fcast compute exp(dependend variable) possible?

[Christian Danner]

I am forecasting some price variable based on VECM. All my variables are transformed to ln , but I would like to forecast the price and not the lnprice , i.e., exp(lnprice):
VECM

vec lnprice lnx1 lnx2 lnx3 lnx4 lnx5, trend(constant) rank(2) lag(4)

-fcast compute- generates the forecasted variable then which I use in a nice graph applying -fcast graph- then.

However, I do not want to graph the forecasted lnprice but the price = exp(lnprice) also still showing the confidence bands (related to price) in the graph.

Any comments how to implement this would really help me.

Thanks for any support in advance!

[Christian Danner]

My second problem related to the -fcast compute- : I actually want to "backcast" , i.e. use the estimation for previous time period to compare the estimation with observed values. STATA output gives me the error then:

fcast compute fc, step(500) dynamic(w(2000w1)) replace

"cannot specify a time prior to the beginning of the estimation sample r(198);"