Detection of Shocks of Coin Price in Iran using Multivariate Time series

This study seeks to analyze the effectiveness of outliers types, shocks, innovational (IO), additive (AO), level shift (LS), temporary change (TC) in model assignment and multivariate time series, MTS, parameters. Repetitious approach given by Tessy (2000) is considered for detecting different types of outliers, AO, IO, LS, TC in multivariate time series model. The performance of the proposed method has been shown in comparing with classical one-variable approach suggested by Chen and Liu. Price of the golden coin can be a good example for the time series data subjected to the outliers. First, there is a highly dependence between the variations of price of the golden coin and the global price of the gold. Second, as the golden coin has been used as a gift in fiestas and national ceremonies, its price can be varied with different sorts of seasonal patterns. These and some other reasons lead us to possibly detect different shocks in the price of the, quarter, half and full, golden coin in Iran. The data were taken from the Bank of Markazi of Iran and includes 117 observations from Farvardin 1378 to Azar 1387. The data have been analyzed to detect outliers and compared with results obtained from the classical approach. The numerical results showed that the this approach is more sensitive than one-variable approach to detect outliers.