This project is read-only.

Density of conditional forecasts with jforecast

Topics: Reporting
Nov 4, 2015 at 3:00 PM

I'd have two questions regarding conditional forecasts with jforecast.

1) Could you please redirect me to which algorithm IRIS follows to select the combination of shocks that respects the conditioning plan? Does it minimize the variance-covariance of the possible shocks?

2) Which source of uncertainty is accounted for in the std's from jforecast's output? Out of i) uncertainty over the parameters, ii) uncertainty over the state (initial condition), and iii) uncertainty over the future shocks, could you please confirm that i) is not accounted for e ii) can be switched off with the option 'initCond=','fixed'? If this were the case, the bands would just reflect the variance of the shocks conditional on the constraints that derive from the conditioning plan, that is, the one that you minimize as of 1). Do I get this right?

I would find useful to get the full variance-covariance of future shocks conditional on the constraints when doing conditional forecasts with jforecast. In this way, one could sample from the resulting multivariate distribution and easily obtain densities of conditional forecasts.

Thank you,