The Research Of Seasonal Index's Estimation And It's Application

In the real economic life, there are always some regular cyclical changes which appear in every year because of the influence of climate condition, production condition, holiday factor, resident life and folkways. The changing regulations are similar in every year, which are always called seasonal variation, sometimes also called cyclical variation. In factor, most economic time series data have the ingredient of seasonal variation. Nowadays, there are a lot of researches about economic phenomena which including seasonal variation, but most researchers regard the ingredient of seasonal variation as"data interruption". Researchers primary research how to weed these data, which is namely"seasonal adjustment", and research the forecasting of seasonal time series. However, in a large amount of conditions, we want to know the magnitude of seasonal variation in economic phenomena, so that we can take according measures to cope with the seasonal variation in different seasons. Seasonal index is just the index to measure the magnitude of seasonal variation. So researching precisely the seasonal index have grate significance, not only in theory, but also in application.The basic train of thought about seasonal index is averaging method according season, the research of this paper also develops based on this train of thought. But the economic data in real economic life are various, so averaging method according season can't be applied directly. At the some time, the seasonal index from averaging method according season is just a descriptive statistic, whose application in economy statistic analysis is limited. So we must research the deductive statistic about seasonal index.On the base of averaging method according season, this paper put forward three kinds of seasonal index estimation methods according different data, and each method can execute deductive statistic.The first and second methods both make use of seasonal dummy variable model to estimate the seasonal index. By use of the parameters estimation of seasonal dummy variable model to estimate the seasonal index indirectly. Both of the two methods can not only estimate the point estimation and interval estimation of seasonal index, but also execute the deductive test about whether the economic time series have seasonal variation. In these two method, the former method suit to the economic data from long years in a row, and the later suit to the short-term panel data. Both of the methods estimate and analyze the seasonal index by use of the thought of regression.The third method of this paper estimates the seasonal index by use of sampling estimation, estimating the totality seasonal index indirectly according to sampling data, which not only lowing the request of data largely, but also having larger field of application.