On the Effects of Missing Values on Estimates of Trend Parameters and Seasonal Indices in Descriptive Time Series Analysis

In literature, when missing values are observed in a series, emphasis has always been on obtaining the estimates of the missing values while little or no attention has been given to assessing the effects of the missing values on parameter estimates. The aim of this study therefore is to at assess the effect of missing values on the estimates of trend parameters and seasonal indices in descriptive time series analysis when trending curve is linear and the decomposition model is additive. Estimates of trend parameters and seasonal indices were obtained using descriptive time series methods, while the performances of the estimates in the presence and absence of missing values are assessed using the values of summary statistics (MSE, RMSE and MAE), based on deviation of estimates from the actual parameters used in simulation. The results show that estimates of the trend parameter, error mean and standard deviation appear not to be affected when the number of missing values is less than ten. Estimates of seasonal indices also seem to be affected only slightly. For ten or more missing values, the summary statistics are only slightly higher especially when the missing follow consecutives. Specifically, the study found that trend parameters, error mean and error standard deviation used in simulation are recovered better than seasonal indices. It has therefore been recommended that when using the descriptive time series methods to obtain estimates of trend parameters and seasonal indices in presence of missing values there is no need to obtain estimates of the missing values first.