首页 | 本学科首页   官方微博 | 高级检索  
     检索      


A comparison of the forecasting performance of WEFA and ARIMA time series methods
Institution:1. Dept. of Pharmaceutical Engineering, College of Chemical Engineering, Qingdao University of Science and Technology, Qingdao, China;2. Institute of Pharmacy, Dept. of Pharmaceutics, Biopharmaceutics and NutriCosmetics, Freie Universität Berlin, Berlin, Germany;3. Advance Pharma GmbH and Steiner Deutsche Arzneimittel GmbH, Berlin, Germany
Abstract:This paper examines the forecasting performance of the Wharton model (MARK III) over the period 1973 through 1975 and compares it with that of ARIMA models' performance over the same period. Despite strong intimation in the literature to the contrary, we find that this econometric model, at least, exhibits greater accuracy in every respect relative to ARIMA methods, in terms of its forecasts cum constant adjustments. When constant adjustments are disallowed then its forecasts are still more accurate than ARIMA forecasts over a 4- and 8-quarter forecasting horizon, but less accurate over a 1-quarter horizon. The comparison was carried out over twenty three macrovariables, under a slight handicap for the Wharton Model, in that the latter's parameters were estimated over a sample ending in 1969.3 while the ARIMA models were reidentified and reestimated as of the quarter immediately preceding the forecast.
Keywords:
本文献已被 ScienceDirect 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号