Table 4.

Return equation using GARCH and PARCH model for NSE return Table 4

GARCH (NSE)PARCH (NSE)
Pre-L2Pre-L1Post-L1Post-L2Pre-L2Pre-L1Post-L1Post-L2
Panel A: Mean equation
Constant, Β0−0.080*−0.081*0.077*0.081*0.060*0.065*0.064*0.066*
Buddha Purnima0.0930.188−0.306−0.0530.1160.173−0.294−0.101
Christmas−0.0320.2810.1200.343*0.330*0.424**0.1560.379*
Diwali0.141**0.1070.041−0.0390.145**0.1180.049−0.043
Dussehra−0.213−0.190−0.016−0.162−0.211−0.214−0.013−0.188
Eid al -Fitr0.189**−0.0950.3110.0440.027−0.0910.2770.010
Ganesh Chaturthi0.180**0.2650.456**0.366*0.203**0.2850.458**0.379*
Good Friday−0.179−0.2790.014−0.044−0.159−0.2660.000−0.061
Gurunanak Jayanti0.0140.095−0.0120.176−0.0190.093−0.0260.126
Holi0.0020.055−0.005−0.0500.0270.074−0.055−0.090
Mahashivratri0.2230.2110.129−0.0820.2120.2180.133−0.117
Mahavir Jayanti−0.086−0.036−0.051−0.0800.083−0.024−0.054−0.066
Muharram0.121***−0.111−0.330***−0.0410.048−0.093−0.339***−0.047
Panel B: Variance equation
C0.044*0.044*0.021*0.045*0.033*0.034*0.0220.036*
Resid(−1)^20.109*0.110*0.097*0.110*0.120*0.120*0.108*0.121
Garch(−1)0.841*0.8783*0.895*0.851*0.268*0.265*0.183*0.251*
γ    0.888*0.887*0.899*0.885*
δ    1.127*1.171*1.601*1.237**
Diagnostic test for specified model
Ged parameter1.1861.1861.4581.1871.239651.2881.4601.189
Log likelihood10,331.410,331.910,163.510,326.918,380.8810,297.310,144.810,292.8
ARCH-LM (10)0.9910.9690.8730.9690.9971.000.9210.972
Q2 (10)0.9700.9960.9010.9700.9950.9980.9220.991
Wald test0.000.000.000.000.000.000.000.00

Notes: Return equation using GARCH and PARCH model for NSE return;

This table represents the estimate for regression specification using the GARCH and PARCH Model. The parameter γ captures asymmetry in the model (γ < 0, γ ≠ 0 ∼ asymmetric). The power term δ capture both standard deviation (δ = 1) and conditional variance (δ = 2). EViews legacy technique was adopted to maximize the log-likelihood function of the GED. The ARCH LM test the null hypothesis of no autocorrelation up to 10 lags. The Ljung-Box Q-square statistics up to order 10 failed to rejected the null hypothesis of no autocorrelation. The null hypothesis of Wald test is H0: β1 = β2 …. = β12 = 0;

Panel A *, ** and *** represent statistical significance at 1, 5 and 10% levels, respectively

Source: Table created by authors

or Create an Account

Close Modal
Close Modal