Chapter 3 Data and Methodologies
3.3 Methodologies
3.3.1 Smooth earnings and Regulatory capital
To determine whether banks use realized gains and losses on AFS securities to smooth earnings and whether banks with regulatory capital decrease the volatility of earnings. All of the panel regression models in this study with bank fixed effect is used to estimate the earnings smoothing and employ residual standard errors clustered by bank and year-quarter to construct t-statistics. Moreover, the regression analysis to test earnings smoothing hypothesis will be divide into two stages which are the year before the announcement of capital proposal of Basel III from year 2008 to 2012 and after the announcement of capital proposal of Basel III from 2013 to 2017.
0 1 2 3 1 1 2 1
RegCapLow and it also has UL, UG, Liquid, and Sec as controls for bank characteristics
likely associated with realized gains and losses on AFS securities. In addition, equation (1) includes macroeconomic control variables because securities gains and losses likely depend on economic condition such as VIX, Unemp, TAIBOR, and Crisis.RGL is realized gains and losses on AFS securities and NI is net income before taxes and RGL, both deflated by beginning-of-quarter total assets. RegCapLow is an indicator variable that equals one if the bank’s regulatory capital ratio—which bank regulators define as allowable Tier 1 plus allowable Tier 2 regulatory capital, deflated by risk-weighted assets—before RGL and after taxes, is in the lowest decile for the
quarter, and zero otherwise.
UG is accumulated unrealized gains on AFS securities, UL is accumulated
unrealized gains and losses on AFS securities. Liquid is the sum of cash items in process of collections. Sec is the sum of held-to-maturity, available-for-sale, and trading securities. The variables of UG, UL, Liquid, and Sec are deflated by beginning-of-quarter total assets. VIX is implied volatility of options on the Taiwan Options Index.Unemp is the number of people actively looking for a job as a percentage of the labor
force in Taiwan. TAIBOR is Taipei Interbank Offered Rate. Crisis is financial crisis which occured in 2008 and 2009. These variables are quarterly, which eliminates need to include quarter fixed effects. ε is error term. The terms i and t denote bank and period.If banks attempt to perform smoothing earnings, this study predicts there will have a negative relationship within RGL and NI whereas positive relationship within RGL and RegCapLow. The relationship within RGL and interaction variables
NI×RegCapLow depends on whether earnings smoothing is dominated by positive or
negative NI banks. If it is dominated by positive NI banks recognizing more losses or fewer gains, then the relationship is positive. On the other hands, if it is dominated by negative NI banks recognizing fewer losses or more gains, then the relationship is negative. In this study excepts bank with more unrealized gains and losses are more likely to realize them to smooth earnings.3.3.2 Engage in a big bath for managing earnings
To determine whether banks with negative earnings use realized AFS securities losses and gains to engage in big bath for managing earnings if their unrealized gains are insufficient to offset the negative earnings. Therefore, this study proceeds to expand equation (1) by separating negative and positive NI banks and to determine whether
banks with negative NI realize fewer losses or more gains, which is consistent with earnings smoothing, or whether they realize more losses or fewer gains, which is consistent with taking a big bath. In other words, this study expects banks with positive
NI to smooth earnings. Equation (2) uses panel regression models with bank fixed effect
is used to estimate big bath earnings management. The regression analysis to test earnings smoothing hypothesis will be divide into two stages which are the year before the announcement of capital proposal of Basel III from year 2008 to 2012 and after the announcement of capital proposal of Basel III from 2013 to 2017.0 1 2 3 1 4 1
RGL negNI negNI RegCapLow negNI UL negNI UG
B posNI posNI UL RegCapLow RegCapLow UG
where in equation (2) includes the main independent variables such as negNI, posNI,
RegCapLow, UG, UL, Size, Liquid, and Sec. posNI (negNI) is NI if NI is greater (less)
than zero, and zero otherwise. The macroeconomic control variables which are VIX,Unemp, TAIBOR, and Crisis that similar to earnings smoothing model.
If bank with negative NI take a big bath (smooth earnings), this study predicts that there will have a positive (negative) relationship for RGL and negNI. Low regulatory capital provides incentives to realize more gains, which mitigates the incentive for negative NI banks to take big baths. Thus, this study predicts that there will have negative relationship within RGL and negNI×RegCapLow. Because there is more opportunity for big bath earnings management when negative NI banks have more accumulated unrealized losses, this study predicts that there will have negative relationship for RGL and negNI×
UL
t−1. Because there is more opportunity for earnings smoothing when negative (positive) NI banks have more accumulated unrealized gains (losses), this study predicts that there will have negative (positive) relationship for RGLand negNI×
UG
t−1 (posNI×UL
t−1). Because there is more opportunity to realize gains to increase regulatory capital when the bank has more accumulated unrealized gains, this study predicts that there will have a positive relationship for RGL andRegCapLow× UG
t−1. All other predictions are as in equation (1).3.3.3 Long-term performance
To determine whether the return on assets (ROA) of banks, the return on equity (ROE), and capital buffer (Buffer) affect the long-term performance. The two-stage least squares (2SLS) regression model in this study with bank fixed effect is used to estimate the long-term performance. The regression analysis to test earnings the long-term performance of banks will also be divide into two stages which are the year before the announcement of capital proposal of Basel III from year 2008 to 2012 and after the announcement of capital proposal of Basel III from 2013 to 2017.
2
where in equation (3) 、(4) 、(5) includes the main dependent variables are similar as equation (2). The NI variable is divided into negNI and posNI. The RGL is endogenous variable. The negNI, posNI, RegCapLow, UG, UL, Size, and Liquid are instrumental variables. The VIX, Unemp, TAIBOR, and Crisis are macroeconomic control variables.