Coefficent incorrect on the first forecasted value..help!

[gmn734]

Hi,

I have run a simple regression (log(dep_variable) c log(indep_variable)) and forecast in Eviews but my first forecasted value is incorrect. My calculated coefficient is 1.8 but first forecasted value (for 2014) is 28 times the growth rate of the independant variable. 2015/16/17/18 all have a growth rate 1.8 times the independant variable. I have attached a picture of the regression output here and the forecast output here

regression forecast.png (138.84 KiB) Viewed 2507 times

I am writing it in a program file as i want to add a lot more models in. So im guessing it is something to do with my programming, here is the relevant excerpt. Can anyone help? Any help is appreciated, thanks

[EViews Gareth]

Could you post bigger pictures? It is pretty difficult to make out the numbers in the images you provided.

[gmn734]

Sorry they were a bit cr*p, clearer ones uploaded

[EViews Gareth]

Calculation looks correct to me.

Your equation is log(dep) = -2.66666 + 1.7999*log(ind)

For 2014, ind is equal to 96.2. Thus log(dep) = -2.6666 + 1.7999*log(96.2) = 5.55. If log(dep) = 5.55, then dep = exp(5.55) = 257.97, which is what you get.

[gmn734]

Thanks Gareth, very helpful

Another question if you don't mind...

The output using the C makes the forecast somewhat unrealistic for 2014, but the profile of the forecast in the years afterwards seems appropriate. Is it bad practice for me to manually substitute C for the previous periods dependant variable, so dep = dep(-1) + 1.79999*(ind)

If i change my model in Eviews to log(dep) c log(dep(-1)) log(ind) i still get a big movement in 2014, which again from a qualitative point of view is unrealistic.

Thanks