The Empirical Analysis Of China's GDP Time Series Based On Structural Breaks Theory

Gross domestic product (GDP) refers to a period of time (a quarter or a year), a national or regional production of all resident units in all the final results. Measure of a country is often considered an important indicator of overall economic strength. It can not only reflect a country's economic performance, but also reflect a country's strength and wealth. This indicator in our national economic accounting system is an extremely important position, for judging whether the healthy development of the national economy plays a key role. Therefore, accurate analysis and forecast GDP has important theoretical and practical significance.Time series refers to a phenomenon of a certain statistical indicators at different times on different values, arranged in chronological order to form sequences. Time series analysis of dynamic data analysis is an important way to deal with it as the theoretical foundation of probability statistics to analyze the random data sequence (or sequence of dynamic data), and its mathematical model, and further used to predict, since adaptive control and other aspects, is a very high practical value of applied research. Time series forecasting methods is through the phenomenon of the sequence of historical data reveal the variation over time, the law extends to the future, thus to predict the future of the phenomenon. The traditional time series analysis application in the economy is mainly deterministic time series analysis methods, including exponential smoothing, moving average, time series decomposition. With the social and economic development, many uncertainties in the economic life of the increasing influence, must be attracted widespread attention. ARMA model by the American statistician George EPBox with British statisticians GunlymM.Jenkins's, also known as B-J model, it is a dynamic model is a dynamic description of the random process is used when both AR MA method and a model. However, because of living in a lot of data is a non-smooth, making the need for modeling the original data before differential treatment, it is time series analysis of the most commonly used model is the ARIMA model.This paper choose from 1952 to 2006,55 years of GDP data in China as the research object, first of all, the establishment of a conventional time series analysis ARIMA (2,1,0) model and ARIMA (0,1,2) model, and then a more rational structure of the mutant containing the trend stationary model, and mutations were analyzed included the slope and jump mutation model for two different mutations and found that the optimal model for the mutation point in 1978 occurred in a smooth slope change in the trend model. Then use the residuals established a time-series model, after comparison test showed that the optimal model for the residual ARMA (1,2), then the estimated value of the residuals contain structure into the trend stationary series model of mutation, and ultimately reduced from a forecast of China's real GDP value. Through the forecast, inspection, comparison, contains the trend of structural change stationary sequence model in China from 1952 to 2006 GDP data simulation, the effect of forecasting the optimal.