Evaluation of consumer demand for selected end-use markets for cotton

Abstract

The U.S. cotton industry is facing a rapidly diminishing share of the domestic and foreign textile markets. To become more competitive in these markets, the textile industry should know where consumer demand is being directed. The objectives of this study were to quantitatively examine the demand for cotton in twelve selected end-uses from 1973 to 1997. Four structural models were developed to explain demand, two of which were built using the results from directed graphs. Changes in fashion and governmental policy were also assessed during this time period. Economic theory and prior literature suggested that the following factors influence demand at the end-use level: Disposable Personal Income, the Consumer Price Index for Apparel and Upkeep, cotton fiber prices lagged one year, polyester fiber prices lagged one year, and population segmented by age and gender. Regression results indicate that the Consumer Price Index, the lagged polyester prices, and the population variables most significantly contribute to the demand for cotton. A negative polyester coefficient was associated in all cases of significance, characterizing cotton and polyester fiber as complements in end-uses. Out-of-sample forecasts were generated for the years 1993 to 1997, and then evaluated using the Theil's U statistic, the Root Mean Square Error, and the Mean Absolute Percentage Error. The forecasts from the full models, those with all variables, outperformed the directed graph models in terms of predictive power. Differences in significant variables and variability in forecasting accuracy among the different end-uses suggests that there is an inherent difficulty in modeling the demand for a fiber in its end-use. A common saying in the fashion industry is that "the only constant is change." A constantly changing industry thus presents difficulty in quantifying, modeling, and forecasting.