I have pre-existing monthly forecasts, but now need to make daily forecasts. The ending value of my daily forecasts 30 days from now should match the first monthly value. (I only have to make a 30 day forecast so I am not worried about further monthly values)
If I take my daily data and just straight line it out 30 days (ie, just regress on time), I would very likely end up with a balance amount not equal to my 1st monthly amount.
Is there a way I can use -forecast solve- to estimate one equation that straight lines my y variable (ie, reg y t ... but more generally, possibly some other variables included) but with the constraint that its value at day 30 equals my first monthly forecast value?
(I don't want to use interpolation methods cause in general I'd like to add other variables to influence my intra-month daily regression model)
I was thinking of something like this
reg y t
estimates store y_equation
forecast estimates y_equation
forecast identity y = 4567 if tin(31jan2015)
forecast solve
But I am guessing I can not use "if tin( )" clauses in a forecast identity statement.
Or perhaps I could do
forecast identity y = y_monthly
where I set up the variable y_monthly to be all zeros but the value I want on day 30?
Any help appreciated, thanks in advance.
If I take my daily data and just straight line it out 30 days (ie, just regress on time), I would very likely end up with a balance amount not equal to my 1st monthly amount.
Is there a way I can use -forecast solve- to estimate one equation that straight lines my y variable (ie, reg y t ... but more generally, possibly some other variables included) but with the constraint that its value at day 30 equals my first monthly forecast value?
(I don't want to use interpolation methods cause in general I'd like to add other variables to influence my intra-month daily regression model)
I was thinking of something like this
reg y t
estimates store y_equation
forecast estimates y_equation
forecast identity y = 4567 if tin(31jan2015)
forecast solve
But I am guessing I can not use "if tin( )" clauses in a forecast identity statement.
Or perhaps I could do
forecast identity y = y_monthly
where I set up the variable y_monthly to be all zeros but the value I want on day 30?
Any help appreciated, thanks in advance.