Many variables have values that change with time such as the monthly unemployment figures, the daily production rates for a factory etc. The changing value of such variables over a period of time is known as time series. Thus time series is a sequence of values of some variables taken at successive time periods. Company: Orange is one of the world’s leading mobile communications companies, well positioned for the future. ‘Orange’ is the first choice in wirefree(tm) communications.
Orange innovations like simple Talk Plans that offered real value for money, per second billing, Caller ID, itemised billing free of charge, and direct customer relationships changed people’s attitudes about mobile communications. In 1996, Orange plc underwent its first initial public offering with the shares being listed on the London and Nasdaq markets. In August 2000, France Ti?? li?? com acquired Orange plc for a total consideration of i?? 25. 1 billion. Orange plc’s wirefree(tm) interests were merged with the majority of those of France Ti?? li?? com to form the new Orange SA group.
The Orange brand now operates in the United Kingdom, France, Switzerland, Romania, Slovakia, the Netherlands, Thailand, the Ivory Coast, the Dominican Republic, Cameroon, Botswana and Madagascar. The Orange group also has minority interests in Portugal (Optimus), Austria, (Connect Austria), and Mumbai/India (BPL Mobile). Orange makes a difference to people’s lives by creating simple and innovative services that help people communicate and interact better. Data Analysis and Time Series Graph The data below displays the total turnover (in million euros) of ‘Orange’ group on a quarterly basis for the period 2000 to 2003.
For the multiplicative model to be reliable, the actual sales as a percentage of the trend for the particular quarter should be same or approximately very close. After replacing the figures in the above formula we see that the Multiplicative Model gives us big differences in the percentages in the same quarter for the four years. For example in 2001 Q2, we get an effect of 99. 34% compared to 94. 89 % in 2002 Q2, which is not very close. With these major differences, the multiplicative model would not provide a very reliable forecast and therefore it is advisable to use the Additive Model of time series.
In table 2. 2 (Mean Seasonal Effect) we note that the seasonal effect is different for the same quarter in the four years. This is due to the random element. Thus the Mean Seasonal Effect (MSE) shown in table 2. 2 is not a perfect result and has a total error of 18. 08. So we need to correct this error by smoothening out the seasonal variations to arrive at values without errors using the additive model of time series. We therefore divide the error (18. 08) equally amongst the four quarters to arrive at an adjusted MSE. This will enable us to calculate a more accurate forecast for quarter 1 of 2004.
We then calculate the Residual Error, as shown in table 2. 1, in the following way: Using the additive model of time series we get: Y = T + C +S + E where: Y = Actual sales T = Long term, secular, trend C = Cyclical variation S = Seasonal variation E = Residual error After calculating the trend and the seasonal effect, we substitute the information in the formula (Y = T + C + S + E) to get the Residual errors. Here the Cyclical variation is zero. Forecasting Our purpose of the time series analysis above is to use the results to forecast future values of the series using the decomposition model.
The procedure for this is to extrapolate the trend into the future and then apply the seasonal component to the forecast trend. To calculate the forecasted sales for Orange group for Q1 of 2004 we take the difference between the last trend (2003) and the first trend (2000) as shown in table 2. 1 (4476. 5 – 3119. 38) which is1357. 13. This difference is then divided by 11 (no. of changes between the quarters) to give us 123. 38. This is the difference on an average between the quarterly trends. which is then added to the last trend figure (4476. 5) to give us the next trend figure of 4599. 88 (4476.5 + 123. 38).
Thus the forecasted sales revenue for Orange in quarter 1 of 2004 is 5172 million euros with the forecasted trend being 4599. 88 million euros as shown in the graph above. We have now evaluated and analysed the sales data for Orange over 16 quarters from 2000 to 2003. By suggesting an approach to the analysis we have also projected the forecast for quarter 1 of 2004 and we shall now move on to comparison. After calculating the forecast I compared the forecasted figure to the actual sales for the first quarter of 2004 which was given as 5000 million euros as shown in the graph above.
As we can see, this is relatively optimistic and close to our calculations. The reason for the forecast being close to the actual sales, I believe, is the use of additive model of moving averages, which smoothens the data leaving no room for seasonal errors thus arriving at a seasonally adjusted forecast for quarter 1 of 2004. Referring to the actual sales in Q1 of 2004 I noticed that the merger with France Telecom significantly improved the group’s profitability and operating margins. Solid resilience of Fixed Line, Distribution, Networks, Large Customers and Carriers segment also enabled an increase in the consolidated revenues of Orange.
Methods of forecasting based on an analysis of historical data can only provide a valid basis for predicting the future to the extent that the factors remain unchanged and the same general trends are followed in the future. There are changes and seasonal errors in the trend, which could arise due to external factors that are not taken into consideration i. e. market demand, consumer preferences, competition or changes in legislation etc. However in calculating and evaluating the forecasted sales revenue for Orange, we have taken into account the nature of residual errors which gives us a more accurate result.
Also short term forecasts are more reliable than long term forecasts as seen in our comparison of Actual v/s Forecast for quarter 1 of 2004. Thus the above approach takes into consideration seasonal errors, those we cannot control and also distributes the same between all quarters to arrive at an adjusted seasonal variation allowing an error free and a more precise forecast. We can therefore conclude that forecasting is a method by which a business can evaluate its performance which is used as s decision support process and is almost never the actual outcome but provides an idea as to what is expected under normal course of actions.
Morris, C. (1983) Quantitative Approaches in Business Studies 4th edition, Great Britain: Pitman Publishing. Oakshott, Les. (1998) Essential Quantitative Methods for Business Management and Finance 2nd edition, New York: Palgrave Publishers Ltd. Friend, D. V. (1987) Quantitative Methods (Longman exam guides) 2nd edition, United States of America: Longman Inc. , Information and sales revenue data of the Orange group (2004), Available: www. orange. com Last accessed: 18th November 2004