Abstract
Developing a dairy supply model assuming deterministic trend and seasonality a priori could lead to model misspecification. A structural time series methodology was used to examine the role of stochastic trend and seasonality in a dairy supply response model using quarterly time series data. Four versions of a time-linked model of dairy supply response were compared. A dairy supply model with stochastic seasonality and deterministic trend performed the best, as evidenced by diagnostic tests, goodness-of-fit measures, and forecasting accuracy. Our analysis suggested against the classic approach of incorporating the deterministic seasonality component in dairy supply models.