How to inteprete the results of component GARCH

[cklint]

Hi

I estimate the effect of government intervention on the volatility and level of exchange rates. And I've got results below:

My question is:
(1) Is it Ok to have the negative R-squared (-1.236584)? To my understanding, negative R-squared is not unusual in GARCH-type estimation. Is there anything to impy the misspecification in the estimation result. (2) If these negative values are problem, what should I do to solve the problem? I only know the correlogram-Qstatistics on residual and squred residual to see whether mean and variance equations are properly specified. anything else?

Thanks in advance.


---------estimation result -------------------

Dependent Variable: DLOG(FXWON)
Method: ML - ARCH (BHHH) - Normal distribution
Date: 01/13/11 Time: 17:30
Sample (adjusted): 9/06/2001 3/23/2010
Included observations: 2125 after adjustments
Failure to improve Likelihood after 9 iterations
Presample variance: backcast (parameter = 0.7)
Q = C(11) + C(12)*(Q(-1) - C(11)) + C(13)*(RESID(-1)^2 - GARCH(-1)) +
C(14)*INTB/10000 + C(15)*CUMDB*(INTB/10000) + C(16)*SIZEDB
*(INTB/10000) + C(17)*INTS/10000 + C(18)*CUMDS*(INTS/10000) +
C(19)*SIZEDS*(INTS/10000) + C(20)*D(DIFF) + C(21)*D(CDS)
GARCH = Q + (C(22) + C(23)*(RESID(-1)<0))*(RESID(-1)^2 - Q(-1)) + C(24)
*(GARCH(-1) - Q(-1)) + C(25)*INTB/10000 + C(26)*CUMDB*(INTB
/10000) + C(27)*SIZEDB*(INTB/10000) + C(28)*INTS/10000 + C(29)
*CUMDS*(INTS/10000) + C(30)*SIZEDS*(INTS/10000) + C(31)
*D(DIFF) + C(32)*D(CDS)


Variable Coefficient Std. Error z-Statistic Prob.
GARCH 2.647897 0.383380 6.906723 0.0000
C 0.000266 0.000212 1.256061 0.2091
INTB/10000 -0.016946 0.044274 -0.382764 0.7019
CUMDB*(INTB/10000) -0.063217 0.062281 -1.015017 0.3101
SIZEDB*(INTB/10000) -0.023504 0.016660 -1.410869 0.1583
INTS/10000 0.935460 0.702622 1.331385 0.1831
CUMDS*(INTS/10000) -4.08E-06 0.001631 -0.002502 0.9980
SIZEDS*(INTS/10000) -0.936187 0.702753 -1.332171 0.1828
D(DIFF) 0.001340 0.000889 1.507509 0.1317
D(CDS) 0.000153 2.30E-05 6.641351 0.0000

Variance Equation
C(11) 3.11E-05 2.06E-06 15.11257 0.0000
C(12) 0.558576 0.017329 32.23389 0.0000
C(13) 0.083012 0.010468 7.930376 0.0000
C(14) 0.022706 0.003280 6.922430 0.0000
C(15) -0.000351 0.002171 -0.161727 0.8715
C(16) -0.011716 0.002916 -4.017570 0.0001
C(17) -0.002372 3.92E-07 -6050.184 0.0000
C(18) -2.37E-05 1.26E-05 -1.881518 0.0599
C(19) 0.002388 9.85E-06 242.2857 0.0000
C(20) -2.21E-06 1.38E-05 -0.160536 0.8725
C(21) 1.80E-07 1.81E-07 0.993863 0.3203
C(22) -0.037272 0.015447 -2.412852 0.0158
C(23) 0.038779 0.032266 1.201850 0.2294
C(24) -0.050028 0.183079 -0.273257 0.7847
C(25) -0.009851 0.004149 -2.374277 0.0176
C(26) 0.000489 0.005408 0.090423 0.9280
C(27) 0.013150 0.003846 3.419364 0.0006
C(28) 0.001290 2.52E-06 511.6020 0.0000
C(29) 1.82E-05 1.42E-05 1.278745 0.2010
C(30) -0.001305 9.82E-06 -132.9635 0.0000
C(31) 1.59E-05 1.64E-05 0.966763 0.3337
C(32) -2.59E-08 2.68E-07 -0.096637 0.9230

R-squared -1.236584 Mean dependent var -5.60E-05
Adjusted R-squared -1.246101 S.D. dependent var 0.007117
S.E. of regression 0.010666 Akaike info criterion -5.987229
Sum squared resid 0.240598 Schwarz criterion -5.901973
Log likelihood 6393.431 Hannan-Quinn criter. -5.956021
Durbin-Watson stat 1.656275