Evidence from Chinese commercial bank loan loss provisions
Introduction
Earnings are the profits of a company. Firms with good earnings in the future will typically have higher stock prices than those with bad prospects. The firm's ability of generate earnings in the future plays a very important role in determine the stock's price. Earnings management occurs when managers use judgment in financial reporting and in structuring transactions to alter financial reports either to mislead some stakeholders about the underlying economic performance of the company or to influence contractual outcomes that depend on reported accounting numbers (Healy and Wahlen [1999],p.369). Earnings management involves the intentional intervention of management to alter the reported earnings number (through making operating decisions or through changing accounting procedures) for what it would otherwise have been in the absence of such intervention. Earnings management also involves using the discretionary accrual components of accounting earnings to show mangers' private information about the future performance of the firm (Subramanyam 1996). Earnings management is recognized as attempts by management to influence or manipulate reported earnings by using specific accounting methods, recognizing one-time non-recurring items, deferring or accelerating expense or revenue transactions, or using other methods designed to influence short-term earnings. The objective of this essay is to investigate Chinese bank managers' incentives to manage earnings to reduce their variability. Prior research indicates that share prices reflect a risk premium associated with earnings variability (Collins and Kothari, 1989; Easton and Zmijewski, 1989; Barth, Landsman and Wahlen, 1995). Therefore, managers can reduce the cost of raising capital by using their discretion in valuating the certain components of earnings to reduce earnings variability. Bank managers have private information regarding default risks inherent in the loan portfolio; thus their judgment is necessary in estimating the loan loss provisions each period .Managers will increase reported earnings when pre-managed earnings are low and will reduce them when pre-managed earnings are high. Managers can use their inside knowledge of firms current performance to make the report usefulness for decision making and share prices. Moyer (1990) provides evidence consistent with bank managers exercising over the timing of report loan loss provisions to avoid regulatory capital constraints. In this study I will investigate Chinese bank managers' use of discretion in estimating the loan loss provisions to reduce the earnings variability.
Loan loss provision (LLP) is a non-cash expense for banks to account for future losses on loan defaults. Banks indicate that a certain percentage of loans will default or become pay slowly. Thus banks enter a percentage as an expense when calculating their pre-tax incomes. Through this can guarantee a bank's solvency and capitalization if and when the defaults occur. The loan loss provision allocated each year increases with the bank gives the riskiness of the loans. This mean a bank making a large number of risky loans will have a high loan loss provision compared to a bank taking lower risks. Loan loss provision is one of the largest accruals of banks. Loan loss provision can be used to study such smoothing behavior. The smoothing behavior is mentioned before bank manager' use of discretion is estimating the loan loss provision. Bank managers can reflect changes in expected future loan losses by estimating loan loss provision. Therefore bank managers have wide latitude for discretion in estimation of loan loss provision. The U.S. Securities and Exchange Commission (SEC) and bank regulatory agencies have devoted considerable attention to how bank managers use that discretion. Comparing with western countries' banks Chinese banks need to do more works on improving Chinese bank systems. Since China joint WTO foreign banks would enter Chinese market compete with Chinese banks. During the improving of our economy and finance, Chinese banks' accounting regulatory is become more and more close to international accounting regulatory. The China Banking Regulatory Commission is preparing first-ever standards for lending institutions that would require a minimum 2.5 percent of loan loss provision ratios. Though this essay is study about Chinese banks but almost theories provide by other researchers according to western countries are also applicable.
Income smoothing used by Chinese commercial banks' managers is studied to investigate earnings management to reduce earnings variability. Since there is no evidence indicates that cross-sectional target better than time series, in this study, I investigate income smoothing towards a cross-sectional target and a time series.
In 2004 Kiridaran Kanagaretnam, DeGroot School of Business, McMaster University, Ham, Ontario, Cananda L8S 4M4; Gerald J.Lobo, Whitman School of Management, Syracuse University, Syracuse, NY 13244-2130, U.S.A and Robert Mathieu, School of Business and Economics, Wilfrid Laurier University, Waterloo, Ontario, Canada N2L 3C5 have a investigate on earnings management to reduce earnings variability and evidence from bank loan loss provisions. The data they collected are from bank holding companies listed on the FDIC web site and period from 1992 to 2001. Since different countries have different rules about loan loss provision depend on the country's performance, in this study, I investigate earnings management to reduce earnings variability by evidence from Chinese commercial banks loan loss provisions.
My results are close to early research (Kiridaran Kanagaretnam, DeGroot School of Business, McMaster University, Ham, Ontario, Cananda L8S 4M4; Gerald J.Lobo, Whitman School of Management, Syracuse University, Syracuse, NY 13244-2130, U.S.A and Robert Mathieu, School of Business and Economics, Wilfrid Laurier University, Waterloo, Ontario, Canada N2L 3C5, 2004) and consist with my hypothesis that Chinese commercial banks use loan loss to smooth income.
Related literature
One of the first definitions on earnings management was given by Schipper (1989, 92), who defined it as ''…purposeful intervention in the external financial reporting process, with the intent of obtaining some private gain". Schipper (1989) explicitly states, and which can be understood from the Healy and Wahlen (1999) definition, earnings management need not be directly connected to reported earnings, but has an impact through other accounting numbers. Several presentations on earnings management use the term in connection with managerial discretion that has the aim to communicate information to investors that is may not opportunistic (e.g. Dechow and Skinner, 2000 and Scott, 2003). Subramanyam (1996) refers to earnings management only in relation to opportunistic behaviour but not when managerial discretion is used to improve earnings persistence and predictability. Earnings management has been found in connection to financial transaction, for example, management buy-outs (Perry and Williams, 1994). Managers may manage earnings to meet analysts' expectations (Kasznik, 1999) or use income smoothing to make the company show a less risky investment than it really is (Trueman and Titman, 1998). Earnings management can be driven by contracts written in terms of accounting numbers, such as debt covenants (Defond and Jiambavlo, 1994) or bonus contracts (Healy, 1985), and by the attempt to reduce political costs (Jones, 1991). In the earnings management review article by Healy and Wahlen (1999) and SEC both expressed that earnings management is something opportunistic and harmful that is used to mislead at least some stakeholders. Therefore the extent literature on the use of loan loss provisions to manage earnings has mainly focused on detecting earnings management.
Dechow (1994) and some other studiers provided evidence that earnings are more highly related to stock price than actual cash flow. Therefore, in order to maximizing shareholders' welfare managers need to smooth earnings. In prior researches, Ronen and Sadan (1981) indicated that managers can do two kinds of income smoothing activities: (a) intertemporal smoothing by discretionary timing of expenditures such as development and research, repairs and maintenance and asset disposal decisions; intertemporal smoothing also has discretionary accounting accrual decisions; (b) classificatory smoothing through discretionary choice of, say, classifying an item as a component of income form continuing operations, or by treating an item as a "special" item.
If losses of principal appear probable and reasonably estimable, then loan loss provisions are recorded to increase the loan loss allowance and reduce current income (per FAS 5). Actually, the loan loss provision is one of the items under scrutiny by the SEC's task force on earnings management (The Wall Street Journal, November 16, 1998). Though there are difference between Chinese commercial banks and western countries' commercial banks, for example different kinds of provisions. But loan loss provision is also an item under scrutiny by the China Banking Regulatory Commission's task force on earnings management accounting to the rules provided by China Banking Regulatory Commission. The SEC's investigate of reserves stems from its broader concerns about earnings management in banking and other industries. When some banks have been adjusting loss reserves to increase earnings and bolster their banks' return on equity, there is evidence that other banks have been conservative by overly providing for loan losses (American Banker, June 29, 1998). "Although management's process for determining loan loss allowance is judgmental and results in a range of estimated loss, it must not be used to manipulate earnings or mislead investors…" is the statement issued by the SEC and four bank regulatory agencies for loan losses of depository institutions after much debate on whether the alleged discretion over their major accrual is beneficial or detrimental to sound banking (Federal Reserve Release, November 24, 1998). Early research on earnings management concerned the detection of changes in accounting methods that were usually easily observable to outsiders. Thus, it is no surprise that in general this research did not find the assumed manipulation to affect stock prices (for an overview of this early research on earnings management, see Watts, 1998). This is in contrast to more recent investigates focusing on accruals, which has found that earning management goes unconsidered many times by the market (Healy and Wahlen, 1999).
Recent empirical studies get evidence consistent with a positive relation between stock returns and loan loss provisions (cf., Beaver et al. 1989; Elliott et al. 1991; Griffin and Wallach 1991; Johnson 1989). This evidence negates the view that loan loss provisions are interpreted as expenses that reflect expected future loan losses (Wahlen, 1994). Wahlen (1994) found evidence suggests that when future cash flow prospects improve, bank managers increase the discretionary component of unexpected loan loss provisions. Wahlen (1994) also found new evidence on earnings management and earnings management effect on the capital market. Not only Wahlen (1994) but also Collins et al. (1995), among others, find evidence that banks use loan loss provisions to manage income, but Beatty (1995), Moyer et al. 1990), and Ahmed et al.(1999) do not find support for the earnings management hypothesis.
Methodology
In order to study income smoothing, I hypothesize that the motivations to smooth incoming are greater for banks with high or low levels of pre-managed income. Furthermore, I test whether Chinese commercial banks with relatively low pre-managed income have low loan loss provisions (i.e., negative discretionary component of loan loss provisions) and Chinese commercial banks with relatively high pre-managed income have high loan loss provisions (i.e., positive discretionary component of loan loss provisions).
I study three factors that may have effects on the income smoothing behavior of Chinese commercial banks' managers. First, I test whether bank managers use gains or losses from sale of securities as an alternative method for smoothing earnings. Second, I test a bank's capitalization influence on the extent of income smoothing. Poorly capitalized banks have more restrictions and more regulatory supervision therefore limiting them less room to smooth earnings than well capitalized banks. Third, I test the motivations what to reduce the cost of external borrowing to finance growth and to enlarge the lending portfolio of the bank.
Managers of Chinese commercial banks have the incentives to smooth income to reduce the variability in reported earnings. Since earnings variability is a key indicator of risk then banks managers can reduce perceived risk by reducing earnings variability. Managers of banks use their inside knowledge of the bank's current state and business circumstances to prepare the information. When actual income is low, a bank manager takes actions to increase reported income and takes actions to decreases reported income when actual income is high. If income is in line with expectations a bank manager takes no action to adjust income. These are methods and situations for a bank manager to smooth income. As mentioned before in this study I investigate income smoothing towards cross-sectional target. Implied risk premium is consistently higher for commercial banks and that variability and predictability of earnings are key factors in explaining cross-sectional differences in the implied risk premium which evidence provided by Gebhardt, Lee and Swaminathan (2001).
As we all know that earnings variability is a key indicator of risk. Therefore financial practitioners often regard the variability of reported earnings as a source of risk for firm valuation. Furthermore earnings variability is likely to capture fundamental cash flow risk (Gebhardt et al., 2001). Bank shareholders will require a higher risk premium for the increased risk perceived from a more variable earnings stream which was argued by Barth, Landsman and Wahlen (1995). Combining these ideas it can be get that managers of banks with higher earnings variability will have stronger incentives to smooth earnings through loan loss provision. According to these and since I investigate Chinese commercial banks, I make two following hypotheses:
H1 a: The propensity to increase earnings by decreasing discretionary component of loan loss provisions is high for Chinese commercial banks with relatively low pre-managed earnings.
H1 b: The propensity to decrease earnings by increasing discretionary component of loan loss provision is high for Chinese commercial banks with relatively high pre-managed earnings.
As mentioned before, bank managers use their discretion over timing of realization of gains and losses on securities held for sale on smooth income (Beatty and Harris, 1999; Warfield and Linsmeier, 1992; Barth, Beaver, Wolfson, 1990). This is because that securities gains and losses are an alternative means of smoothing income. Therefore there are two ways for Chinese commercial bank managers to smooth income. One way is using discretionary loan loss provisions and the other way is realized gains and losses on the sale of securities. It is obviously that discretionary component of loan loss provisions and securities gains and losses can be instead of each other and they have a negative relation. This suggests me make the following hypothesis:
H 2: The degree of income smoothing through discretionary component of loan loss provisions has a negatively relation to realize gains and losses on securities held for sale.
Nothing can be accomplished without norms or standards. Therefore there are also regulatory to restrict banks. Kim and Kross (1998) indicated that the regulatory treatment of banks differs cross-sectionally depending on their capitalization levels. Regulatory actions such as restrict the least ratio of provisions, the frequency of examinations on banks' activities will different depending on the well- capitalized. In 2010, the China Banking Regulatory Commission is preparing first-ever standards for lending institutions that would require a minimum 2.5 percent of loan loss provision ratio. This is a signal that there will be more restrictions on Chinese banks in the future. In this study I do not need to consider this because the rule is just a plan not in applying. Bank regulators evaluated a bank well-capitalized will has less frequency audited than the bank evaluated as not well-capitalized. In 2003, The Wall Street Journal said that Standard & Poor's evaluated all Chinese mainland banks as rubbish level. In fact this is not a fair report. As financial crisis happen the famous commercial bank which evaluated high level by international evaluate organizations are in trouble. However, Bank of China which was evaluated low level by those organizations is still keeping good liquidity and making profit in the financial crisis. The most use method camel rating has limitation when evaluate Chinese commercial banks. Different regions may have different regulatory but well-capitalized banks have more ability to smooth reported income than not well-capitalized banks is the fact. I define banks with capital ratios above the sample mean total capital ratio as banks that are well-capitalized the same as Liu et al (1997). Then I hypothesize that:
H3: The degree of income smoothing through discretional loan loss provisions is greater for relatively well-capitalized banks than for relatively not-well-capitalized banks.
Internal financing and external financing are two ways of financing. There are many debates about financing. In recent years it is often argued that the need for external financing is an important motivation for income smoothing. Dechow et al. (1996) find that banks' financing needs have not received much attention in prior academic research and the cost of financing is a function of bank's perceived risk. Furthermore he also notes that bank managers have an incentive to smooth income to reduce fluctuations. It is easy to understand that if the bank can increase additional financing on more beneficial terms, shareholders can benefit from this behavior. Thus, I think need for external financing is another incentive for income smoothing by Chinese banks' managers. I consider a bank's total loans relative to its total deposits in order to measure the need for external financing. When the ratio of a bank's total loans divided by total deposits is greater than one, then the bank need to borrow money to finance its loan portfolio. Therefore I use the ratio of total loans divided by total deposits as my measure of need for external financing. Thus, I hypothesis that:
H4: The degree of income smoothing through discretionary loan loss provisions is positively related to the need for external financing.
These hypothesizes are consist with prior research (Kiridaran Kanagaretnam, DeGroot School of Business, McMaster University, Ham, Ontario, Cananda L8S 4M4; Gerald J.Lobo, Whitman School of Management, Syracuse University, Syracuse, NY 13244-2130, U.S.A and Robert Mathieu, School of Business and Economics, Wilfrid Laurier University, Waterloo, Ontario, Canada N2L 3C5 2004). They collect data form FDIC web site and have 22,640 bank-year observations.
From the hypothesis it can be seen that I study three bank-specific factors that may influence the income smoothing behavior of bank managers. These three factors are the incentives of managers of Chinese commercial banks demand for external financing, alternatives to income smoothing through loan loss provisions and Chinese commercial banks' regulatory restrictions.
Sample Description
The sample is composed of ten Chinese commercial banks. They are Bank of China, Industrial and Commercial Bank of China, Agricultural Bank of China, China Construction Bank, Communications Bank of China, SPD Bank, CIB Bank, SD Bank, Everbright Bank of China, China Merchant Bank. The data are collected every year period from 2001 to 2010. Therefore there are a total of 100 bank-year observations. I just chose ten commercial banks one reason is the time limit another reason is the ten commercial banks are famous commercial banks in China can approximately show the commercial banks' situation. In 2001 Ministry of Finance promulgate <<The financial enterprise bad debt prepares to withdraw and the bad debt cancels after verification the policing method>> to improve loan risk valuation and loan loss valuation system. This method contributes to the growth of Chinese commercial bank. This is the reason I chose the period from 2001 to 2010. I lose 10 observations because I deflate the variables used either by beginning assets or beginning loans. Therefore this leaves me with a final simple of 90 bank-year observations for my empirical analysis. I search the data by Baidu web site. Some data are provided by the bank's performance report and some are provided by securities companies' estimation of the bank. The data of loan loss provision can not get directly from the statement. It is the result of nonperforming loans multiply by nonperforming loans coverage.
Model
Since there is the discretionary component of loan loss provisions (DLLP) and nondiscretionary component of loan loss provisions (NDLLP). In this study I will do two multiple regressions. One is used to test nondiscretionary component of loan loss provisions. In this regression there have three variables one is change in nonperforming loans (CHNPL), the other two are the beginning balance of nonperforming loans (NPLt-1) and change in total loans (CHLOAN) to estimate the nondiscretionary component of loan loss provisions. This idea is consist with prior research by Kim and Kross (1998).
I use the following model to estimate nondiscretionary Loan loss provision:
LLPit= α0+α1NPLit-1+α2CHNPLit+α3CHLOANit+εit [1]
Where,
LLPit =provisions for loan losses deflated by beginning loans;
NPLit- =beginning of period nonperforming loans deflated by beginning loans;
CHNPLit =change in the value of nonperforming loans deflated by beginning loans;
CHLOANit =change in the value of loans deflated by beginning loans.
Then I use the statistical analysis software Stata to do the regression. Since the independent variables account for the nondiscretionary component of loan loss provisions in equation [1], then the discretionary component of loan loss provisions is given by the residual term. After analysis the first regression, I regresses the residuals from equation [1], which represent the discretionary component of loan loss provisions (DLLP), on the three factors identified above that can potentially influence income smoothing behavior. I use the following model:
DLLPit =β0 + β1RGLASSit +β2L/DEPit+ β3WELLit + β4EBTPit +β5LSIZEit + εit [2]
DLLPit =discretionary component of provision for loan losses estimated as the residual of equation [1];
RGLASSit =realized gains and losses on securities for sale deflated by beginning total assets;
L/DEPit = ratio of loans to deposites;
WELLit =a dummy variable which equals 1 when a bank's capital ratio is above the sample mean and equals 0 otherwise;
EBTPit =earnings before tax and provisions deflated by beginning total assets;
LSIZEit =the natural logarithm of total assets.
RGLASS, L/DEP and WELL, the first three variables are picked to explain cross-sectional differences in discretionary component of loan loss provisions. The fourth variable, EBTP, uses to find the income smoothing incentive related to pre-managed performance and the fifth variable, LSIZE, is a control variable. Before run the regression some predictions are made. I predict a negative (positive) sign for the coefficient of RGLASS, a positive (negative) sign for the coefficient of WELL and also a positive (negative) sign for the coefficient of L/DEP for Chinese commercial banks with a high (low) pre-managed relative performance. These predictions are based on my above hypothesis. Since bank managers will decrease discretionary component of loan loss provision (DLLP) when earnings before tax and provisions (EBPT) is low and increase DLLP when EBPT is high to reduce earnings variability, I predicts EBTP is positively related to DLLP.
Results
Table 1 Descriptive Statistics
Variables
Number of observation
Mean
Standard Deviation
LLP
90
0.0065
0.0109
NPL
90
0.0755
0.0345
CHLOAN
90
0.1221
3.3252
CHNPL
90
0.0056
0.0458
EBTP
90
0.0323
0.0856
REGLASS
90
0.0008
0.0063
L/DEP
90
0.6874
0.1980
Notes: LLP=loan loss provisions deflated by beginning total loans; NPL=non performing loans deflated by beginning total loans; CHLOAN=change in loans deflated beginning total loans; CHNPL=the change in nonperforming loans deflated by beginning total assets; RGLASS=realized gains and losses on securities for sale deflated by beginning total assets; L/DEP=ratio of loans to deposits.
Table 1 shows descriptive statistics for the ten Chinese commercial banks from 2001 to 2010. As indicated in table 1, the loan loss provision on average shows 0.65 percent of beginning loans bigger than foreign banks data from period 1992 to 2001. Almost Chinese commercial banks decrease the nonperforming loans year by year and increase the loan loss provisions in the recent ten years.
Table 2 Estimation of discretionary loan loss provisions
LLPit= α0+α1NPLit-1+α2CHNPLit+α3CHLOANit+εit
Intercept
NPLit-1
CHNPL
CHLOAN
Expected sign coefficient
+
0.00622
+
0.12356
+
0.37052
+
0.00146
T-statistic
25.32***
48.52***
62.23***
34.79***
Adjusted R2
0.4697
Observations
90
Notes: LLP=loan loss provisions deflated by beginning total loans; NPLit-1
= beginning nonperforming loans deflated by beginning total loans; CHLOAN=change in total loans deflated by beginning total loans; CHNPL=change in nonperforming loans deflated by beginning total loans. ***indicates significance at p=0.01
As results of estimating equation [1] are presented in Table 2, that results are consist with expected. It can be seen that the coefficients of the three variables are positive and significant at the 0.01. Therefore an increase in nonperforming loans results in an increase in provisions for loan losses. Loan loss provisions is equal to nonperforming loans multiply by nonperforming loans coverage in Chinese banks statement. If nonperforming loans coverage is constant then nonperforming loans is positively correlated with the loan loss provisions. The result is consistent with it.
Before the second multiple regression I partition the level of discretionary of loan loss provisions by using the change in pre-managed earnings. Hypothesis 1 predicts that Chinese commercial banks with low pre-managed performance have incentives to increase reported earnings. Therefore I need to observe a low level of discretionary of loan loss provisions for these Chinese commercial banks. Hypothesis 1 also predicts that Chinese commercial banks with high pre-managed performance have incentives to decrease the reported earnings. Thus, I also need to observe a high level of discretionary loan loss provisions for these Chinese commercial banks. In order to observe these I classify Chinese commercial banks into two groups (referred to as low and high groups) based on the relative change in annual performance for each year. Here I only classify into two groups because of small samples. The following table 3 presents the results.
Table 3 Discretionary loan loss provisions partitioned by annual relative performance
Change in Earnings before Tax and Provisions
High
Low
Mean
0.00065
-0.00032
Median
0.00013
-0.00094
Standard deviation
0.00536
0.00389
Observations
47
43
Student's t
(p-value)
5.26
(0.00)
-4.18
(0.00)
It can be found that the results in table 3 are consistent with my predictions. Chinese commercial banks with relatively high change in performance have significantly positive discretionary loan loss provisions (t=5.26>2.76; p<0.01). Comparing with Chinese commercial banks with low change in performance discretionary loan loss provisions is negative and significant at 0.01 level (t=-4.18<-2.76). Therefore it can be said that Chinese commercial banks with high pre-managed performance have incentives to decrease the reported earnings and with low pre-managed performance have incentives to increase the reported earnings.
The results present before are the evidence that Chinese commercial bank managers use of discretionary component of the loan loss provision to smooth income. Furthermore, I find the results show that Chinese commercial bank managers decrease (increase) earnings by increasing (decreasing) discretionary component of loan loss provision when the Chinese commercial bank's relative performance is high (low). Every Chinese commercial bank has their own rules to limit managers' ability to smooth earnings. There are two ways for bank managers to smooth earnings. One way is using discretionary component of loan loss provision and the other way is using gains and losses on the sale of securities. Then I test the combined effects of these factors on income smoothing through discretionary component of loan loss provision. The following table shows the results of estimating equation [2] for high and low relative performance groups that are classified based on change in earnings before tax and provisions.
Table 4
DLLPit =β0 + β1RGLASSit +β2L/DEPit+ β3WELLit + β4EBTPit +β5LSIZEit + εit
Panel A: Partition with High relative Change in Earnings before Tax and Provisions
Model 1
Model 2
Model 3
Model 4
Sign
Coefficient
t-statistic
Coefficient
t-statistic
Coefficient
t-statistic
Coefficient
t-statistic
Intercept
+
-0.0065
-4.87
***
-0.0072
-4.57
***
-0.0046
-2.37
**
-0.0066
-4.66
***
L/DEP
+
0.0057
5.86
***
0.0048
4.96
***
RGLASS
-
-0.3425
-10.26
***
-0.3356
-9.58
***
WELL
+
-0.0021
-6.47
***
-0.0025
-4.21
***
EBPT
+
0.0453
50.23
***
0.1635
28.63
***
0.2473
56.37
***
0.0521
30.54
***
LSIZE
0.0002
2.13
**
0.0001
4.86
***
0.0005
2.36
**
0.0003
1.91
*
Adjusted R2
0.3241
0.3258
0.3238
0.3532
Observations
47
47
47
47
Table 4
DLLPit =β0 + β1RGLASSit +β2L/DEPit+ β3WELLit + β4EBTPit +β5LSIZEit + εit
Panel B: Partition with Low relative Change in Earnings before Tax and Provisions
Model 1
Model 2
Model 3
Model 4
Sign
Coefficient
t-statistic
Coefficient
t-statistic
Coefficient
t-statistic
Coefficient
t-statistic
Intercept
_
-0.0095
-9.67
***
-0.0131
-13.46
***
-0.0068
-9.86
***
-0.0095
-12.54
***
L/DEP
_
0.0032
4.36
***
0.0013
1.98
**
RGLASS
+
0.2217
3.26
***
0.1532
4.38
***
WELL
_
-0.0032
-9.58
***
-0.0032
-9.24
***
EBPT
+
0.0386
6.34
***
0.1356
7.53
***
0.2664
6.77
***
0.0436
6.44
***
LSIZE
0.0007
9.62
***
0.0004
10.26
***
0.0005
9.36
***
0.0006
7.92
*
Adjusted R2
0.0473
0.0532
0.0497
0.0545
Observations
43
43
43
43
Notes: DLLP=discretionary loan loss provision estimated as the residual from equation [1]; RGLASS=realized gains and losses on securities for sale deflated by beginning total assets; L/DEP=ratio of loans to deposits; LSIZE=natural logarithm of total assets; WELL=a dummy variable which equals 1 when a bank's capital ratio is above the sample mean and equals 0 otherwise; EBPT=earnings before tax and provisions deflated by beginning total assets.
Each year, observations are partitioned into two equal-sized groups on the basis of change in earnings before tax and provisions.
*** indicates significant at p=0.01; ** indicates significant at p=0.05; * indicates significant at p=0.1
Table 4 is the final results for my study. In table 4 there are four models and the first three models test each variable's effect on discretionary component of loan loss provision. The final model tests all the variables' effect on discretionary component of loan loss provision. Table 4 consist of two panels, panel A shows the results for the high relative change in performance and panel B shows the results for the low relative change in performance. It can be found that the variable EBTP is positive and significant at the 0.01 level in all four models both in panel A and panel B. This indicates that the change in performance of Chinese commercial banks is an important part for explaining managers' use of discretionary component of loan loss provision to smooth income, no matter there are other factors considered in analysis.
In panel A the variables RGLASS and L/DEP are both significant at level 0.01 and are same as the predicted sign. This result consists with my hypothesis 2 and 4. The coefficient of WELL is negative different from the positive predicted sign. Therefore the result do not support hypothesis 3. This may because of Chinese bank regulators are more restrict to Chinese commercial banks in good economic environment. This behavior is like Chinese people often say that be prepared for danger in times of peace. In the 2007 financial crisis many foreign banks are in trouble and some even close down but Chinese commercial banks still make profits. Chinese regulators' prudence help Chinese commercial banks make profits in the financial crisis.
In panel B it can be found that the coefficient of L/DEP is positive and significant at the 0.01 level in model 1. In model 4 the coefficient of L/DEP is positive and significant at the 0.05 level. Therefore the coefficient of L/DEP is different from the predicted sign, I can not reject hypothesis 4. In model 2 and model 4 the coefficient of RGLASS is positive and significant at the 0.01 level. This result is consists with my hypothesis 2. In model 3 and model 4 the coefficient of WELL is negative the as the predicted sign and it also significant at the 0.01 level. This result is consists with my hypothesis 3 different from the result get from panel A. The coefficient of control variable LSIZE is positive and significant in all models.
Conclusion
Earnings management occurs when managers use judgment in financial reporting and in structuring transactions to alter financial reports either to mislead some stakeholders about the underlying economic performance of the company or to influence contractual outcomes that depend on reported accounting numbers (Healy and Wahlen [1999],p.369). Through earnings management that Chinese commercial banks' managers can make the bank looks better than the really performance which may disappoint shareholders and investors. Chinese commercial banks' managers often use loan loss provisions to smooth earnings. In this study, I examine Chinese commercial banks' managers' motivations to reduce earnings variability by manage earnings through the use of discretionary loan loss provision. Prior researches have provided evidence that market requires a premium for earnings variability. Thus Chinese commercial banks' managers have motivations to smooth income in order to reduce the cost of financing. Managers want to make the bank look better, thus, they will increase the reported earnings when pre-managed earnings are low and will decrease the reported earnings when pre-managed earnings are high. Since loan loss provision is the largest accrual for most Chinese commercial banks, loan loss provision can be used to investigate income smoothing behavior.
I use two multiple regressions to do the research. The residual of first multiple regression (equation [1]) is the dependent variable (DLLP) for the second multiple regression. I choose three bank specific factors that could have effect on income smoothing through loan loss provisions. These three factors are corresponding to three variables in equation [2], L/DEP, RGLASS, and WELL. I test the three factors individually and then test three factors combined together. In table 4 model 1, 2, and 3 are present the tested results for the three individual variables.
The four tables present the results I have got and these results are consist with my hypothesis that Chinese banks smooth earnings. In conclusion, Chinese commercial banks' managers use the discretionary component of loan loss provisions to reduce the variability in earnings.
In the last ten years, Chinese commercial banks make progress every year. Since 2001, Chinese commercial banks have paid much attention on loan risk management. Therefore Chinese commercial banks almost make profits in the 2007 financial crisis. From some Chinese commercial banks report, it can be know that nonperforming loans is decrease year by year in recent years but the loan loss provisions is increase by increase provision coverage. This shows the prudent rules of Chinese commercial banks. For some banks the China Regulatory Bank Commission also give higher provision coverage than others.
There are some limitations of my study. One is the sample of my study is too small. I only observe 10 Chinese commercial banks from period 2001 to 2010. The number of banks I observe too small compare with prior researches. I only partition discretionary based on change in earnings before tax and provision, do not test discretionary loan loss provision based on the level of earnings before tax and provisions. This may lead some mistake because in different years the worth of money is always different. This is another limitation. In my study, I only focused on managing earnings variability by discretionary accruals not through real actions by management. I only chose ten famous Chinese commercial banks which are mostly do well in the market. In fact, some small Chinese commercial banks are in trouble. This is also a limitation.
