AFE6014-B - Empirical Methods in Accounting and Finance
Tutorial 7 notes
1. Purpose
Know how to use EViews 12 to conduct panel data analysis.
2. Research question
Investigate the impact of a firm’s CEO age (“CEOAGE”) on a firm’s cash holding level (“CASH”)
3. Data and variables
Excel file: Data_Firms Cash Holding Level
Variables: cash ceoage firmsize leverage roa
4. Descriptive statistics
Quick----Group statistics--- descriptive statistics--- common sample (try individual sample)
5. Correlation matrix
View---Covariance Analysis
Covariance Analysis: Ordinary
Date: XXX Time: XXX
Sample: 2009 2012
Included observations: 107
6. Graph (Relation between Cash and each independent variables)
View---Graph—Graph Options—Graph type: Specific:
Scatter Details:
Fit lines: Regression Line
Multiple series: Multiple graphs - First vs. All – click OK
7. Pooled OLS
Object---New Object --- Equation --- Name for object: OLS – click OK
Key in the following variables in Equation specification:
cash c ceoage firmsize leverage roa year03 year04 year05 year06 year07 year08 year09 year10 year11 year12 year13 year14
Estimation settings:
Method: LS - Least Squares (NLS and ARMA)
Dependent Variable: CASH
Method: Panel Least Squares
Date: XXX Time: XXX Sample: 2003 2015
Periods included: 13
Cross-sections included: 18
Total panel (unbalanced) observations: 107
8. Residual checks
View --- Actual,Fitted,Residual --- Actual,Fitted,Residual Table
View --- Actual,Fitted,Residual --- Actual,Fitted,Residual Graph
View --- Actual,Fitted,Residual --- Residual Graph
9. Residual Diagnostics tests for OLS Normality
View ---- Residual Diagnostics ------- Histogram - Normality Test
Serial Correlation
View ---- Residual Diagnostics ------- Serial Correlation LM Test, try different lag 1,2,3 …
Breusch-Godfrey Serial Correlation LM Test:
Null hypothesis: No serial correlation at up to 1 lag
Breusch-Godfrey Serial Correlation LM Test:
Null hypothesis: No serial correlation at up to 2 lags
Heteroskedasticity
View ---- Residual Diagnostics ------- Heteroskedasticity Tests, Test type: White, untick
Include White cross terms
Heteroskedasticity Test: White
Null hypothesis: Homoskedasticity
View ---- Residual Diagnostics ------- Heteroskedasticity Tests, Test type: Breusch-Pagan- Godfrey
Heteroskedasticity Test: Breusch-Pagan-Godfrey
Null hypothesis: Homoskedasticity
10. Coefficient Diagnostics tests for OLS Multicollinearity
View ---- Coefficient Diagnostics ------- Variance Inflation Factors
Variance Inflation Factors
Date: 11/01/22 Time: 00:10
Sample: 1 107
Included observations: 107
11. Interpretation, analysis and discussion
Is this a reliable regression fit? i.e. the OLS output in Step 7 above