Distribution in simulation

[pkpio24]

I am trying to simulate the prices of Oil and Gas. When simulating in LOGprices I get a very big confidence interval once I transform back to real prices, so I would like to sample a distribution from my historic prices and create a distribution on my simulated prices? How do I do this in Eviews?

Another question: is it possible to simulate a simulated cointegrated price path or do I only get the baseline mean?

[EViews Glenn]

Not entirely sure from the question if this is what you want, but series objects in EViews offer a resample proc (Proc/Resample...)

[pkpio24]

Yes, I wonder how I can by resampling the residuals from the model create the distribution and confidence intervals for my simulated prices? Could you please tell me the steps.

My confidence intervals from solving the model and transforming back from LOG are too wide, so I would like to resample the residual distribution from my sample to estimate confidence intervals around my forecasted mean.

[EViews Glenn]

To resample from the residual distribution, save the residuals in a series, and then use the series proc to resample.

[pkpio24]

I want to create a distribution and confidence interval around the forecasted mean with bootstrapping. Please tell me if this is the correct way:
1. Create a VECM
2. Make residuals (from the VECM)
3. resample the residuals to my forecast period (2015-2040)
4. Add the resampled residuals to my forecasted mean.


Is this the right way?

[pkpio24]

Could you please help?
The confidence intervals from the solved VECM model become too large when transformed back from LOG. Hence I need another way to calculate confidence intervals around my mean forecast.