEC action
occurred). Our long term sample period begins in the first quarter of 1992, when the current rules for
how loan loss reserves enter into bank capital first went into effect.24 The long term sample period
ends in the fourth quarter of 2013. Loan loss provisions increased dramatically during the 2007–2009
financial crisis, particularly among publicly-held banks (see Fig. 1), potentially skewing our results. In
our extensions, we discuss results from treating the crisis and post-crisis periods separately.
From these samples, we drop banks located outside of the continental United States and banks
that are not active lenders, defined as banks for which total loans never exceed 5% of total assets. We
drop de novo bank observations, defined as observations in the first five years of a bank’s existence,
because this portion of a bank’s life cycle may be associated with atypical loan loss provisioning.
For a similar reason, we drop failing banks, defined as the final quarter of a bank’s existence and
the previous three quarters. As noted above, we drop publicly-held banks not traded on one of the
three largest exchanges. We remove outliers by dropping observations for which the value of any
of the ratio variables defined below are below the 0.1st percentile or above the 99.9th percentile.25
Taken together, these sample criteria eliminate approximately 17.2% of the original observations.26
The final dataset includes over 640,000 bank-quarter observations from 13,916 banks. As consolidation
in the banking industry progressed through our sample period, the number of banks in our sample
declined from 9807 in 1992 to 6468 in 2013. Approximately 75% of our sample banks are privately-
held throughout the long term sample period and 16% are publicly-held throughout. The remaining
banks switch ownership structure, with the vast majority switching from private to public.
Table 1 presents summary statistics for the variables used in the empirical analysis. Panel A presents
statistics for all banks in the sample, while Panels B and C present statistics for banks with real assets
under $10 billion and over $10 billion, respectively. The dependent variables are loan loss provisions
as a percentage of average assets from the previous quarter (PLL/Assets) and loan loss reserves as
a percentage of total assets from the previous quarter (LLR/Assets). Public is an indicator equaling
one for publicly-held banks or banks owned by publicly-held holding companies. The percentage of
banks that were publicly-held at the start of our sample was 15%, reached a high of 17% in 1994, and
then declined to a low of 7% in 2013. AfterST equals one in quarters following the SEC action related
to the SunTrust case. PPNR is pre-provision net revenue as a percentage of average assets from the
previous quarter, and denotes our main “earnings” variable. A positive relationship between earnings
21
The mapping is available at http://www.newyorkfed.org/research/banking research/datasets.html, and is valid from
January 1990 to September 2011. We extend the mapping through the end of 2013 by researching ownership status changes
of firms in SNL Financial.
22
We thank Andrew Meyer and Mark Vaughan for sharing their SDC Platinum-based dataset on bank IPOs as a comparison
dataset.
23
We thank Richard Rosen for this suggestion.
24
Note that some financial data from 1991 are used to construct lagged variables.
25
Holod and Peek (2007) drop observations with variable values beyond four standard deviations from the mean. Applying
this approach to our data still left observations with extreme values for some variables.
26
De novo banks represent the majority of the dropped observations (8.9 percent). The next largest source of dropped
observations is bank failures (4.2 percent).
Table 1
Summary statistics. PLL/Assets is provision for loan losses as a percentage of average assets from the previous quarter. LLR/Assets
is loan loss reserves as a percentage of total assets from the previous quarter. AfterST equals 1 after 1998Q4, 0 otherwise.
Public equals 1 if the firm is publicly-held or if the bank-holding company or financial-holding company owning the firm is
publicly-held, 0 otherwise. PPNR is pre-provision net revenues as a percentage of average assets from the previous quarter.
NCO/Assets is net charge-offs as a percentage of average assets from the previous quarter. NPTL/Assets is non-performing loans
as a percentage of total assets from the previous quarter. Loans/Loans is the change in total loans as a percentage of total
loans from the previous quarter. Loans/Assets is total loans as a percentage of total assets from the previous quarter. GDP is
percentage change in real GDP in the previous quarter. EQ is shareholder equity as a percentage of total assets from the previous
quarter. Size is the natural log of total assets (in thousands) in the previous quarter. Merger equals 1 if the firm merged with
another firm during the current quarter, 0 otherwise. The short term and long term sample periods are 1996Q4–2000Q4 and
1992Q1–2013Q4, respectively. Both sample periods exclude 1998Q4, the quarter in which the SEC action occurred.
Variable Short term sample Long term sample
Mean Median Std. dev. Minimum Maximum Mean Median Std. dev. Minimum Maximum
Panel A – all banks
PLL/Assets 0.191 0.108 0.388 −1.459 7.651 0.236 0.116 0.475 −1.475 7.651
LLR/Assets 0.884 0.813 0.455 0.000 6.447 0.935 0.844 0.509 0.000 6.461
Public 0.143 0.000 0.350 0.000 1.000 0.125 0.000 0.331 0.000 1.000
AfterST 0.472 0.000 0.499 0.000 1.000 0.612 0.000 0.487 0.000 1.000
PPNR 1.899 1.853 0.914 −4.559 21.629 1.691 1.672 0.992 −4.650 21.850
NCO/Assets 0.128 0.041 0.365 −1.262 6.847 0.184 0.052 0.454 −1.263 6.862
NPTL/Assets 0.617 0.379 0.786 0.000 11.772 0.841 0.475 1.130 0.000 12.062
Loans/Loans 2.720 2.187 6.666 −37.468 131.892 1.893 1.401 6.547 −38.635 134.530
Loans/Assets 61.371 62.822 15.067 0.139 134.466 60.970 62.470 16.029 0.003 134.847
GDP 1.063 0.989 0.454 0.128 1.889 0.682 0.762 0.597 −2.151 1.889
EQ 10.503 9.621 3.709 2.355 89.523 10.528 9.703 3.723 2.231 90.155
Size 11.368 11.217 1.293 6.890 20.224 11.559 11.408 1.360 6.812 21.411
Merger 0.009 0.000 0.097 0.000 1.000 0.009 0.000 0.095 0.000 1.000
Number of institutions 9432 13,916

Number of observations 128,978 646,178
Panel B – banks under $10 billion in assets

PLL/Assets 0.189 0.107 0.383 −1.459 7.651 0.233 0.115 0.466 −1.475 7.651
LLR/Assets 0.883 0.812 0.454 0.000 6.447 0.932 0.843 0.504 0.000 6.461
Public 0.137 0.000 0.344 0.000 1.000 0.119 0.000 0.324 0.000 1.000
AfterST 0.471 0.000 0.499 0.000 1.000 0.611 1.000 0.488 0.000 1.000
PPNR 0.856 1.850 1.101 −4.400 21.629 1.684 1.669 0.974 −4.650 21.850
NCO/Assets 0.126 0.040 0.359 −1.262 6.847 0.180 0.051 0.445 −1.263 6.862
NPTL/Assets 0.617 0.377 0.788 0.000 11.772 0.839 0.472 1.130 0.000 12.062
Loans/Loans 2.707 2.189 6.603 −37.468 131.892 1.883 1.401 6.498 −38.635 134.530
Loans/Assets 61.333 62.774 15.044 0.139 134.466 60.946 62.430 16.002 0.003 134.847
GDP 1.063 0.989 0.454 0.128 1.889 0.683 0.762 0.597 −2.151 1.889
EQ 10.518 9.635 3.712 2.355 89.523 10.530 9.706 3.719 2.231 90.155
Size 11.325 11.207 1.192 6.890 16.382 11.510 11.397 1.251 6.812 16.524
Merger 0.008 0.000 0.090 0.000 1.000 0.008 0.000 0.090 0.000 1.000
Number of institutions 9378 13,863

Number of observations 128,054 640,818
Panel C – banks over $10 billion in assets

PLL/Assets 0.472 0.263 0.804 −0.689 5.855 0.630 0.269 1.041 −1.396 7.607
LLR/Assets 1.105 1.007 0.608 0.014 5.086 1.208 1.004 0.872 0.001 6.265
Public 0.903 1.000 0.297 0.000 1.000 0.839 1.000 0.368 0.000 1.000
AfterST 0.535 1.000 0.499 0.000 1.000 0.787 1.000 0.409 0.000 1.000
PPNR 2.615 2.349 1.688 −4.415 13.468 2.503 2.063 2.143 −4.528 19.331
NCO/Assets 0.446 0.251 0.790 −0.204 6.384 0.632 0.284 0.982 −0.204 6.856
NPTL/Assets 0.616 0.516 0.429 0.004 3.567 1.117 0.741 1.178 0.000 11.915
Loans/Loans 4.520 1.958 12.498 −32.540 126.615 3.005 1.375 10.889 −36.944 127.257
Loans/Assets 66.638 68.009 17.288 9.637 125.370 63.824 66.577 18.790 0.160 133.086
GDP 1.063 0.989 0.483 0.128 1.889 0.618 0.687 0.615 −2.151 1.889
EQ 8.512 9.399 2.472 4.652 29.478 10.250 9.145 4.230 3.845 60.319
Size 17.328 17.150 0.837 16.088 20.224 17.414 17.150 1.069 15.712 21.411
Merger 0.186 0.000 0.389 0.000 1.000 0.130 0.000 0.336 0.000 1.000
Number of institutions 88 210

Number of observations 924 5360
and provisions indicates that banks on average increase their loan loss provisioning when earnings
are higher, consistent with both earnings management and countercyclical loan loss management.
Other explanatory variables are typical for studies of loan loss provisioning (e.g., Adams, Carow, &
Perry, 2009; Bikker & Metzemakers, 2005; Fonseca & González, 2008; Laeven & Majnoni, 2003; Nichols,
Wahlen, & Wieland, 2009), and include controls for bank balance sheet and income statement variables
relevant to provisioning, as well as controls for broader bank characteristics and macroeconomic
activity. NCO/Assets is net charge-offs as a percentage of average assets from the previous quarter.
To the extent that historical loan losses proxy for contemporary loan losses and therefore the need
for current provisioning, we expect a positive relationship between net charge-offs and provisions.
NPTL/Assets captures non-performing loans, defined as the sum of loans 90+ days past due and loans in
nonaccrual status. We likewise expect it to be positively related to provisions. The change in total loans,
captured by Loans/Loans, could be negatively associated with provisions if loan growth indicates an
expansion of profitable investment opportunities, or it could be positively associated with provisions
if loan growth indicates deteriorating underwriter quality.27
Loans/Assets captures the importance of loans in a bank’s overall portfolio of assets. The percentage
change in real GDP over the previous quarter, GDP, is obtained from the Federal Reserve Bank of Saint
Louis. EQ, shareholder equity as a percentage of total assets from the previous quarter, captures bank
capital structure and the level of high quality bank capital for supervisory purposes. Size measures
bank size as the natural log of total assets (in thousands) from the previous quarter. Merger equals one
if the bank merged with another bank during the current quarter.
There is a quarterly pattern in the percentage of observations with loan loss provisioning equal
to zero. Zero could be the most appropriate provision value for a given bank in a given quarter, but
a zero provision could also indicate a bank that as a matter of policy provisions less frequently than
quarterly. Nearly 30% of first quarter observations show provisions equal to zero, and that percentage
declines steadily to 15% of fourth quarter observations. We address this pattern in two ways. First, we
perform analyses for all quarters and include quarter indicators to control for seasonality. Second, we
repeat our analyses using only observations from the fourth quarter of each year as a robustness check.
Anecdotal evidence and the sample characteristics just described suggest that a bank that provisions
less frequently than quarterly is most likely to provision in the fourth quarter.
We use panel estimation with fixed effects at the bank level. Much of the previous literature uses
random effects rather than fixed effects, but when we utilized random effects Hausman tests uniformly
indicated that the estimates may be inconsistent. (We discuss results from random effects analysis in
our robustness checks.) Laeven and Majnoni (2003) and Fonseca and González (2008) both use the
GMM estimation technique developed by Arrellano and Bond (1991) to examine dynamic models of
loan loss provisioning, but the size of our dataset (two orders of magnitude greater than those authors’
datasets) makes that technique computationally intractable for our long term sample. We are able to
use this technique for our short term sample with a limited number of instruments, also discussed in
our robustness checks. Laeven and Majnoni (2003) present results from both panel estimation with
random effects and Arellano–Bond estimation, with similar results across the two models.28 They also
find evidence of second-order correlation in the first-differenced errors for their U.S. bank sample, sug-
gesting that Arellano–Bond estimates for that sample may be biased. In our Arellano–Bond estimations
for our short term sample, we also consistently found evidence of second-order correlation.
5. Empirical analysis
5.1. Bivariate analysis
Panel A of Table 2 presents correlation coefficients between PLL/Assets and PPNR for all publicly-
held and privately-held banks before and after the SEC action. In each case, the correlation coefficient
27
See Keeton (1999) and Foos et al. (2010).
28
Specifically, in their pooled international sample, all of their explanatory variables have the same signs and levels of signif-
icance across the two models. In their sample of U.S. banks only, their coefficient estimate for loan growth changes sign across
models, but all other explanatory variables retain their signs and levels of significance. Our sample contains only U.S. banks.
Table 2
Pairwise correlation coefficients between provisions for loan losses and pre-provision net revenues by bank type and
sample period. The pre-SunTrust and post-SunTrust sample periods for the short term sample are 1996Q4–1998Q3
and 1999Q1–2000Q4, respectively. The pre-SunTrust and post-SunTrust sample periods for the long term sample are
1992Q1–1998Q3 and 1999Q1–2013Q4, respectively. Variables are as defined in Table 1.
Short term sample Long term sample
Publicly-held banks Privately-held banks Publicly-held banks Privately-held banks
Panel A: Correlation of PLL/Assets and PPNR – all banks

Pre-SunTrust 0.430* 0.182* 0.337* 0.158*
Post-SunTrust 0.352* 0.198* 0.218* 0.057*
Panel B: Correlation of PLL/Assets and PPNR – banks under $10 billion in assets
Pre-SunTrust 0.396* 0.182* 0.329* 0.158*
Post-SunTrust 0.284* 0.189* 0.100* 0.043*
Panel C: Correlation of PLL/Assets and PPNR – banks over $10 billion in assets
Pre-SunTrust 0.829* 0.157 0.616* 0.536*
Post-SunTrust 0.729* 0.957* 0.611* 0.728*
*
Indicates correlation coefficients significant at the 0.01% level.
is approximately two to three times higher for publicly-held banks than for privately-held banks,
suggesting greater earnings management on the part of publicly-held banks. For the short term sam-
ple, the correlation between provisions and earnings was lower for publicly-held banks in the eight
quarters following the SEC action compared to the eight quarters before, while the correlation coef-
ficient for privately-held banks rose slightly. These results are consistent with Hypothesis 1a in that
the relationship between earnings and provisions appears to have weakened for publicly-held banks
relative to privately-held banks. Given the small change in correlation coefficients for privately-held
banks, the results are reasonably consistent with Hypothesis 1b as well. For the long term sample, the
correlation coefficients fall substantially for both types of banks after the SEC action, consistent with
Hypothesis 2.
The pattern of results in Panel B, which shows correlation coefficients for small banks, is similar to
that in Panel A. The short term sample results in Panel B are even more consistent with Hypotheses
1a and 1b, in that the coefficient for publicly-held banks falls more sharply and the coefficient for
publicly-held banks is more stable than is the case in Panel A. Panel B also shows steeper falls for the
long term sample, especially for publicly-held banks.
The coefficients for large banks, presented in Panel C, are quite different from those on the previous
panels. In both the short term and long term samples, the correlation coefficients fall slightly after the
SEC action for publicly-held banks, but rise dramatically for privately-held banks. In some sense this
is consistent with Hypothesis 1a (the relationship between provisions and earnings become relatively
weaker for publicly-held banks, but primarily due to the relationship becoming stronger for privately-
held banks), but more broadly this pattern does not fit with the hypotheses. Differences in the results
between small and large banks suggest that the strengthening of accounting constraints may have
affected large banks and small banks differently.
5.2. Multivariate analysis of provisions over the short term
Table 3 presents results from panel estimations using the short term sample in which the baseline
specification is:
PLL/Assetsit = ˛ + ˇ1 Publicit + ˇ2 AfterSTit + ˇ3 PPNRit + ˇ4 NCO/Assetsit
+ ˇ5 NPTL/Assetsit + ˇ6 Loans/Loansit + ˇ7 Loans/Assetsit + ˇ8 GDPit + ˇ9 EQit
+ ˇ10 Sizeit + ˇ11 Mergerit + q Quarterq + i Fixedi + εit

Table 3
Panel regressions of provision for loan losses scaled by average assets (PLL/Assets) using the short term sample. The sample
period is 1996Q4–2000Q4, with 1998Q4 (the quarter in which the SEC action occurred) excluded. Variables are as defined in
Table 1. Specifications include fixed effects by bank, robust standard errors, quarter dummies, and a constant term. T-statistics
appear in brackets.
(1) (2) (3) Both (4) Both (5) Both (6) Both (7) Both
Publicly-held Privately-held types types types types types
banks banks
Public 0.0401** 0.0557*** 0.0763** 0.0405** 0.0565

[2.408] [3.132] [2.153] [2.438] [1.192]
AfterST 0.0714** −0.0144 −0.0152*** −0.0129** −0.0145*** 0.00391 −0.00783
[2.528] [−0.902] [−3.042] [−2.553] [−2.820] [0.285] [−0.498]
PPNR 0.0424** 0.0340*** 0.0267*** 0.0272*** 0.0315*** 0.0320*** 0.0332***
[2.451] [3.322] [3.158] [3.203] [3.154] [3.602] [3.380]
Public*AfterST −0.0243*** 0.0332
[−3.173] [0.937]
PPNR*Public −0.0200 −0.00432
[−1.154] [−0.193]
PPNR*AfterST −0.0364*** −0.00201 −0.0102 −0.00224
[−2.889] [−0.231] [−1.427] [−0.257]
PPNR*Public −0.0260
*AfterST [−1.533]
NCO/Assets 0.359*** 0.307*** 0.314*** 0.314*** 0.315*** 0.314*** 0.314***
[6.567] [19.25] [20.06] [20.07] [20.07] [20.10] [20.22]
NPTL/Assets 0.126*** 0.0682*** 0.0718*** 0.0718*** 0.0721*** 0.0720*** 0.0722***
[6.087] [12.17] [13.19] [13.19] [13.25] [13.20] [13.24]
Loans/Loans −0.00158** −0.00252*** −0.00249*** −0.00247*** −0.00245*** −0.00246*** −0.00240***
[−2.167] [−9.488] [−9.769] [−9.755] [−9.141] [−9.829] [−9.262]
Loans/Assets 0.00523*** 0.00542*** 0.00564*** 0.00562*** 0.00563*** 0.00561*** 0.00555***
[4.638] [10.50] [11.66] [11.64] [11.44] [11.80] [11.57]
GDP −0.00167 2.10e−05 −0.00121 −0.00103 −0.00124 −0.00117 −0.00105
[−0.436] [0.0125] [−0.784] [−0.666] [−0.798] [−0.753] [−0.674]
EQ −0.00843*** −0.0132*** −0.0125*** −0.0124*** −0.0126*** −0.0123*** −0.0124***
[−2.831] [−4.641] [−6.036] [−5.989] [−6.049] [−5.992] [−5.985]
Size 0.0254 0.110*** 0.0714*** 0.0751*** 0.0707*** 0.0731*** 0.0750***
[1.318] [5.268] [4.226] [4.402] [4.221] [4.374] [4.469]
Merger 0.00193 0.0236 0.0226** 0.0222** 0.0238** 0.0221** 0.0224**
[0.170] [1.555] [2.216] [2.179] [2.310] [2.178] [2.161]
Observations 18,422 110,761 129,183 129,183 129,183 129,183 129,183

Banks 2013 7860 9440 9440 9440 9440 9440
R2 0.516 0.253 0.294 0.289 0.295 0.293 0.290
*
Indicates level of significance at 10%.
**
***
where i = bank identifier; t = time identifier (observations are quarterly); Quarter = quarter indicator,
with q taking values 1–3 (the fourth quarter is the omitted category); Fixed = fixed effect at the bank
level; ε = error term; and other variables defined as in Table 1.
Models 1 and 2 separate the short term sample into publicly-held and privately-held banks, respec-
tively, and add the interaction term PPNR*AfterST to the baseline specification. Hypothesis 1a predicts
that the coefficient estimate PPNR*AfterST should be negative for publicly-held banks, and should be
lower for publicly-held banks than for privately-held banks. Hypothesis 1b predicts that the coefficient
estimate PPNR*AfterST should not be significant for privately-held banks. Those predictions are borne
out. PPNR is positive and significant in both models (and in subsequent models), indicating that both
types of bank managed earnings through loan loss provisioning prior to the SEC action. The coefficient
estimate PPNR*AfterST is negative and significant in model 1, while in model 2 it is not significantly
different from zero, either statistically or economically. The sum of the coefficient estimates for PPNR
and PPNR*AfterST in model 1 is not significantly different from zero, suggesting that following the SEC
action publicly-held banks no longer managed earnings. The sum of those estimates is significant for
model 2, suggesting that privately-held banks managed earnings in the eight quarters before and after
the SEC action. AfterST is positive and significant in model 1 but is not significant in model 2, indicating
that the average level of provisions increased for publicly-held banks following the SEC action, but did
not significantly change for privately-held banks. The signs of the estimates for the other explanatory
variables here and in subsequent tables are consistent with expectations.
Models 3–7 pool all observations from the short term sample, with models 4–7 differing from
the base specification in the presence of the three possible bilateral interaction terms among Public,
AfterST, and PPNR, and the interaction of all three variables. In model 7, Hypotheses 1a predicts that
the triple interaction term PPNR*Public*AfterST is negative and significant. This hypothesis is only
weakly supported, with the coefficient estimate being not quite significant at conventional levels,
with p = 0.125. Hypothesis 1b predicts that PPNR*AfterST is not significant, which the results show.
The sum of the coefficient estimates for PPNR and the three interaction terms that include PPNR is not
significantly different from zero, consistent with the finding from model 1 that publicly-held banks do
not appear to have managed earnings following the SEC action. The sum of the coefficient estimates
for PPNR and PPNR*AfterST is positive and significant, suggesting that privately-held banks continued
to manage earnings after the SEC action, as found in model 2.
Table 4 compares specifications similar to models 1, 2, and 9 from Table 3 for small banks (mod-
els 1–3 of Table 4) and large banks (models 4–6 of Table 4). Unsurprisingly, given that small banks
comprise 99% of our total sample, the results for small banks are quite similar to those in Table 3. The
main difference between the results for small banks and for all banks is that for small banks the triple
interaction term PPNR*Public*AfterST has a slightly lower level of significance (p = 0.144). Also of note
is that while Public is not significant in model 9 of Table 3, for small banks it is positive and significant,
indicating that on average during the sample period, small publicly-held banks had higher provisions
than small privately-held banks.
The large bank results from models 4–6 show both similarities and differences with respect to the
results for small banks. In models 4 and 5, the coefficient estimate for PPNR*AfterST is negative for
publicly-held banks and is lower for publicly-held banks than for privately-held banks as Hypothesis
1a predicts, but neither estimate is statistically significant. The lack of significance of PPNR*AfterST for
privately-held banks is consistent with Hypothesis 1b. As in model 1 for small publicly-held banks, the
sum of the coefficient estimates of PPNR and PPNR*AfterST in model 4 is not significantly different from
zero, suggesting a lack of earnings management for large publicly-held banks following the SEC action.
For large privately-held banks, PPNR itself is not significant, nor is the sum of the coefficient estimates
of PPNR and PPNR*AfterST, implying no evidence of earnings management for large privately-held
banks either before or after the SEC action.29
Model 6 of Table 4 pools all large banks, but Public dropped out of the model due to collinearity as
a result of the lack of variation in ownership type in large banks during the short term sample period.
The triple interaction term is negative and significant as predicted by Hypothesis 1a. PPNR*AfterST is
positive and significant, contradicting Hypothesis 1b and indicating an increase in earnings manage-
ment among large privately-held banks following the SEC action. The sum of the coefficient estimates
of PPNR*AfterST and the triple interaction term are not significantly different from zero, indicating that
large publicly-held banks showed no change in earnings management after the SEC action.
The final model in Table 4 examines the very largest banks, those over $100 billion in real assets,
which due to their size and complexity may behave differently than even most other banks over $10
billion.30 We cannot address differences in reactions to the SEC action across ownership types for the
largest banks because they are all publicly-held throughout the short term sample, but nevertheless
provisioning practices for these banks before and after the SEC action are of interest. The negative and
significant coefficient estimate for PPNR suggests that prior to the SEC action the largest banks provi-
sioned procyclically, rather than managing earnings through countercyclical provisioning. Following
the SEC action, the largest banks appear to have stopped provisioning procyclically, as indicated by
the sum of the coefficient estimates for PPNR and PPNR*AfterST not differing significantly from zero.
29
An important caveat to model 5 is that the sample size is particularly small, greatly reducing the precision of the estimates.
30
Removing banks with over $100 billion in real assets does not substantively change any of the results in this paper, either
for all banks or for banks over $10 billion.
Table 4
Panel regressions of provision for loan losses scaled by average assets (PLL/Assets) for small versus large banks using the short
term sample. The sample period is 1996Q4–2000Q4, with 1998Q4 (the quarter in which the SEC action occurred) excluded.
Variables are as defined in Table 1. Specifications include fixed effects by bank, robust standard errors, quarter dummies, and a
constant term. Public and its interaction with AfterST were omitted from specifications for banks over $100 billion due to a lack
of variation in ownership type. T-statistics appear in brackets.
Banks under $10 billion in assets Banks over $10 billion in assets Banks over
$100 billion in
assets
(1) (2) (3) Both (4) (5) (6) Both (7) Both types
Publicly- Privately- types Publicly- Privately- types
held held held held
banks banks banks banks
Public 0.0925**
[2.378]
AfterST 0.0601** −0.0186 −0.0122 0.107** −0.189 −0.479** −0.245*
[2.252] [−1.255] [−0.836] [2.186] [−0.787] [−2.350] [−2.032]
PPNR 0.0262* 0.0325*** 0.0316*** 0.0451* 0.0483 −0.0123 −0.183**
[1.899] [3.222] [3.266] [1.779] [0.321] [−0.101] [−2.571]
Public*AfterST 0.0186 0.590***
[0.596] [2.823]
PPNR*Public −0.0229 0.0575
[−1.373] [0.459]
PPNR*AfterST −0.0333*** 0.000129 −0.000120 −0.0123 0.120 0.254** 0.174**
[−2.871] [0.0161] [−0.0148] [−0.667] [1.058] [2.286] [2.744]
PPNR*Public −0.0214 −0.267**
*AfterST [−1.460] [−2.372]
NCO/Assets 0.302*** 0.306*** 0.307*** 0.321*** 0.784*** 0.369*** 0.0264
[7.076] [19.17] [20.36] [7.444] [9.615] [6.526] [0.160]
NPTL/Assets 0.126*** 0.0682*** 0.0721*** 0.213** −0.234 0.190** 0.470**
[5.780] [12.16] [13.17] [2.346] [−0.849] [2.159] [2.538]
Loans/Loans −0.00095** −0.00251*** −0.0023*** 0.00048 0.00798 0.00145 0.00425*
[−2.006] [−9.427] [−9.691] [0.450] [1.795] [1.256] [1.984]
Loans/Assets 0.00413*** 0.00542*** 0.00537*** 0.00377* −0.00330 0.00236 −0.00105
[6.532] [10.49] [11.81] [1.810] [−0.464] [1.172] [−0.186]
GDP −0.000303 −2.10e−05 −0.000960 −0.0342** 0.0226* −0.0287* −0.0290
[−0.0779] [−0.0125] [−0.614] [−2.063] [2.022] [−1.907] [−1.093]
EQ −0.00784*** −0.0133*** −0.0123*** −0.0105 −0.0301 −0.0145 −0.0378
[−2.592] [−4.633] [−5.878] [−0.738] [−1.345] [−1.435] [−1.222]
Size 0.0320 0.112*** 0.0806*** 0.0222 −0.0992 0.0231 −0.0546
[1.508] [5.263] [4.704] [0.505] [−0.692] [0.528] [−0.661]
Merger 0.00349 0.0235 0.0259** 0.00448 −0.00339 0.00381 0.0201
[0.271] [1.530] [2.390] [0.256] [−0.0702] [0.230] [0.407]
Observations 17,587 110,671 128,258 835 90 925 87

Banks 1963 7856 9386 79 9 88 8
R2 0.450 0.248 0.275 0.813 0.949 0.778 0.419
*
**
***
Overall, the evidence in Tables 3 and 4 is supportive of Hypothesis 1a for all banks, small banks,
and large banks. The relationship between earnings and provisions does appear to have weakened
for publicly-held banks relative to privately-held banks in the short term following the SEC action,
although some of the coefficient estimates on which that conclusion is based do not quite achieve
conventional levels of significance. The evidence is strongly supportive of Hypothesis 1b for all banks
and small banks. For these groups, the relationship between earnings and provisions for privately-held
banks appears unaffected in the short term following the SEC action. The same is not true for large
privately-held banks, for which earnings management appears to have increased following the SEC
action.
Table 5
Panel regressions of provision for loan losses scaled by average assets (PLL/Assets) using the long term sample. The sample
period is 1992Q1–2013Q4, with 1998Q4 (the quarter in which the SEC action occurred) excluded. Variables are as defined in
Table 1. Specifications include fixed effects by bank, robust standard errors, quarter dummies, and a constant term. T-statistics
appear in brackets.
(1) (2) (3) Both (4) Both (5) Both (6) Both (7) Both
Publicly- Privately- types types types types types
held held
banks banks
Public 0.0125 0.00938 0.0793*** 0.0131* 0.0447**

[1.573] [1.046] [5.150] [1.647] [2.168]
AfterST 0.0516** −0.00141 −0.0248*** −0.0251*** −0.0224*** −0.00134 −0.00624
[2.331] [−0.165] [−8.885] [−8.892] [−7.814] [−0.178] [−0.758]
PPNR 0.0427*** 0.0491*** 0.0380*** 0.0380*** 0.0446*** 0.0465*** 0.0504***
[4.738] [11.44] [12.08] [12.01] [13.17] [12.16] [12.33]
Public*AfterST 0.00482 0.0557**
[0.715] [2.340]
PPNR*Public −0.0403*** −0.0257**
[−4.900] [−2.473]
PPNR*AfterST −0.0271*** −0.0114** −0.0129*** −0.00892*
[−2.711] [−2.313] [−3.009] [−1.887]
PPNR*Public −0.0241**
*AfterST [−2.005]
NCO/Assets 0.493*** 0.411*** 0.431*** 0.431*** 0.431*** 0.431*** 0.430***
[30.11] [64.75] [71.98] [71.96] [72.17] [71.76] [72.18]
NPTL/Assets 0.168*** 0.0943*** 0.100*** 0.100*** 0.101*** 0.100*** 0.100***
[21.02] [49.64] [53.49] [53.49] [53.66] [53.43] [53.62]
Loans/Loans −0.00147*** −0.00206*** −0.00195*** −0.00195*** −0.00190*** −0.00194*** −0.00189***
[−4.821] [−14.66] [−15.36] [−15.36] [−14.87] [−15.35] [−14.86]
Loans/Assets 0.00472*** 0.00363*** 0.00376*** 0.00376*** 0.00373*** 0.00374*** 0.00372***
[9.525] [30.56] [31.60] [31.67] [31.03] [31.93] [31.44]
GDP −0.0774*** −0.0272*** −0.0334*** −0.0335*** −0.0336*** −0.0331*** −0.0334***
[−14.69] [−20.50] [−25.09] [−25.07] [−25.21] [−24.73] [−24.86]
EQ −0.0103*** −0.0106*** −0.00969*** −0.00969*** −0.00988*** −0.00964*** −0.00987***
[−4.739] [−13.95] [−14.01] [−14.02] [−14.21] [−13.96] [−14.17]
Size 0.00326 0.0413*** 0.0366*** 0.0364*** 0.0364*** 0.0366*** 0.0361***
[0.352] [13.17] [12.17] [12.06] [12.07] [12.23] [12.05]
Merger −0.00322 0.0175** 0.0140** 0.0140** 0.0172*** 0.0139** 0.0170***
[−0.341] [2.314] [2.328] [2.329] [2.835] [2.312] [2.803]
Observations 81,284 567,084 648,368 648,368 648,368 648,368 648,368

Banks 3831 11,701 13,996 13,996 13,996 13,996 13,996
R2 0.552 0.393 0.425 0.425 0.422 0.425 0.423
*
**
***
5.3. Multivariate analysis of provisions over the long term
Table 5 presents results from specifications identical to those in Table 3, but here based on the
long term sample. PPNR*AfterST is negative and significant in all of the models in which it appears,
indicating that, over the long term, earnings management weakened for both publicly-held banks
and privately-held banks following the SEC action. This is consistent with Hypothesis 2. The sum
of the coefficient estimates of PPNR and PPNR*AfterST is positive and significant in models 1 and 2,
indicating that although earnings management lessened for both publicly-held and privately-held
banks following the SEC action, a positive relationship between earnings and provisioning persisted.
In model 9, the triple interaction term is negative and significant, indicating that while the relationship
between earnings and provisioning weakened for both types of banks, it weakened more for publicly-
held banks than for privately-held banks. The sum of the coefficient estimates of PPNR and PPNR*AfterST
is once again positive and significant, suggesting that in the long term privately-held banks continued
Table 6
Panel regressions of provision for loan losses scaled by average assets (PLL/Assets) for small versus large banks using the long
term sample. The sample period is 1992Q1–2013Q4, with 1998Q4 (the quarter in which the SEC action occurred) excluded.
Variables are as defined in Table 1. Specifications include fixed effects by bank, robust standard errors, quarter dummies, and a
constant term. Public and its interaction with AfterST were omitted from specifications for banks over $100 billion due to a lack
of variation in ownership type. T-statistics appear in brackets.
Banks under $10 billion in assets Banks over $10 billion in assets Banks over $100
billion in assets
(1) (2) (3) Both (4) (5) (6) Both (7) Both types
Publicly- Privately- types Publicly- Privately- types
held held held held
banks banks banks banks
Public 0.0542** 0.912**

[2.301] [2.372]
AfterST 0.0672** −0.00154 −0.00586 0.132** 0.666** 0.925** −0.357*
[2.250] [−0.181] [−0.710] [2.179] [2.615] [2.499] [−1.777]
PPNR 0.0411*** 0.0487*** 0.0492*** 0.0614** 0.388*** 0.447*** −0.133**
[3.668] [11.37] [12.13] [2.278] [5.608] [5.240] [−2.382]
Public*AfterST 0.0606** −0.800**
[2.006] [−2.137]
PPNR*Public −0.0285** −0.395***
[−2.397] [−4.594]
PPNR*AfterST −0.0384*** −0.0117** −0.00960** −0.00736 −0.354*** −0.351*** 0.313***
[−2.714] [−2.404] [−2.036] [−0.362] [−4.846] [−4.082] [3.488]
PPNR*Public −0.0297* 0.341***
*AfterST [−1.922] [3.901]
NCO/Assets 0.476*** 0.410*** 0.425*** 0.541*** 0.473*** 0.544*** 0.268
[28.22] [66.79] [73.43] [8.079] [3.357] [8.446] [1.386]
NPTL/Assets 0.170*** 0.0943*** 0.0998*** 0.188*** 0.140** 0.182*** 0.261***
[21.12] [49.68] [53.47] [5.309] [2.635] [5.807] [3.770]
Loans/Loans −0.00160*** −0.00206*** −0.0019*** 0.000310 0.000746 0.000820 −0.00157
[−5.421] [−14.68] [−15.23] [0.263] [0.331] [0.801] [−0.516]
Loans/Assets 0.00479*** 0.00363*** 0.00370*** 0.00381** 0.0116** 0.00403** 0.00483
[9.366] [30.43] [31.59] [2.282] [2.495] [2.377] [0.738]
GDP −0.0646*** −0.0268*** −0.0313*** −0.219*** −0.220*** −0.222*** −0.282***
[−12.45] [−20.28] [−23.67] [−9.124] [−4.461] [−10.08] [−7.168]
EQ −0.0106*** −0.0108*** −0.0102*** −0.0118 −0.0337** −0.0144 0.0306
[−4.672] [−15.63] [−15.17] [−0.981] [−2.332] [−1.483] [0.838]
Size 0.00523 0.0424*** 0.0384*** −0.0472 0.218** −0.0273 −0.142*
[0.503] [13.72] [12.77] [−1.384] [2.651] [−0.938] [−1.822]
Merger −0.00670 0.0171** 0.0176*** 0.0259 0.0144 0.0212 0.0335
[−0.648] [2.260] [2.830] [1.345] [0.194] [1.140] [0.687]
Observations 76,783 566,213 642,996 4501 871 5372 816

Banks 3788 11,675 13,943 169 57 210 34
R2 0.517 0.389 0.409 0.733 0.681 0.732 0.617
*
**
***
to manage earnings through provisioning after the SEC action, consistent with the conclusion from
model 2. For publicly-held firms, the results indicate that earnings management was present prior
to the SEC action (the sum of the estimates of PPNR and PPNR*Public is positive and significant), but
following the SEC action there is no evidence of earnings management (the sum of the estimates
of PPNR and the three interaction terms that include PPNR is not significantly different from zero),
contrasting with the result from model 1.
In Table 6, the models using only small banks have results very similar to those in Table 5 in terms
of the signs and significance of the key variables, again consistent with Hypothesis 2. Two notable
differences do arise concerning the sums of coefficient estimates of PPNR and its interactions. First,
in model 1, the sum for PPNR and PPNR*AfterST is not significantly different from zero. This indicates
that while both types of small banks appear to have managed earnings prior to the SEC action, there
is no evidence of earnings management among small publicly-held banks after the action. Second, in
model 3, the sum of the coefficient estimates for PPNR and the three interaction terms that include
PPNR is negative and significant, suggesting that small publicly-held banks went from countercyclical
provisioning before the SEC action (the sum for PPNR and PPNR*Public is positive and significant) to
procyclical provisioning after the action.
Turning to the results for large banks, the results from model 4 indicate that large publicly-held
banks exhibit no significant change in the relationship between earnings and provisioning following
the SEC action, contradicting Hypothesis 2. The results for large privately-held banks are consistent
with Hypothesis 2 with PPNR*AfterST being negative and significant. The sum of the estimates for
PPNR and PPNR*AfterST is not significantly different from zero, suggesting that large privately-held
banks managed earnings prior to the SEC action but ceased doing so after. The results for all large
banks pooled in model 6 are mostly consistent with those for models 4 and 5, with Hypothesis 2 being
supported for large privately-held banks but not for large publicly-held banks, only for model 6 tests
on sums of coefficient estimates indicate that both types of large banks provisioned countercyclically
following the SEC action.
Model 7 in Table 6 examines the largest banks using the long term sample. The results are con-
sistent with those found for the largest banks using the short term sample. These banks appear to
have provisioned procyclically prior to the SEC action (PPNR is negative), but after the SEC action the
relationship between earnings and provisioning (measured with the sum of PPNR and PPNR*AfterST)
is not significant.
Overall, the long term sample evidence in Tables 5 and 6 are supportive of Hypothesis 2 for all
banks and for small banks. For those groups, PPNR*AfterST is negative in every model regardless of
bank ownership type. The results for those groups also indicate that while earnings management
declined for both types of banks after the SEC action, it declined more for publicly-held banks than for
privately-held banks. For large banks, earnings management declined for privately-held banks only,
and did not change for publicly-held banks.
5.4. Multivariate analysis of reserves over the short and long term
Table 7 presents results from specifications with LLR/Assets as the dependent variable using the
short term sample and the long term sample for all banks and banks split by size. Because changes
in levels of loan loss reserves are the focus here rather than changes in the relationship between
reserves and earnings, there are no PPNR interactions in the specifications. Hypothesis 3 predicts
that Public*AfterST is negative. This prediction is borne out for all banks, and for small banks the
coefficient estimate for Public*AfterST is negative for both samples but only significant for the short
term sample. For large banks, Public*AfterST is not significant for either the short term or long term
sample, contradicting Hypothesis 3 for this group.
5.5. Robustness checks
We perform several additional analyses to check the robustness of our findings. As described below,
results concerning our hypotheses are mostly unchanged in these robustness checks. (Complete results
are available upon request.) As noted above, the reported results are based on panel regressions with
fixed effects, even though much of the previous literature uses random effects. When we employ
random effects instead, PPNR*Public*AfterST changes from not significant to significant in Table 3 and
in Table 4 for small banks, strengthening the support for Hypothesis 1a. In addition, Public*AfterST
becomes negative and significant for small banks using the long term sample in Table 7, strengthening
the support for Hypothesis 3. However, Hausman tests indicated potential inconsistency in the random
effects estimates.
Also noted above, Laeven and Majnoni (2003) and Fonseca and González (2008) use the GMM
estimation technique developed by Arrellano and Bond (1991) to examine provisioning. Due to com-
puting capacity limits, we were only able to use this technique for our short term sample, and only
Table 7
Panel regressions of loan loss reserves scaled by average assets (LLR/Assets) using the short term sample and the long term sample.
The sample period in models 1–4 is 1996Q4–2000Q4, with 1998Q4 (the quarter in which the SEC action occurred) excluded.
The sample period in models 5–8 is 1992Q1–2013Q4, with 1998Q4 excluded. Variables are as defined in Table 1. Specifications
include fixed effects by bank, robust standard errors, quarter dummies, and a constant term. Public and its interaction with
AfterST were omitted from specifications for banks over $100 billion due to a lack of variation in ownership type. T-statistics
appear in brackets.
(1) All (2) Banks (3) Banks (4) Banks (5) All (6) Banks (7) Banks (8) Banks
banks under $10 over $10 over $100 banks under $10 over $10 over $100
billion in billion in billion in billion in billion in billion in
assets assets assets assets assets assets
Public 0.135*** 0.132*** 0.0930*** 0.0741*** 0.130

[6.538] [6.394] [6.206] [5.278] [0.485]
AfterST 0.0126** 0.0127** −0.0953* −0.105 −0.0172*** −0.0179*** −0.375 −0.227*
[2.255] [2.192] [−1.887] [−1.400] [−3.631] [−3.777] [−1.600] [−2.030]
PPNR 0.0138** 0.0113 0.0230 −0.0528 −0.00447 −0.00627* 0.0272** 0.0741***
[2.035] [1.638] [1.309] [−1.821] [−1.340] [−1.872] [2.397] [3.403]
Public*AfterST −0.0203** −0.0166** 0.0651 −0.0375*** −0.0161 0.132
[−2.411] [−2.036] [1.048] [−2.907] [−1.397] [0.541]
NCO/Assets 0.0368*** 0.0305*** 0.113** −0.165 0.128*** 0.118*** 0.479*** 0.443***
[2.990] [2.592] [2.296] [−0.695] [22.55] [21.70] [9.437] [11.18]
NPTL/Assets 0.0707*** 0.0705*** 0.251*** 0.509 0.139*** 0.138*** 0.188*** 0.193***
[15.63] [15.56] [4.356] [1.408] [60.74] [60.42] [6.750] [6.190]
Loans/Loans −0.0019*** −0.00196*** 0.000829 0.000922 −0.0032*** −0.00328*** 0.000727 0.00252*
[−8.465] [−9.345] [0.637] [0.292] [−22.83] [−23.89] [0.693] [1.739]
Loans/Assets 0.00635*** 0.00618*** 0.00783*** 0.0164* 0.00551*** 0.00542*** 0.00876*** 0.00721*
[12.56] [12.69] [3.767] [2.147] [28.67] [28.26] [4.637] [1.756]
GDP −0.00278** −0.00294** −0.00916 0.0569*** 0.0171*** 0.0177*** −0.0223** −0.0517**
[−2.242] [−2.338] [−0.837] [4.207] [16.79] [17.40] [−2.092] [−2.084]
EQ 0.00917*** 0.00921*** 0.00913 −0.0300 0.00648*** 0.00646*** 0.00624 0.00946
[4.200] [4.168] [1.021] [−0.747] [5.620] [5.560] [0.992] [0.657]
Size −0.110*** −0.110*** −0.0812 −0.0188 −0.0349*** −0.0324*** −0.0629 −0.0830
[−4.549] [−4.292] [−1.586] [−0.129] [−6.710] [−6.161] [−1.076] [−1.404]
Merger 0.137*** 0.149*** 0.0327* −0.00428 0.135*** 0.148*** −0.0224 −0.0330*
[13.89] [14.30] [1.899] [−0.171] [19.29] [21.78] [−0.909] [−1.832]
Observations 129,183 128,258 925 87 648,368 642,996 5372 816

Banks 9440 9386 88 8 13,996 13,943 210 34
R2 0.103 0.110 0.546 0.871 0.291 0.286 0.729 0.847
*
**
***
with a limited number of lagged variables as instruments.31 In each of these analyses, the only key
variable that was statistically significant was PPNR. Also in each of these analyses, we found evidence
of second-order correlation in the first-differenced errors, suggesting bias in the estimates.
Among previous studies of provisions, some scale provisions and lending-related explanatory vari-
ables by assets (e.g., Bikker & Metzemakers, 2005; Fonseca & González, 2008; Laeven & Majnoni, 2003)
and others scale by loans (e.g., Adams et al., 2009; Beatty et al., 2002; Nichols et al., 2009). When we
scale our variables by loans rather than assets, the triple interaction term is further from significance
in Table 3 and for small banks in Table 4, weakening the support for Hypothesis 1a. Scaling by loans
did not substantively affect any results for large banks.
Given that banks that provision less frequently than quarterly are most likely to provision in the
fourth quarter, we repeated our analyses using only fourth quarter observations. This had no sub-
stantive effect on the results for all banks. For small banks, the coefficient estimate for Public*AfterST
31
Specifically, completed attempts included two to eight lags of PLL/Assets and the same or smaller numbers of lags of PPNR
and its interaction terms as instruments.
in model 2 was essentially unchanged (going from −0.0166 to −0.0176), but it slipped below con-
ventional levels of significance (p = 0.108), plausibly due to the loss of three-fourths of the total
observations. For large banks, the only changes occurred in Table 4. In model 6, both PPNR*AfterST and
the triple interaction term lose significance, simultaneously weakening the support for Hypothesis 1a
and strengthening the support for Hypothesis 1b.
While in reported results we omit observations from the fourth quarter of 1998 (the quarter of the
SEC action), we also repeat our analyses after omitting observations for 1999 as well on the premise
that banks may have required multiple quarters to adjust their provisioning policies following the
SEC action. For the short term sample, this entails dropping one fourth of the total observations. In
Table 7, Public*AfterST loses significance in models 1 and 2, weakening the support for Hypothesis
3 for all banks and small banks. For large banks, PPNR*AfterST and the triple interaction term lose
significance in model 6 of Table 4, once again simultaneously weakening the support for Hypothesis
1a and strengthening the support for Hypothesis 1b.
Our long term sample period begins in the first quarter of 1992, when the current laws regulating
how loan loss reserves enter into bank capital first went into effect. Banks may have required con-
siderable time to fully adjust their levels of loan loss reserves to fit the new requirements, and this
adjustment would be reflected in their provisioning. To ensure that this potential adjustment period
does not influence our results, we repeat our long term analyses after omitting observations from
1992 and 1993.32 (Note that this does not affect our short term analyses.) PPNR*AfterST is no longer
negative and significant for all banks in Table 5 or for large banks in Table 6. Results for Table 7 are
unchanged.
To help control for the systematic difference in sizes of publicly-traded banks versus privately-
traded banks (most of the largest banks are publicly-held and most of the smallest banks are privately-
held), in each quarter Kwan (2004) drops observations for publicly-held banks with assets greater than
the assets of the largest privately-held bank, and observations for privately-held banks with assets less
than the assets of the smallest publicly-held bank. There are no substantive changes to our results when
we employ this condition to our analyses.
Some previous analyses of provisions (e.g., Adams et al., 2009; Beatty et al., 2002; Nichols et al.,
2009) control for the level of loan loss reserves from the previous period, which could plausibly affect
the subsequent magnitudes of loan loss provisions. Including lagged loan loss reserves in the specifi-
cations strengthens the support for Hypothesis 1a as the triple interaction term becomes statistically
significant, but weakens the support for Hypothesis 3 as Public*AfterST loses significance for all banks
and for small banks.
6. Extensions
In this subsection we present analyses that do not directly evaluate the paper’s hypotheses, but are
nonetheless germane to the relationship between bank provisioning and earnings management. First,
we examine how the relationship between earnings and provisioning changed during and after the
recent financial crisis. Second, we examine whether the strengthening of the accounting constraint
associated with the SEC action coincided with any changes in the relationship between earnings and
realized security gains, another potential mechanism for earnings management. Neither analysis is
comprehensive, but rather is intended to highlight potentially fruitful topics for related research.
Models 1–4 of Table 8 split the long term sample into four periods: pre-SunTrust, which runs
from the start of the long term sample to the quarter before the SEC action (1992Q1–1998Q3); pre-
recession, which runs from the quarter after the SEC action to the start of the recent financial crisis
(1991Q1–2007Q3); recession, which includes the quarters of the recession as identified by the NBER
(2007Q4–2009Q2); and post-recession, which runs from the end of the recession to the end of the
dataset (2009Q3–2013Q4). Relative to the baseline specification, these specifications drop AfterST of
necessity, and include PPNR*Public to identify differences in the relationship between earnings and
32
FAS 114, which implemented key elements of the incurred-loss approach, dates from March 1993. See the Appendix for
details. This robustness check therefore also eliminates observations from before this change in accounting rules.
Table 8
Panel regressions of provision for loan losses scaled by average assets (PLL/Assets) using the long term sample. The sample
period is 1992Q1–2013Q4, with 1998Q4 (the quarter in which the SEC action occurred) excluded. Pre-recession equals 1 for
1999Q1–2007Q3, 0 otherwise. Recession equals 1 for 2007Q4–2009Q2, 0 otherwise. Post-recession equals 1 for 2009Q3–2013Q4,
0 otherwise. Variables are as defined in Table 1. Specifications include fixed effects by bank, robust standard errors, quarter
dummies, and a constant term. T-statistics appear in brackets.
(1) Pre- (2) Pre- (3) (4) Post- (5) Full

SunTrust recession Recession recession sample
Public 0.0504** 0.0264 0.0802 0.322*** 0.0608***

[2.049] [0.999] [0.856] [3.945] [3.901]
Pre-recession −0.0169***
[−5.940]
Recession 0.00728
[1.532]
Post-recession 0.00277
[0.566]
PPNR 0.0428*** 0.0474*** −0.00119 0.0337*** 0.0458***
[8.273] [9.108] [−0.0890] [3.283] [13.20]
Public*Pre-recession −0.0182***
[−2.744]
Public*Recession 0.212***
[11.17]
Public*Post-recession 0.0505***
[3.526]
PPNR*Public −0.0235* −0.00126 −0.0820** −0.0588** −0.0332***
[−1.932] [−0.0871] [−2.130] [−2.004] [−4.182]
PPNR*Pre-recession −2.52e−08
[−0.473]
PPNR*Recession 6.04e−07***
[6.763]
PPNR*Post-recession −6.09e−08
[−0.941]
PPNR*Public 2.95e−08
*Pre-recession [0.551]
PPNR*Public −5.62e−07***
*Recession [−6.201]
*Post-recession [0.777]
NCO/Assets 0.341*** 0.368*** 0.304*** 0.375*** 0.429***
[39.80] [31.80] [17.95] [39.57] [71.14]
NPTL/Assets 0.0815*** 0.0804*** 0.163*** 0.113*** 0.0985***
[25.66] [24.46] [21.99] [24.22] [52.04]
Loans/Loans −0.00214*** −0.00147*** −0.00292*** −0.00760*** −0.00190***
[−13.45] [−8.012] [−5.464] [−13.94] [−14.84]
Loans/Assets 0.00408*** 0.00365*** 0.00120 0.0158*** 0.00373***
[18.11] [18.75] [1.232] [21.44] [31.26]
GDP 0.00464*** −0.00648*** −0.0649*** −0.0549*** −0.0256***
[2.640] [−5.682] [−16.85] [−14.61] [−22.30]
EQ −0.0170*** −0.0120*** −0.0241*** −0.0400*** −0.0105***
[−11.50] [−10.87] [−5.455] [−9.113] [−14.81]
Size 0.00922 −0.0123** 0.130** 0.152*** 0.0234***
[1.122] [−2.488] [2.280] [4.239] [6.703]
Merger 0.0392*** 0.00494 0.175*** 0.0899*** 0.0175***
[4.717] [0.658] [5.036] [4.431] [2.912]
Observations 251,295 244,178 42,822 110,073 648,368

Banks 0.370 0.402 0.358 0.328 0.429
R2 11,750 9043 6476 7259 13,996
*
**
***
provisioning between banks of different ownership types. The results from models 1 and 2 are largely
consistent with earlier results. Both types of banks manage earnings in both periods (both PPNR and
the sum of PPNR and PPNR*Public are positive and significant). Publicly-held banks exhibit less earnings
management than privately-held banks before the SEC action, but in the post-SunTrust period there
is no significant difference. In the recession period, privately-held banks do not appear to manage
earnings, while publicly-held banks appear to provision procyclically. In the post-recession period
privately-held banks return to managing earnings, while there is no significant relationship between
earnings and provisioning for publicly-held banks (based on the sum of PPNR and PPNR*Public).
Model 5 of Table 8 uses the entire long term sample, and adds to the specification interactions of
Public with time period indicators, interactions of PPNR with those indicators, and triple interaction
terms of Public, PPNR, and the time period indicators. PPNR*Recession and PPNR*Public*Recession are
both statistically significant, but are several orders of magnitude smaller than the estimates for PPNR
and PPNR*Public, suggesting a more limited economic impact of changes in earnings management over
time than is conveyed in models 1–4.
Table 9 presents results for the same specifications as Table 8 for small banks only. The only notable
differences with Table 8 are that the sum of PPNR and PPNR*Public is not quite statistically significant
in model 1, and in model 5 PPNR*Public*Recession also loses significance.
Table 10 shows the results for large banks. The results for model 1 are similar to those from model
1 of Table 8, with both types of banks managing earnings but privately-held banks doing so more than
publicly-held banks. In the period following the SEC action, privately-held banks no longer exhibit
earnings management, but the sum of the coefficient estimates for PPNR and PPNR*Public is signif-
icantly different from zero, indicating earnings management on the part of publicly-held banks. In
models 3 and 4, neither PPNR nor its sum with PPNR*Public is significant, implying no relationship
between earnings and provisioning for either type of bank in those periods. In model 5, PPNR*Public
is not significant as it was in Table 8, but otherwise the PPNR-related variables are similar.
The second extension concerns whether changes in earnings management via mechanisms other
than provisioning can be discerned following the SEC action, when earnings management via pro-
visioning became more constrained. Previous studies find that banks realize lesser (greater) security
gains when earnings are higher (lower), and this negative relationship between realized security gains
and earnings is evidence of earnings management.33 Table 11 presents correlation coefficients similar
in nature to those in Table 2, only in Table 11 they refer to correlations of realized security gains and
earnings, where earnings is calculated as net revenue before extraordinary items prior to both realized
security gains and provisioning for loan losses.34 Panel A uses all banks, and Panels B and C use small
and large banks, respectively.
Starting with the short term sample results, Panel A indicates that before the SEC action, both
publicly-held and privately-held banks manage earnings via realized security gains, and that their
correlation coefficients were of similar magnitude. Following the SEC action, earnings management fell
(the correlation coefficients became higher) for both types of banks, but fell far more dramatically for
publicly-held banks. In the long term, there was little change in the correlation of realized security gains
and earnings for publicly-held banks following the SEC action, while earnings management appears to
have become stronger for privately-held banks. The results for small banks in Panel B generally mirror
those in Panel A, the only exception that in the long term sample earnings management appears to
have slightly increased for publicly-held banks after the SEC action.
For large banks using the short term sample, it remains the case that earnings management via
realized security gains increased for banks of both ownership types following the SEC action, but some
differences with Panels A and B are present. First, the correlation coefficient is much lower for large
privately-held banks, indicating greater earnings management relative to large publicly-held banks.
Second, the magnitude of the change in coefficients from before to after the SEC action is much more
33
For examples, see Barth et al. (1990), Moyer (1990), Beatty et al. (1995), Beatty et al. (2002), and Cornett et al. (2009).
34
We also examined correlations measuring earnings as net revenue before extraordinary items prior to realized security
gains, net revenue before extraordinary items prior to provisioning, and net revenue before extraordinary items, with similar
patterns of results.
Table 9
Panel regressions of provision for loan losses scaled by average assets (PLL/Assets) using the long term sample for banks under $10
billion in assets. The sample period is 1992Q1–2013Q4, with 1998Q4 (the quarter in which the SEC action occurred) excluded.
Pre-recession equals 1 for 1999Q1–2007Q3, 0 otherwise. Recession equals 1 for 2007Q4–2009Q2, 0 otherwise. Post-recession
equals 1 for 2009Q3–2013Q4, 0 otherwise. Variables are as defined in Table 1. Specifications include fixed effects by bank,
robust standard errors, quarter dummies, and a constant term. T-statistics appear in brackets.
(1) Pre- (2) Pre- (3) (4) Post- (5) Full

Public 0.0509** 0.0387 0.104 0.330*** 0.0564***

[2.029] [1.394] [1.039] [4.006] [3.243]
Pre-recession −0.0180***
[−6.376]
Recession −0.00135
[−0.272]
Post-recession −0.000500
[−0.0939]
PPNR 0.0424*** 0.0477*** −0.00373 0.0314*** 0.0447***
[8.196] [9.089] [−0.275] [3.037] [13.17]
Public*Pre-recession −0.00740
[−0.787]
Public*Recession 0.171***
[7.092]
Public*Post-recession 0.105***
[5.059]
PPNR*Public −0.0247** −0.00812 −0.113** −0.0739** −0.0333***
[−1.983] [−0.510] [−2.096] [−2.321] [−3.909]
PPNR*Pre-recession −6.22e−07**
[−2.040]
[3.329]
[−0.539]
*Pre-recession [0.263]
PPNR*Public −1.13e−06
*Post-recession [−1.592]
NCO/Assets 0.340*** 0.363*** 0.292*** 0.370*** 0.424***
[39.53] [31.23] [17.17] [39.22] [72.70]
NPTL/Assets 0.0812*** 0.0802*** 0.161*** 0.110*** 0.0979***
[25.43] [24.33] [21.49] [23.71] [51.85]
Loans/Loans −0.00216*** −0.00151*** −0.00308*** −0.00792*** −0.00193***
[−13.50] [−8.030] [−5.711] [−14.51] [−15.22]
Loans/Assets 0.00408*** 0.00363*** 0.00121 0.0161*** 0.00369***
[18.01] [18.60] [1.237] [21.45] [31.46]
GDP 0.00453** −0.00614*** −0.0628*** −0.0534*** −0.0249***
[2.568] [−5.395] [−16.49] [−14.11] [−21.69]
EQ −0.0169*** −0.0121*** −0.0258*** −0.0388*** −0.0106***
[−11.39] [−10.81] [−5.721] [−8.825] [−15.71]
Size 0.0120 −0.0117** 0.147*** 0.187*** 0.0292***
[1.431] [−2.462] [2.660] [5.048] [7.741]
Merger 0.0412*** 0.00627 0.205*** 0.0703*** 0.0182***
[4.811] [0.774] [5.602] [3.281] [2.941]
Observations 250,152 241,926 42,380 108,538 642,996

Banks 11,721 8973 6409 7168 13,943
R2 0.367 0.371 0.330 0.314 0.414
*
**
***
Table 10
Panel regressions of provision for loan losses scaled by average assets (PLL/Assets) using the long term sample for banks over $10
billion in assets. The sample period is 1992Q1–2013Q4, with 1998Q4 (the quarter in which the SEC action occurred) excluded.
Pre-recession equals 1 for 1999Q1–2007Q3, 0 otherwise. Recession equals 1 for 2007Q4–2009Q2, 0 otherwise. Post-recession
equals 1 for 2009Q3–2013Q4, 0 otherwise. Variables are as defined in Table 1. Specifications include fixed effects by bank,
robust standard errors, quarter dummies, and a constant term. T-statistics appear in brackets.
(1) Pre- (2) Pre- (3) (4) Post- (5) Full

Public −0.154 −0.0349 −0.0747 0.163

[−0.777] [−0.269] [−0.372] [0.636]
Pre-recession 0.0928
[0.560]
Recession 0.447*
[1.658]
Post-recession 0.122
[0.580]
PPNR 0.515*** 0.0140 0.0480 0.0412 0.0644**
[16.13] [0.934] [1.360] [0.649] [2.138]
Public*Pre-recession 0.0151
[0.0878]
Public*Recession 0.133
[0.467]
Public*Post-recession −0.162
[−0.773]
PPNR*Public −0.474*** 0.0321 −0.0575** 0.0435 −0.0283
[−11.61] [1.213] [−2.163] [0.404] [−1.282]
PPNR*Pre-recession 1.02e−07
[1.648]
[3.943]
[−0.528]
*Pre-recession [−1.612]
PPNR*Public −3.99e−07***
*Post-recession [0.235]
NCO/Assets 0.250*** 0.589*** 0.756*** 0.425*** 0.564***
[4.488] [6.101] [4.867] [4.603] [9.127]
NPTL/Assets 0.208*** 0.175* 0.385*** 0.293*** 0.220***
[5.080] [1.937] [4.709] [5.231] [5.981]
Loans/Loans 0.000396 0.000435 0.000910 −0.000856 0.00145
[0.333] [0.556] [0.219] [−0.251] [1.538]
Loans/Assets 0.00421* 0.00421** 0.0140 0.00980* 0.00204
[1.726] [2.567] [1.147] [1.888] [1.217]
GDP 0.00128 −0.0358*** −0.0963* −0.119*** −0.0852***
[0.0787] [−2.629] [−1.775] [−4.849] [−5.601]
EQ −0.0435*** −0.0112 −0.0157 −0.0833*** −0.0120
[−4.525] [−1.384] [−0.682] [−3.806] [−1.308]
Size 0.102** −0.0273 −0.0786 −0.447** −0.0136
[2.267] [−0.996] [−0.120] [−2.458] [−0.509]
Merger 0.0178 −0.0128 −0.0179 0.153*** 0.0199
[0.756] [−0.626] [−0.179] [2.854] [1.052]
Observations 1143 2252 442 1535 5372

Banks 81 127 77 123 210
R2 0.304 0.876 0.784 0.361 0.757
*
**
***
Table 11
Pairwise correlation coefficients between realized security gains and earnings by bank type and sample period. Earnings here is
defined as net revenue before extraordinary items prior to realized security gains and provision for loan losses. The preSunTrust
and post-SunTrust sample periods for the short term sample are 1996Q4–1998Q3 and 1999Q1–2000Q4, respectively. The pre-
SunTrust and post-SunTrust sample periods for the long term sample are 1992Q1–1998Q3 and 1999Q1–2013Q4, respectively.
With the exception of that for publicly-held banks post-SunTrust for the long term sample, all correlation coefficients in Panel
C are significant at least at the 5% level.
Publicly-held banks Privately-held banks Publicly-held banks Privately-held banks
Panel A: Correlation of realized security gains and earnings – all banks

Pre-SunTrust −0.060* −0.057* −0.052* −0.059*
Post-SunTrust 0.004 −0.044* −0.050* −0.078*
Panel B: Correlation of realized security gains and earnings – banks under $10 billion in assets
Pre-SunTrust −0.060* −0.057* −0.052* −0.059*
Post-SunTrust −0.001 −0.044* −0.059* −0.079*
Panel C: Correlation of realized security gains and earnings – banks over $10 billion in assets
Pre-SunTrust −0.127 −0.597 −0.090 −0.291
Post-SunTrust 0.097 −0.285 −0.015 −0.077
*
Indicates correlation coefficients significant at the 0.01% level.
similar for both types of banks in Panel C than in earlier panels, possibly indicating more similar
reactions by both types of large banks. For the long term sample, both types of large banks appear to
have reduced earnings management, with privately-held banks showing the larger change.
The results in Table 11 indicate that in the short term following the SEC action, there was a reduction
in earnings management via realized security gains that was more concentrated among publicly-held
banks than privately-held banks. This is analogous to the findings supportive of Hypothesis 1a, in
which the relationship between earnings and provisioning weakened relatively more for publicly-
held banks than for privately-held banks in the short term after the SEC action. Banks do not appear
to have responded to strengthened accounting constraints against earnings management through
provisioning by substituting it with earnings management via realized security gains, and instead
may have responded by reducing earnings management more generally. More thorough analysis of
the relationship between realized security gains and earnings is required before drawing any strong
conclusions, but developing a model for realized security gains is beyond the scope of this paper.
7. Conclusion
In this paper, we examine the implications of the SEC action in its SunTrust Bank earnings restate-
ment decision for the relationship between earnings and provisioning for publicly-held banks versus
privately-held banks. This episode represents a natural experiment in which a regulatory shift tight-
ened accounting constraints against earnings management via provisioning that at first applied only
to publicly-held banks directly under the SEC’s purview.
We present evidence that in the short term, the relationship between earnings and provisions
weakened for publicly-held banks, but not for privately-held banks, consistent with a reduction in
earnings management confined solely to publicly-held banks. We also present evidence that over the
long term, that relationship weakened for both publicly-held and privately-held banks, consistent
with a reduction in earnings management among banks of both ownership types. We also present
evidence in our main results that the level of loan loss reserves fell for publicly-held banks relative to
privately-held banks in the short term following the SEC action, and the difference in reserves persisted
through the years leading up to and including the financial crisis. Overall, this evidence is consistent
with the joint dialog between the SEC and bank supervisory agencies that occurred following the SEC
action, which resulted in supervisory guidance that asserted the importance of accounting priorities
in supervisory standards applicable to all banks. We must note a caveat, though, that events unrelated
to the SEC action and the subsequent supervisory interagency guidance (e.g., technological or legal
changes relevant to bank practices) may influence our long term results, although there is no obvious
reason to suspect that such events would have had differential impacts on provisioning by ownership
type.
When we split our sample banks by asset size, the results for small banks are largely in accord
with the results described above for the pooled sample. The large bank results are in some cases
not consistent with the hypotheses. Large publicly-held banks exhibited no significant change in the
relationship between earnings and provisions in either the short term or the long term. For large
privately-held banks, the relationship between earnings and provisions strengthened in the short term
following the SEC action, but in the long term the relationship weakened. One possible explanation is
that, due to SEC supervision or market scrutiny, large publicly-held banks were already constrained
in their ability to manage earnings. The strengthening of the accounting constraints associated with
the SEC action therefore required no change in their provisioning policies. Large privately-held banks,
which were not similarly constrained in the short term but read the SEC action as likely to result in
greater constraints in the future, became more aggressive in their earnings management in the short
term. Once the strengthening of accounting constraints was applied to all banks, large privately-held
banks found those constraints binding, resulting in the long term decline in earnings management. This
scenario is speculative, and it is worth reiterating a caveat from earlier, that the number of observations
for large privately-held banks in the short term sample is quite small, and so any conclusions about
large banks by ownership type should perhaps be treated tentatively.
To the best of our knowledge, this is the first empirical investigation of provisioning and earnings
management in the wake of the SEC action, and as such it provides new evidence in the debate over
why banks entered the financial crisis of 2007–2009 with what in retrospect is seen as insufficient
loan loss reserves. Our main evidence supports the argument that in the years before the crisis, banks
faced accounting constraints against the early recognition of loan losses that prevented them from
increasing provisions during good economic times. During the economic downturn, the incurred-loss
approach required banks to increase provisions procyclically, potentially exacerbating the downturn.
A more forward-looking approach to provisioning that gives greater priority to the early recognition
of loan losses would likely reduce the procyclicality of provisioning.
Disclaimers
The views expressed belong to the authors and do not represent the views of the Federal Reserve
Bank of Richmond, the Federal Reserve System, the Office of the Comptroller of the Currency, or the
United States Treasury Department.
Acknowledgments
We thank seminar participants at the Federal Reserve Bank of Richmond and participants at the
2011 Federal Reserve System Committee on Financial Structure and Regulation conference, particu-
larly Jeff Gunther, for helpful comments. Susan Maxey and Jordan Nott provided excellent research
assistance.
Appendix A.
Provisioning for loan losses in the United States is accounted for under FAS Statement 5, Accounting
for Contingencies (issued in March 1973), and FAS 114, Accounting by Creditors for Impairment of a
Loan – an amendment of FASB Statements Nos. 5 and 15 (issued in May 1993). Impaired loans evaluated
under FAS 114, which provides guidance on estimating losses on loans evaluated individually, must
be valued based on the present value of cash flows discounted at the loan’s effective interest rate, the
loan’s observable market price, or the fair value of the loan’s collateral if collateral dependent. Loans
individually evaluated under FAS 114 that are not found to be impaired are transferred to homogenous
groups of loans that share common risk characteristics, which are evaluated under standard FAS 5.
FAS 5 provides for accrual of losses by a charge to the income statement based on estimated losses if
two conditions are met:
(1) information available prior to the issuance of the financial statements indicates that it is probable
that an asset has been impaired or a liability has been incurred at the date of the financial statement,
and
(2) the amount of the loss can be reasonably estimated.35
Both FAS 114 and 5 allow banks to include environmental or qualitative factors in consideration
of loan impairment analysis. Examples of these factors include, but are not limited to, underwriting
standards, credit concentration, staff experience, local and national economic business conditions. In
addition, FAS 5 allows for the use of loss history in impairment analysis.36 These elements provide
bankers with flexibility in determining the level of provisions taken against incurred losses when they
are well substantiated by relevant data or documentation required by supervisors and accountants.
Banks identify losses by categorizing loans based on their payment status (i.e. current, 30 days past
due, 60 days past due, etc.) and the severity of delinquency (which can vary by asset class) and, assess
whether a provision should be taken on loans they expect to experience a loss, if the loss is probable
and estimable.
The Basel Accord of 1988 set current rules for bank capital regulation and the role of loan loss
reserves (LLR) in capital regulation. In 1991, FDICIA enacted these changes into law. LLR were no longer
counted as a component of Tier 1 capital but were counted toward Tier 2 capital, up to 1.25% of the
bank’s risk-weighted assets. If a bank increases its LLR, the effect is therefore to increase Tier 2 capital
while reducing retained earnings and Tier 1 capital.37 If, as a result of this transfer, Tier 1 regulatory
thresholds become binding, (usually in bad economic times), bank supervisors would require the bank
to issue more capital or reduce its measured risk. Laeven and Majnoni (2003) have argued that since
FDICIA “. . .from the perspective of compliance with regulatory capital requirements, it became much
more effective for U.S. banks to allocate income to retained earnings (entirely included in Tier 1 capital)
than to loan loss reserves (only partially included in Tier 2 capital).”38 FDICIA reserving rules became
effective in 1992, which is when we begin our sample, and those rules remain constant throughout
our sample. We thus abstract away from a discussion of loan loss provisions as a means of capital
management, focusing on earnings management instead.
References
Adams, B., Carow, K. A., & Perry, T. (2009). Earnings management and initial public offerings: The case of the depository
institution. Journal of Banking and Finance, 33, 2363–2372.
Ahmed, A. S., Takeda, C., & Thomas, S. (1999). Bank loan loss provisions: A reexamination of capital management, earnings
management, and signaling effects. Journal of Accounting and Economics, 28, 1–25.
Arrellano, M., & Bond, S. (1991). Some tests of specification for panel data: Monte Carlo evidence and an application to employ-
ment equation. Review of Economic Studies, 58, 277–297.
Asea, P., & Blomberg, B. (1998). Lending cycles. Journal of Econometrics, 83, 89–128.
Barth, M., Beaver, W., & Wolfson, M. (1990). Components of bank earnings and the structure of bank share prices. Financial
Analysts Journal, 46(3), 53–60.
Basel Committee on Banking Supervision. (2011). Basel III: A global regulatory framework for more resilient banks and banking
systems.
Basel Committee on Banking Supervision. (2006). Basel II: International convergence of capital measurement and capital standards:
A revised framework – Comprehensive version.
Beatty, A. L., Chamberlain, S. L., & Magliolo, J. (1995). Managing financial reports of commercial banks: The influence of taxes,
regulatory capital, and earnings. Journal of Accounting Research, 333, 231–262.
Beatty, A. L., Ke, B., & Petroni, K. R. (2002). Earnings management to avoid earnings declines across publicly and privately held
banks. Accounting Review, 77, 547–570.
Benston, G. J., & Wall, L. D. (2005). How should banks account for loan losses? Federal Reserve Bank of Atlanta. Economic Review,
Fourth Quarter, 19–38.
35
Financial Accounting Standards Board (1973).
36
SR 06-17: Interagency Policy Statement on the ALLL, December 13, 2006. SR 01-17: Policy Statement on ALLL Methodolo-
gies and Documentation for Banks and Savings Institutions, July 2, 2001. SR 99-22: Joint Interagency Letter on the Loan Loss
Allowance, July 26, 1999.
37
Ahmed et al. (1999) provide evidence that, in line with reduced incentives in the new regulatory regime of risk based capital,
banks engaged in less capital management through loan loss provisions in 1991–1995 compared to 1985–1990.
38
Laeven and Majnoni (2003), p. 194.
Berger, A. N., & Udell, G. F. (2003). The institutional memory hypothesis and the procyclicality of bank lending behavior. Journal
of Financial Intermediation, 13, 458–495.
Bernanke, B. (2009). Financial reforms to address systemic risk. Remarks before the Council on Foreign Relations, Washington, D.C.,
March 10, 2009.
Bikker, J. A., & Metzemakers, P. A. J. (2005). Bank provisioning behavior and procyclicality. Journal of International Financial
Markets, Institutions, and Money, 15, 141–157.
Bouvatier, V., & Lepetit, L. (2008). Banks’ procyclical behavior: Does provisioning matter? Journal of International Financial
Markets, Institutions, and Money, 18, 513–526.
Bouvatier, V., & Lepetit, L. (2012). Provisioning rules and bank lending: A theoretical model. Journal of Financial Stability, 8,
25–31.
Bouvatier, V., Lepetit, L., & Strobel, F. (2014). Bank income smoothing, ownership concentration and the regulatory environment.
Journal of Banking and Finance, 41, 253–270.
Cornett, M. M., McNutt, J., & Tehranian, H. (2009). Corporate governance and earnings management at large US bank holding
companies. Journal of Corporate Finance, 15, 412–430.
Dye, R. A. (1988). Earnings management in an overlapping generations model. Journal of Accounting Research, 26, 195–235.
Financial Accounting Standards Board. (1973, March). Summary of Statement No. 5: Accounting for contingencies.
Financial Accounting Standards Board. (1993, May). Statement of Financial Accounting Standards No. 114: Accounting by creditors
for impairment of a loan – An amendment of FASB Statement Nos. 5 and 15.
Financial Accounting Standards Board. (2014, February). Project update on accounting for financial instruments – Joint Project of
the FASB and IASB. Available at: http://www.fasb.org/fi improvements to recognition and measurement.shtml#summary
Fonseca, A. R., & González, F. (2008). Cross-country determinants of bank income smoothing by managing loan-loss provisions.
Journal of Banking and Finance, 32, 217–228.
Foos, D., Norden, L., & Weber, M. (2010). Loan growth and riskiness of banks. Journal of Banking and Finance, 34, 2929–2940.
Gebhardt, G., & Novotny-Farkas, Z. (2011). Mandatory IFRS adoption and accounting quality of European banks. Journal of
Business Finance and Accounting, 38, 289–333.
Greenwald, M. B., & Sinkey, J. F., Jr. (1988). Bank loan loss provisions and the income smoothing hypothesis: An empirical
analysis, 1976–1984. Journal of Financial Services Research, 1, 301–318.
Gunther, J. W., & Moore, R. R. (2003a). Early warning models in real time. Journal of Banking and Finance, 27, 1979–2001.
Gunther, J. W., & Moore, R. R. (2003b). Loss underreporting and the auditing role of bank exams. Journal of Financial Intermediation,
12, 153–177.
Holod, D., & Peek, J. (2007). Asymmetric information and liquidity constraints. Journal of Banking and Finance, 31, 2415–2451.
International Accounting Standards Board. (2014, July). IASB completes reform of financial instruments accounting. Available at:
http://www.ifrs.org/Alerts/Publication/Documents/2014/PR-IASB-completes-reform-financial-instruments-accounting-
July-2014.pdf
Keeton, W. (1999). Does faster loan growth lead to higher loan losses? The Federal Reserve Bank of Kansas City. Economic Review,
Second Quarter, 57–75.
Kwan, S. H. (2004, September). Risk and return of publicly held versus privately owned banks. The Federal Reserve Bank of New
York. Economic Policy Review, 97–107.
Laeven, L., & Majnoni, G. (2003). Loan loss provisioning and economic slowdowns: Too much, too late? Journal of Financial
Intermediation, 12, 178–197.
Lambert, R. A. (1984). Income smoothing as rational equilibrium behavior. Accounting Review, 59, 604–618.
Lown, C., & Morgan, D. (2006). The credit cycle and the business cycle: New findings using the loan officer opinion survey.
Journal of Money, Credit, and Banking, 38, 1575–1598.
Moyer, S. (1990). Capital adequacy ratio regulations and accounting choices in commercial banks. Journal of Accounting and
Economics, 13, 125–154.
Nichols, D. C., Wahlen, J. M., & Wieland, M. M. (2009). Publicly-traded versus privately-held: Implications for conditional
conservatism in bank accounting. Review of Accounting Studies, 14, 88–122.
Rajan, R. (1994). Why bank credit policies fluctuate: A theory and some evidence. Quarterly Journal of Economics, 109, 399–441.
Securities and Exchange Commission Staff Accounting Bulletin: No. 102 – Selected Loan Loss Allowance Methodologies and
Documentation Issues, July 6, 2001. Available at: http://www.sec.gov/interps/account/sab102.htm
Supervision and Regulation Letter 06-17: Interagency Policy Statement on the ALLL, Board of Governors, Federal Reserve System,
December 13, 2006.
Supervision and Regulation Letter 01-17: Policy Statement on ALLL Methodologies and Documentation for Banks and Savings
Institutions, Board of Governors, Federal Reserve System, July 2, 2001.
Supervision and Regulation Letter 99-22: Joint Interagency Letter on the Loan Loss Allowance, Board of Governors, Federal
Reserve System, July 26, 1999.
Trueman, B., & Titman, S. (1988). An explanation for accounting income smoothing. Journal of Accounting Research, 26, 127–139.
Wall, L. D., & Koch, T. W. (2000). Bank loan loss accounting: A review of theoretical and empirical evidence. Federal Reserve
Bank of Atlanta. Economic Review, Second Quarter, 1–19.
Wahlen, J. (1994). The nature of information in commercial bank loan loss disclosures. Accounting Review, 69, 455–478.

Loan Loss Provisioning Accounting Features

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Loan Loss Provisioning Accounting Features

Hochgeladen von

Copyright:

Verfügbare Formate

Journal of Economics and Business 78 (2015) 92–117

Contents lists available at ScienceDirect

Journal of Economics and Business

Loan loss provisions, accounting constraints,

Article history: We examine bank-level changes in the relationship between earn-

∗ Corresponding author. Tel.: +1 410 455 8485.

All Banks Public Banks Private Banks

4. Data and methodology

Variable Short term sample Long term sample

Number of institutions 9432 13,916

Panel B – banks under $10 billion in assets

Number of institutions 9378 13,863

Panel C – banks over $10 billion in assets

Number of institutions 88 210

5.1. Bivariate analysis

Short term sample Long term sample

Publicly-held banks Privately-held banks Publicly-held banks Privately-held banks

Panel A: Correlation of PLL/Assets and PPNR – all banks

5.2. Multivariate analysis of provisions over the short term

PLL/Assetsit = ˛ + ˇ1 Publicit + ˇ2 AfterSTit + ˇ3 PPNRit + ˇ4 NCO/Assetsit

+ ˇ5 NPTL/Assetsit + ˇ6 Loans/Loansit + ˇ7 Loans/Assetsit + ˇ8 GDPit + ˇ9 EQit

+ ˇ10 Sizeit + ˇ11 Mergerit + q Quarterq + i Fixedi + εit

Public 0.0401** 0.0557*** 0.0763** 0.0405** 0.0565

Observations 18,422 110,761 129,183 129,183 129,183 129,183 129,183

Observations 17,587 110,671 128,258 835 90 925 87

Public 0.0125 0.00938 0.0793*** 0.0131* 0.0447**

Observations 81,284 567,084 648,368 648,368 648,368 648,368 648,368

5.3. Multivariate analysis of provisions over the long term

Public 0.0542** 0.912**

Observations 76,783 566,213 642,996 4501 871 5372 816

5.5. Robustness checks

Short term sample Long term sample

Public 0.135*** 0.132*** 0.0930*** 0.0741*** 0.130

Observations 129,183 128,258 925 87 648,368 642,996 5372 816

(1) Pre- (2) Pre- (3) (4) Post- (5) Full

Public 0.0504** 0.0264 0.0802 0.322*** 0.0608***

Observations 251,295 244,178 42,822 110,073 648,368

(1) Pre- (2) Pre- (3) (4) Post- (5) Full

Public 0.0509** 0.0387 0.104 0.330*** 0.0564***

Observations 250,152 241,926 42,380 108,538 642,996

(1) Pre- (2) Pre- (3) (4) Post- (5) Full

Public −0.154 −0.0349 −0.0747 0.163

Observations 1143 2252 442 1535 5372

Short term sample Long term sample

Publicly-held banks Privately-held banks Publicly-held banks Privately-held banks

Panel A: Correlation of realized security gains and earnings – all banks

Das könnte Ihnen auch gefallen

Public 0.0401 0.0557* 0.0763 0.0405 0.0565

Public 0.0542 0.912

Public 0.135* 0.132* 0.0930* 0.0741* 0.130

Public 0.0504 0.0264 0.0802 0.322* 0.0608***

Public 0.0509 0.0387 0.104 0.330* 0.0564***