Restricting the dependent variable

[jacarandas]

Hi,

So, I have been trying to figure this out. I know it's possible to impose restrictions on the coefficients of the independent variables. But, is there any way to have a constraint on the dependent variables of a model of equations, in which all the dependent variables must sum up to 1 by construct?

Thanks!

[EViews Gareth]

This makes very little sense. Can you explain what you mean in more detail?

[jacarandas]

I've a set of ratios that sums up to 1 in history. I would like to forecast them ahead of time. But, how do I ensure that these forecast ratios in the model sums up to 1 and remain positive? Is there an estimation method I can use or an identity I can append to my model?

Currently, I am using a least square estimation to forecast ahead.

I know I am able to restrict coefficients on independent variables. But, I don't know if the same thing works for forecasts/dependent variables.

Thanks!

[KrilleJ]

Perhaps logit-type specifications could work:

fraction_i=exp(x_i*beta_i)/exp(sum_i x_i*beta_i)


/K