Application And Comparative Study Of Multi-Scale Model Family

The financial market changes rapidly and there are many random influencing factors.Most of the existing researches on its complex volatility set up various financial market volatility spillover models from a single time scale and estimated the volatility spillover effect of the financial market.However,the volatility correlation of a single time scale does not mean that this relationship exists on all time scales.Therefore,most of the existing volatility spillover models are difficult to fully and accurately grasp the volatility spillover effect between financial markets and form an effective decision scale.In this thesis,wavelet analysis and multi-scale theory and method are introduced,maximum overlapping discrete wavelet transformation is selected and combined with MSV model,and single-scale MSV model family is extended to multi-scale MSV model family to measure the volatility spillover effect between Chinese stock market and U.S.stock market.At the same time,according to DIC rule,this thesis makes a comparative study on multi-scale MSV model family and obtains a series of results.First of all,this thesis selects the weekly closing quotation data of Shanghai Composite Index and S&P Index,selects the optimal wavelet basis,uses the maximum overlapping discrete wavelet to transform and decompose the time series of return rate of each index,and analyzes the wavelet variance and wavelet correlation coefficient,draws the following conclusions:(1)The optimal wavelet bases of Shanghai Composite Index and S&P Index are db2 and db4 respectively.(2)Chinese stock market and U.S.stock market volatility of each scale order from large to small is: Chinese stock market in short term,Chinese stock market in short-mid-term,U.S.stock market in short-term,Chinese stock market in mid-term,U.S.stock market in short-mid-term,Chinese stock market in mid-long-term,U.S.stock market in mid-term,Chinese stock market in long-term,U.S.stock market in mid-long-term,U.S.stock market in long-term;(3)At the same scales,there is a certain correlation between the Chinese stock market and the us stock market yield;The correlation coefficient of Shanghai stock index on scale 2,3 and 4 is higher than that on scale 1 and 5,indicating that for Chinese stock market and us stock market,the correlation of medium-term investment is higher than that of short-term and longterm investment.Secondly,this thesis uses the multi-scale MSV model family to make an empirical study on the volatility spillover effect of each scale between Chinese stock market and U.S.stock market.Among them,most of the multi-scale MSV models are convergence.But with the increase of scale,the convergence of the model becomes worse and worse.From the model comparison results of multi-scale MSV model families,the DGC-t-MSV model of each scale has the best fitting effect,but the complexity is the highest.Finally,the conclusions of the study on the spillover effect of volatility on each scale between China's stock market and the U.S.s0 tock market are as follows:(1)The volatility between the Chinese stock market and the U.S.stock market shows a strong dynamic correlation,and the dynamic correlation between the volatility on different scales shows certain long-term memory characteristics;(2)The volatility of Chinese stock market and U.S.stock market is relatively persistent,both of which have the characteristics of volatility aggregation;(3)There are granger causalities between the Chinese stock market and the U.S.stock market,the Chinese stock market is more the source of volatility spillover.