The relationship between the revenue being forecast and the predictors of the revenue must be very strong. Moreover the shape of the relationship should be established. For example, in a study of the water/sewer rate structure in a Florida local government, I discovered a strong relationship between revenues from water charge and revenues from sewer charge: Sewer Charge = α + 2 × Water Charge while a is a constant. I can use this formula to forecast the revenues from the sewer charge.