Difference between forecast and conditional_forecast

Stochastic simulations
1

Hello to all,

Comparing the results between oo_. forecast.Mean.y (forecast function) and oo_. conditional_forecast.uncond.Mean.y (conditional_forecast function) I get different values. Additionally, even though oo_. forecast.Mean.y converge to 0 in the long-term (as expected) the ones of oo_. conditional_forecast.uncond.Mean.y do not. Does anyone know why is this happening? I upload the data and .mod file.

NKM_conditionalf.mod (1.0 KB)

knm_1.xlsx (94.6 KB)

Thank you very much.

2
  1. You should compare
figure
plot(oo_. forecast.Mean.y)
hold on
plot(oo_. conditional_forecast.uncond.Mean.y(2:end))

because oo_. conditional_forecast.uncond.Mean.y contains the initial condition.
2. The conditional forecasts are generated by simulating with shocks generated from the ergodic distribution. Thus, there is variability that will only go away if replic goes to infinity.

3

Perfect, thank you for your timely reply. Another question, both the conditional_forecast and forecast function use the last observed data point as the initial value?

Thank you,

4

Yes, if specified correctly, they will start at the inferred value of the state variables in the last period.

5

Thank you for your reply.