| Model/Data | Result | Remarks | ||||||
Chi-Square | df | p-Value | |||||||
Model Selection Phase |
|
|
|
|
| ||||
Lagrange Multiplier Test (Breusch-Pagan) | Pooled Model | 1859.9 (Chi-Sq.) | 1 | <2.2e−16
| Rejects null hypothesis (i.e., significant individual effects). Therefore, the pooled model cannot be used. | ||||
Hausman Test | Fixed Effects Model, Random Effects Model | 2.7448 (Chi-Sq.) | 4 | 0.6014 | Unable to reject the null hypothesis. As a result, we should choose the random effects model. | ||||
Measures for Robustness of Selected Model | |||||||||
Breusch-Pagan LM Test for Cross-Sectional Dependence in Panels | Random Effects Model | 833.38 (Chi-Sq.) | 66 | <2.2e−16 | Cross-sectional dependence present. | ||||
Pesaran CD Test for Cross-Sectional Dependence in Panels | Random Effects Model | −2.0132 (Z-Test) |
| 0.04409 | Cross-sectional dependence present. | ||||
Augmented Dickey-Fuller Test for Unit Roots | Panel Data | −17.913 (Dickey-Fuller) | 7 (Lag Order) | 0.01 | Rejects null hypothesis. No unit-roots, and therefore, time series is stationary. | ||||
Breusch-Pagan Test for Heteroskedasticity | Random Effects Model | 689.5 (BP) | 4 | <2.2e−16 | Rejects null hypothesis. heteroskedasticity is present. | ||||
Breusch-Godfrey/Wooldridge Test for Serial Correlation in Panel Models | Random Effects Model | 292.67 (Chi-Sq.) | 29 | <2.2e−16 | Rejects null hypothesis. Serial correlation idiosyncratic errors are present. | ||||