Beruflich Dokumente
Kultur Dokumente
1994; 46:299-305
Circular number 201 (May 31, 1985), Comptroller of the Currency, states bank
managers should follow relevant criteria when determining loan-loss provisions.
We estimate the tie between relevant criteria to the actual provision. Contrary to
results of recent past studies, there is no evidence of income smoothing. We find
that bank managers consider past loan risk, loan quality deterioration, and foreign
risk when determining loan-loss provisions. Bank managers do not consider off-
balance sheet exposure or changes in loan composition. Extra provisions are taken
in 1987, probably to address the LDC crisis. Money-center bank managers are less
likely to consider any of the criteria studied when determining loan-loss provisions.
I. Introduction
The criteria of loan-loss provisions is a topic of interest in the literature [Greena-
walt and Sinkey (1988), Keeton and Morris (1987), Ma (1988), and Scbeiner (1981)].
Ma (1988) and Greenawalt and Sinkey (1988) find an income smoothing effect in
the determination of the loan-loss provision. The strategy behind income smooth-
ing is that bank managers take a large loan-loss provision in a good year so that
extra reserves are available for bad years. Income smoothing is considered desir-
able because it reduces perceived volatility in income [Ronen and Sadan (1975),
Lambert (1984), and Moses (1987)], thereby maintaining stock price stability.
However, income smoothing may not encourage bank managers to accurately
disclose loan losses if income levels are low, resulting in misleading information
about the bank's condition.
To address the disclosure of loan losses, the Comptroller of the Currency
Administrator of National Banks issued Circular Number 201 (May 31, 1985)
stating that banks are required to maintain an adequate allowance for loan losses
and to establish a policy ensuring that relevant criteria were considered in the
loan-loss allowance decision. The criteria are outlined later. 1
Department of Law and Finance, Saginaw Valley State University, University Center, Michigan
(JLW); Department of Finance and Insurance, Michigan State University, East Lansing, Michigan
(JRB).
Address reprint requests to Professor Jill L. Wetmore, Department of Law and Finance, Saginaw
Valley State University, 338 Brown Hall, University Center, Michigan 48710.
1 The text of the Circular is available from the authors on request.
We determine if evidence exists that banks are using criteria described in the
Circular to determine loan-loss provisions and if banks practice income smoothing.
A model is developed and estimated that proxies the major criteria outlined in the
Circular. We find no evidence of income smoothing and the explanations of
loan-loss provisions vary by bank type.
2 The variable "net charge-offs"is loan charge-offsless recoveriesfrom past charge-offs.An average
of the past three years is used.
Loan-Loss Provisions 301
Bank holding companies (banks) are chosen from the top 100 banks by primary
capital in Moody's Bank and Financial Manual (1989). Year-end data are taken
from annual reports [Ma (1988)]. Banks from this list are included if all data are
available for the period studied, 1986-1990 [Greenawalt and Sinkey (1988)],
resulting in 82 banks or 410 data points. 4 All variables except the dummy variables
and LDC are weighted by total assets to reduce problems with heteroskedasticity
[Greenawalt and Sinkey (1988)]. The variable LDC is weighted by total loans. Table
1 shows the average measures of these variables.
Equation (1) is first estimated with all banks using pooled data and a dummy
variable model, where the intercept varies over time [Fomby et al. (1984), p. 328].
Second, we allow all variables to vary by bank type and continue to vary the
3 Nominal off-balance sheet exposure is consistently available whereas at-risk exposure is not.
4 The list of banks is available from the authors on request.
302 ,} t WclmOl-C ai]{J J. R. I:hick
"fable 1. A v e r a g e M e a s u r e s ol h t d c p c n d e n l V a r i a b l e s ot' a S a m p l e oI 82 C o m m e r c i a l B a n k
H o l d i n g C o m p a n i e s (1986 - 19911)"
Standard
Variablc / ~tCall dcviatkm Maximum Minimum
intercept over time. This is an error components model, and dummy variables are
used to divide the banks into groups of regional (REG), superregional (SREG),
and money-center (MONC) banks [Fomby et al. (1984), pp. 332-334].
III. R e g r e s s i o n R e s u l t s
The regression results are jointly significant and there is no serious autocorrelation
(Table 2). In the "all banks" results, the coefficients of the variables ACHG and
NNP are positive and significant at the 5% level as in Ma (1988) and Greenawalt
and Sinkey (1988). This indicates that managers do consider future losses, known
deterioration in credit, historical losses, nonaccruals and delinquencies, and eco-
nomic conditions when determining loan-loss provisions. The coefficient of variable
LDC is positive, but significant only at the 10% level, indicating a lesser concern
for foreign loan risk.
The coefficient of variable OB is significant and negative, indicating that
off-balance sheet exposure is not considered or in inversely related to loan-loss
provisions. Because the variable OB is positively correlated with LLP, the sign is a
p r o b l e m . O n e c a u s e o f this p r o b l e m is that t h e variable O B is highly c o r r e l a t e d
w i t h L D C e x p o s u r e P A n o t h e r w e a k n e s s o f O B is that it m e a s u r e s t h e n o t i o n a l
s The correlation between these two variables is 0.53. Therefore, some problem with multi-
collinearilty may exist. The correlation between the variables LLP and OB is 0.I5, whereas the
correlation between the variables LLP and LDC is 0.18, suggesting that the variable LDC is more
important when considering loan-loss provisions.
Loan-Loss Provisions 303
amount of off-balance sheet activity rather than the amount at risk. The coefficient
of the variable INC is not significantly different from zero, indicating that smooth-
ing of income is not a factor when determining loan-loss provisions, which differs
from the result of Ma (1988) and Greenawalt and Sinkey (1988).
The coefficient of the variable CHCI is not significantly different from zero,
indicating that bank managers do not consider changes in the percentage of
31)4 .~ 1~ W~:tmorc and .I. t,L Brit+
commercial and industrial loans? ~ Finally, .the coefficient for the dummy variable
TIME87 is positive and significant, indicating that the loan-loss provisions ar~
larger in 1987 than in other years. This is explained by the extensive reduction ~f
LDC exposure that occurred in 1987 as documented in studies as Musumeei and
Sinkey (1990).
The results of the error components model reveal the following additiona~
information. ~ First, the insignificant coefficient of A C H G for money-center banks
indicates that these banks do not consider past or future losses when determining
loan-loss provisions. Money-center banks are also less likely to consider loail
deterioration and economic conditions because the coefficient of NNP is smaller
and significant only at the 10% level.
Superregional and regional banks consider LDC exposure more than the
money-center banks. This is probably because these banks had lower LDC expo-
sure than money-center banks and made an effort to eliminate the exposure rather
than a gesture to reduce it. Regional banks show an insignificant coefficient of the
variable TIME87, indicating that they were unaffected by the LDC reductions of
1987. Money-center banks show a lower level of loan-loss provisions in 1988 and
higher in 1989 than other banks. The increase in 1989 is probably caused by
pressure to address the large amounts of nonperforming commercial construction
loans and the order by regulators to write down Argentine exposure in June 1989.
The results indicate that bank managers tend to consider past experience in loan
losses, foreign loan risk, deterioration of quality of loan portfolios, and economic
conditions when determining loan-loss provisions. Money-center banks are less
likely to consider these factors. Bank managers do not consider off-balance sheet
exposure or loan portfolio composition change although this result may be mislead-
ing. Bank loan-loss provisions are larger in 1987 and this is explained by extra
provisions made for LDC loans. There is no evidence of income smoothing.
Further study will determine if income smoothing is no longer used or the
circumstances of this period precluded its use.
References
Fomby, T. B., Hill, R. C., and Johnson, S. R. 1984. Advanced Econometric Methods. New
York: Springer.
Greenawalt, M. B., and Sinkey, J. F. 1988. Bank loan-Loss provisions and the income-
smoothing hypothesis: An empirical analysis, 1976-1984, Journal of Financial Services
Research 1:301-318.
Keeton, W. R., and Morris, C. S. 1987. Why do banks' loan losses differ? Economic Review:
Federal Reserve Bank of Kansas Oty, 3-21.
6 The coefficients of the changes in loans/assest are insignificant and so are not reported. The other
coefficients remain the same. These results are available from the authors on request.
7 Multicollinearilty is not a problem because the correlations with other variables are low. The
coefficients of correlation are available from the authors on request.
8A White test [White (1980)] is performed to determine if heteroskedasticity is a problem in the
regression equation. The value for chi-squared in 17.03 and the critical value is 18.307 for heteroskedas-
ticity at the 5% level. Therefore, heteroskedasticity is not a problem and the dummyvariable regression
can be used.
Loan-Loss Provisions 305