1997 年 27 巻 2 号 p. 49-57
In recent years, applied time series analyses have increasingly relied on univariate time series models of the Box-Jenkins type. When economists and forecasters apply ARIMA models to their data, they frequently fail to take advantage of the qualitative information embodied in the results. This paper provides an example of how such information acquisition may be accomplished. The models were originally proposed as a means to develop forecasting mechanisms for the generation of expected prices in an analysis of fresh vegetable market supply functions. In this paper, a technique is presented showing how these same models provide information about market structure. First, a formal statistical test is proposed. Second, when this test cannot be applied, intuitive observations on general model characteristics are proposed.