Modelling swine supply response using a structural time series approach

Abstract

Swine supply response was analyzed using a structural time series model (STSM) with both seasonal and yearly effects introduced stochastically. Swine production is divided into four production phases. Parameters of each phase were then estimated using a structural time series model. Out-of-sample forecasting robustness of the STSM was compared against a commonly used deterministic trend and deterministic seasonally (DTDS) using root mean square error and mean absolute percentage error criteria. The STSM model produced lower RMSE and MPE values than those obtained using a DTDS modeling approach.