The present research aims to modify the standard measure of the bullwhip effect (BE) in supply chain management (SCM) to evaluate and capture the particular characteristics of non-stationary demand time series. It also focuses on assessing different demand time series in value-added services supply chain (VAS-SC) sector, which has recently been very popular in the mobile telecommunication market.

To analyse different demand cases in this research, the generalised autoregressive conditional heteroskedasticity (GARCH) models are employed to analyse the time-series variance. Moreover, out-of-sample demand forecasting is used to compare the power of each model in predicting VAS-SC demand data. The results show that the modified BE measure fluctuates dramatically over time.

Furthermore, running the GARCH-class models for several mobile VAS (MVAS) cases shows that the best model varies for each case. Finally, this paper proposes that the best model can be judged by statistical testing of the results of the modified BE ratio.