Forecasting monthly sales of cell-phone companies - The use of VAR model
Journal
Second International Conference on Innovative Computing, Information and Control, ICICIC 2007
Date Issued
2008
Author(s)
Abstract
Forecast of earnings is one of the major tasks for financial statement analysts, and sales forecast is the most important step in the process of predicting earnings. Taking one large cell phone company in Taiwan as example and by the use of monthly data to fit a proper VAR model, we found that, when forecasting monthly sales, number of clients is a significant forward-looking information for financial statement analysts of the mobile industry. Compared to a model with only past sales as predicting variable, the forecasts generated by the bi-variate (monthly sales and month-end number of clients) model in this paper reduce the forecasting error by almost 30%. The evidence from the two variable VARMA model also support our hypothesis that there is a bidirectional dynamic relationship between sales and number of clients, i.e., both sales and number of clients are leading indicators of one another. The purpose of requiring listed firms to publicly announce their monthly sales before the 10 th day of the following month in Taiwan is of course to provide investors and analysts with relevant information for them to make rational decision. Monthly sales numbers are relevant because users of this financial information can predict future sales based on their past history. In this paper we target on a large cell phone company in Taiwan to demonstrate that a similar timely disclosure of number of clients should also be required in additional to the monthly sales. Besides, the SEC should careful examine for all industries that whether more forward-looking information should be made available to the public.
Type
conference paper