When estimating the model, we either demean the observable variables or use cyclical part of HP filter. My question is how to interpret the forecast values ? As we compute conditional forecasts by conditioning on certain variables in dynare
Hi, I do not understand the question. Your question is about plain forecasts or conditional forecast? A conditional forecast is a forecast were some of the variables are constrained to follow a given path in the future (for instance the nominal interest rate is assumed to be constant for two years). Obviously this additional information will affect the forecast of the free (unconstrained) endogenous variables.
Thanks Stephane, my question was simply that for example we have demeaned inflation series and we constrain it for say two quarters and later the forecast produced will be like Smoothted series as fractions while in policy discussion perspectives we often say for example ‘YoY’ anticipated change, so how to precisely present the forecast results? Do we need to construct back the series in indices/levels
I am very sorry, but your English is extremely hard to understand. Also, you need to be more systematic. Are we talking about conditional or unconditional forecasts? What is it that your are trying to achieve? Your last question seems to be about data transformations.
Thanks Professor for your answer,
Let me try to make it clearer
For example, I am asked to run a model and produce forecast out of it for policy discussion purposes setting for example
periods 1 2;
values -0.010856513 -0.002514222;
conditional_forecast(parameter_set = posterior_mean, controlled_varexo = (e_xr), replic = 3000,periods=9);
Now dynare produces two results one is unconditional forecast and the other conditional forecast for all the endogenous variables.
Now that results are produced, how do I go about presenting the forecast results as the norm is to forecast in Year-on-Year % increase terms.
Apologies in advance if l am still unable to make it better understandable