Forecasting models for the U.K. pig sector
In this thesis, forecasting models for the UK pigmeat sector are built using various methodologies with particular interest being paid to the relative forecasting ability of time series models compared with the performance of biological and econometric methodologies. The main determinant of the supply of pigmeat in the UK is the size of the breeding herd, the quantity of meat itself being directly attributable to the number of fat pigs slaughtered and to a lesser extent cullings of older sows and boars from the breeding herd. These three key variables are the ones modelled in this thesis. Prior to building forecasting models an explanation is given of the system underpinning the pig sector, in terms of the biology of the breeding herd pig, the mechanism of how supply responds to prices, and consideration of the well documented 'pig cycle'. Thus, the workings of the biological and economic mechanisms are described in the context of an equilibrium framework before the relevant models are built. Having built the various models, their relative forecasting performance is measured by consideration of the size of the forecast errors and the ability of the models to forecast the directional movements of the actual series in a specified out-of-sample period. In the concluding chapter, suggestions are made as to how the models might be developed further and how the various approaches might be combined into a single forecasting model. The availability of data has an important influence on much of the model building methodology and forecasting analysis. Consideration is given at various points in the thesis to circumventing these restrictions.