and finally, (c) A sales forecasting system allows the manager to experiment with alternative policies and to consider the effects of various possible levels of future sales. This simulation of alternative futures helps management to examine the ‘sensitivity’ of various management plans and reconsider their effect on the financial and competitive viability of the organization. APPENDIX | A brief presentation of FORSYS X,, the time series, is decomposed as follows: X= MS, +e (a) The mean, M,, is defined by a moving average process which is basically of exponential smoothing, type. For instance, for a linear model, Af, is defined as: M,=2(MI,) — M2, 2) where Mi, ~ 5 get ve of] =4,- ”” Q) o M2,= y M1, 0% -0 [] (1 0) (4) The smoothing constant «, is given by: O, =M%, + Ax, The values of «, vary as follows: ot” ty, = ap! Aa, = Kap ltl — ge, ph) where of" is a measure of the stability of the series and is defined as: MAI ™, and where MAD, = [ely +(1 —)MAD,_; co? is a normalized measure of the randomness of the series. It is defined as g x2 ]_f i haa | and finally, 2¥ is a tracking signal defined as follows: Hoe ey = wa, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.where BB ls) +e where ys, can be thought of as the coeflicient of decay, that is: 5, = Youll — poh] The seasonal coefficients are found by an exponential smoothing process similar to that of (3) and (4) which is: 5X, S,= x M,- BT By. where By = Bop The forecasting of the series is given by combining the components of (1), that is M, and S,. This results in the following projections: SD. = M(a), + KT(a), + 87 0(a), RO, = M@), + KT(?), + 70), XD, = M(at), + eT), Finally, the forecasts are found by Xian = OO dSian For more details, see Lewandowski (1979). REFERENCES Box, G. E. P., and Tino, G. C., ‘Intervention analysis with applications to economic and environmental problems’, Journal of American Statistical Association, 70 (1975), 70-79. Lewandowski, R., ‘Ein voll adaptationsfiihiges Modell zur kuraftistigen Prognose’, A KOR-Tagung, Aachen (1969). Lewandowski, R., La Prévision @ Court Terme, Paris: Dunod, 1979. Makridakis, S., et al., ‘The accuracy of extrapolation (time series) methods: results of a forecasting competition’, Journal of Forecasting, 1 (1982), 111-153, Wheelwright, S. C., and Clarke, D, G., ‘Corporate forecasting: promise and reality’, Harvard Business Review, November-December (1976). Author's biography: Rudolf Lewandowski, born 1938 in Valencia/Spain, studied mathematics and economics at the Sorbonne in Paris, and the universities of Bochum and Bonn and obtained degrees in both subjects. He came to the Federal Republic of Germany in 1962 as a mathematical adviser for a big French company. In 1966 he became the manager of the operations research, economics and marketing department in a leading software firm in West Germany, and founded in 1973‘ MARKETING SYSTEMS’, and has been its general manager since. In 1973, he obtained a doctoral degree in economics at the Sorbonne. He has published papers on Markov processes, forecasting methodology and forecast systems, and is author of Prognose und Informationssysteme und ihre Anwendungen, 2 vols. (De Gruyter-Verlag, Berlin 1974 and 1980), and La Prévision a Court Terme (Dunod, Paris 1979). He has lectured at various European universities, and is a member of several European Marketing Associations. Author's address: Rudolf Lewandowski, Marketing Systems GMBH, Postfich 230109, Hunsruckstrasse 9a, D-4300, ESSEN 1 (Bredency), Germany. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.