AN ANALYSIS OF FUTURES PRICES USING ARCH MODELS

Agricultural commodity prices are highly volatile and largely unpredictable, which makes it very difficult to extract statistically reliable estimates of price movements. Recently, agricultural economists have made extensive use of the time-series analysis to model agricultural price series (Bessler and Brandt, 1979; Harris and Leuthold, 1985; Shonkwiler and Spreen, 1982). There are a number of reasons for time-series approach to become popular. For instance, these models can be used to gain insights into the dynamic properties of complex systems and require less subjective judgment on the specification of models.' But the most important reason for the wide-spread use of these models is their forecasting accuracy. The improved forecasting accuracy obviously derives from optimal use of past information. However, these forecasting properties• have not been extended to predictions of variances.