Investigating the price of the New Zealand wool clip using modelling approaches

Abstract

The price of the New Zealand wool clip has steadily decreased over the years and the incumbent practitioners are looking for ways to increase the price. The price of the wool is determined by the marketing approaches that are adopted. An auction system is one of the wool marketing approaches and a total of 45% of the New Zealand wool clip is traded via auction (WIN, 2007).This auction is the English open out-cry public auction which has been running for the last 150 years in New Zealand. The key players of the auction (buyers, brokers and growers) as well as the New Zealand government are trying to understand the reasons behind the steady decrease in the price of the New Zealand wool clip. Further, the incumbent practitioners are questioning the auction system and looking for alternative ways of wool marketing. In this study, an attempt is made to model auction data. The data is available from the only auction centre in the South Island of New Zealand in Christchurch. Analytical approaches are used in developing the models from the data. Before fitting the data into the models, the database is thoroughly cleaned and a necessary and sufficient set of the parameters is produced by developing the approximation equations. The probability distributions of the parameters in the data base are observed. Best fitted distributions for all physical parameters are observed in terms of three measures namely mean square error, the chi-square and Kolmogorov-Smirnov (K-S) goodness-of-fit hypothesis tests. The cleaned database is used in the models. The models are developed to predict the price of the different types of wool. The price from the auction centre and the predicted price are compared. The models based on linear regression analysis, multiple regression analysis and principal component analysis are developed. The developed regression models are tested for the goodness-of-fit against coefficient of determination, R² , which describes the proportion of variability in the data set accounted for by the model. The model from the principal component analysis is used for the possible dimensionality reduction in the data set. The developed models and the physical parameters of wool are discussed. The role of physical parameters that account for the price of wool in the auction system is investigated. It is hoped that the analysis of auction database and the developed models will help the practitioners of the New Zealand wool industry in better understanding the role of physical parameters of wool and the price formation of it.