Market Analysis

"Chairman: Mr R. A. PLUMBRIDGE Rapporteur: Dr D. M. HAWKINSPapers:The relevance of computer methods to the economics of the mineral industry by R: A. PerlmanDynamic control of a mining enterprise by R. G. CoyleAn economic model of the mineral sands industry by G. P. A. HoweMr M. Splaine, in opening the discussion on Mr Perlman's paper, asked whether the model, as fitted by data up to 1963, predicted the rise in the price of copper in 1967, and whether the model could be used to capitalize on short-term price fluctuations.Answering the second question first, Mr Perlman said that the horizon used is six months to one year. In the very short term the activities of speculators obscured the effect of the fundamental econometric variables. This effect, however, was removed if one moved to a longer term, say, quarterly. In 1963 the model had predicted a price rise in 1964 and 1965 and a drop in price in 1966. The rise in 1967 had been caused by a strike, and imponderables such as large strikes were difficult to include in the model since they were unpredictable. Model simulation can be used to show the effect of strikes on price and the effect of small strikes is small.Mr B. R. van Rooyen said that because models are both calibrated and tested on historical data a bias was introduced which gave rise to an overoptimistic estimate of the accuracy of the forecasts. To this Mr Perlman replied that there seemed to be no alternative to the use of historical data for calibrating and testing models, and thus little hope of avoiding the bias. A second point made by Mr. van Rooyen was that manage¬ment tended to be suspicious of complex models and that frequently more heuristic models were sufficiently accurate and preferable for investment decisions. He added that if management were not involved in the model building the econometrician might easily go astray, Mr Perlman replied that symbiosis between data analysts and users of the model was essential and that the experience of the practitioner must be reflected in the model.Mr M. R. Fuller-Good remarked that a clear understanding of future price trends was essential for effective capital planning, and asked, first, how one went from the compiled data on supply and demand to the determination of their effect on price, record, whether necessary data, for example, capacities, could be obtained timeously, and, third, when strikes restricted supply, whether stockpiling by consumers induced large price fluctuations."