Table 11 Prais–Winsten FGLS vs ARDL-OLS... | Emerald Publishing

Table 11

Prais–Winsten FGLS vs ARDL-OLS short-run estimates (dependent variable: ΔlnNPLA)

Variable	PW-FGLS Coeff	PW Std. Error	PW t-stat	PW p-value	ARDL-OLS Coeff	ARDL-OLS Std. Error
ΔlnNPLA(−1)	0.251	0.084	2.988	0.004	0.264	0.098
ΔlnGDPA	−0.237	0.089	−2.663	0.009	−0.218	0.104
ΔINF	0.014	0.005	2.8	0.006	0.012	0.006
ΔINT	0.018	0.007	2.571	0.012	0.016	0.008
ΔCAR	−0.022	0.008	−2.750	0.007	−0.019	0.009
RECAP	−0.091	0.033	−2.758	0.007	−0.084	0.038
Constant	0.003	0.007	0.429	0.669	0.004	0.009
$ρ^{ʌ}$ (AR1 coefficient)	0.284				–	–
Durbin–Watson (corrected)	1.98				1.42
Adj. R²	0.478				0.463
F-statistic	12.84***				10.72***
Observations	95				95

Variable	PW-FGLS Coeff	PW Std. Error	PW t-stat	PW p-value	ARDL-OLS Coeff	ARDL-OLS Std. Error
ΔlnNPLA(−1)	0.251	0.084	2.988	0.004	0.264	0.098
ΔlnGDPA	−0.237	0.089	−2.663	0.009	−0.218	0.104
ΔINF	0.014	0.005	2.8	0.006	0.012	0.006
ΔINT	0.018	0.007	2.571	0.012	0.016	0.008
ΔCAR	−0.022	0.008	−2.750	0.007	−0.019	0.009
RECAP	−0.091	0.033	−2.758	0.007	−0.084	0.038
Constant	0.003	0.007	0.429	0.669	0.004	0.009
$ρ^{ʌ}$ (AR1 coefficient)	0.284				–	–
Durbin–Watson (corrected)	1.98				1.42
Adj. R²	0.478				0.463
F-statistic	12.84***				10.72***
Observations	95				95

Note(s): PW-FGLS = Prais–Winsten feasible generalised least squares with AR(1) correction. *** denotes significance at 1%

Source(s): Author's estimation