Tecniche statistiche per effettuare previsioni economiche: un caso di studio

Moschini Cristiano MSc in Agribusiness Management __________________________________________________________________________________________________________________ 7 INTRODUCTION Forecasting the future value of key variables is an integral part of the decision-making activities of management. Organisations establish goals and objectives, seek to predict environmental factors and then select actions that they hope will result in their attainment. A wide variety of forecasting methods is available to management. These range from the most naïve methods, such as use of the most recent observation as a forecast (naïve expectations), to highly complex approaches, such as an econometric estimation of a system of simultaneous equations. The purpose of this paper is to examine two techniques that can be used to forecast, particularly, in the short-run period using very limited data resources. This report then presents and attempt to build forecasting models, using a time series of data of the Italian poultry market of Forli’. Models can be built in different ways, depending on the data gathered and the statistical tools used, statistical tools which allow the prediction of future movements in a variable (i.e. price) by solely analysing its past behaviour are demonstrated. In other words, models are constructed for the time series which do not offer structural explanation for the series’ behaviour in terms of other variable (e.g. using regression models) but which replicate their past behaviour in a way that may help me to forecast their future movements. These models are built by the application of statistical techniques, namely Exponential Smoothing and ARIMA modelling, available in software such as SPSS Trends and Microfit, which I have largely used in my analyses. This report begins with a critical review of the statistical techniques mentioned above in Chapter I and II. Chapter III deals with the practical application of the exponential smoothing technique to a series of data of the Italian Poultry Market of Forli’ while Chapter IV shows the