next up previous
Next: Summary Up: Confidence Limits by Simulation Previous: Monte Carlo Study of

GARCH Model Example

The GARCH model for time series contains several highly constrained parameters. This example presents estimates and confidence limits for a GARCH(1,1) model applied to 20 years of monthly observations on the capitilization weighted returns of the Wilshire 5000 index.

Define the time series

displaymath566

where t = 1, 2, ... T, and tex2html_wrap_inline578 an observed time series with expected value tex2html_wrap_inline580 .

Further define

displaymath567

where tex2html_wrap_inline582 , tex2html_wrap_inline584 , and

displaymath568

The log-likelihood conditional on max(p,q) initial estimates of the conditional variances is, for tex2html_wrap_inline586

displaymath569

The customary constraints applied to the parameters to enforce stationarity and a positive conditional variance are

eqnarray216

To be able to construct a uniform prior, it is necessary to bound tex2html_wrap_inline506 from above and tex2html_wrap_inline580 from above and below. For the example,

eqnarray218

were added.

The prior, then, is constructed from the above constraints:

displaymath570

where A is the hypervolume inscribed by the inequalities, and V = 400 for this example.

table224

The parameter estimates are presented in Table 1 along with confidence limits computed by four different methods. It can be observed there that the confidence limits of the structural parameter, tex2html_wrap_inline580 , is very consistent across the methods. tex2html_wrap_inline580 is not itself constrained. However, it's confidence limits can be affected by estimates of the remaining parameters occuring in the region of their constraint boundaries. This, however, is provided that the distribution of tex2html_wrap_inline580 is correlated more than about .7 with these parameters. Table 2 presents the estimate of the correlation matrix of the parameters calculated from the maximum likelihood estimate. We see there that the largest correlation is quite small, -.05, and thus we should expect that the constraints on the remaining parameters will have little effect on the structural parameter.

table242

  figure248
Figure 6: Kernel density plots of the parameters of a GARCH(1,1) model of the Wilshire 5000 capitalization weighted returns

Kernel density plots were constructed for each of the parameters from the weighted likelihood bootstrap results (Figure 6). The posterior distribution of tex2html_wrap_inline580 appears to be slightly skewed resulting in a slight difference between the mean and the mode. The mean of the distribution, 1.1167, is very close to the maximum likelihood estimate, 1.0947. The mode, however, appears to be about 1.2.


next up previous
Next: Summary Up: Confidence Limits by Simulation Previous: Monte Carlo Study of

R. Schoenberg
Fri Sep 12 09:47:41 PDT 1997