The problems caused by oil price shock have been remained a significant concern for all oil-lconsuming countries since the 1970``s. Thus there have been developed many different methodologies used to address particular questions relevant to price shocks. There exist some supply disruption models analyzing structural changes in oil price movement, i.e. oil price shocks. In this thesis, we examine Vector Autoregressive model to show this problems. But as a result, supply disruption doesn``t explain well price shocks. Instead, we show that as a univariate time series, underlying process of crude oil price has the property of exponential smoothing process which is useful for data analysis. This thesis has an object to demonstrate a practical forecasting model in crude oil price which has structural changes. Especially we concern shortterm analysis in crude oil prices in which shocks happen sequentially. To deal with these problems, dynamic linear model with dummy variable is proposed. Moreover, for the purpose of analyzing crude oil price and model approximation, random walk test for crude oil prices and simple numerical approximation method are used. Some random walk tests propose an adequacy for analyzing oil price as an univariate time series. In the empirical application to the weekly crude oil-Arabian Lightprice, though it is difficult to describe the price shocks caused by various environmental factors, the proposed model gives good materials to analyze short-term price changes in which we concern with the detection of structural change and forecast.