The estimation of the derived demand schedule for raw wool using the Juster Scale : a thesis submitted in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Marketing, Massey University
In this thesis the results of a two-season experiment using the Juster scale to elicit subjective probability estimates of raw wool purchases at auction in New Zealand for five discrete classes of wool are presented. The approach, which involves estimating the rate of purchase of a commodity at various prices, represents a radically different method of estimating demand slopes and price elasticities. Over sixteen four weekly periods during the 1991-92 and 1992-93 wool seasons, a panel of eleven buyers from wool exporting firms in New Zealand were interviewed to obtain four weekly raw wool auction purchase forecasts. The results suggest that the use of a purchase probability scale to develop derived demand schedules for raw wool is possible. An average aggregate price elasticity of demand estimate of - 4.4 was generated for 1991-92 and - 4.6 for 1992-93. These values tend to be a little higher than those generated using an econometric approach and possibly reflect the nature of the wool market (i.e. falling prices and oversupply) and the higher 'information' content associated with the data generated through a survey instrument. It is shown that the panel's forecasts of aggregate wool purchases were reasonably accurate with an under-estimation of 8.3% and 12.9% respectively for 1991-92 and 1992-93. There was consistent under-estimation of aggregate purchases in the fine and fine-medium groups and consistent over-estimation of aggregate purchases in the coarse group. It is quite evident that the ability to forecast purchases within particular micron groups is fraught with difficulty due to the substitutability of wool types between the margins. Overall, the errors tended to fall over the study period reflecting, in part, a growing confidence by the respondents in the use of the survey instrument. An analysis of the qualitative data concurrently collected with the probability survey revealed a great degree of uncertainty and error in variables thought to be 'controllable'. The conclusion is reached that a great deal of the error in the results using the experimental survey instrument is a function more of uncontrollable external factors, rather than of the survey process.