| Cointegration vector is a special unit root process,there is the long-term and balanced relationship between the variable.However,in the real economic problems,the financial and economic data often has the statistical characteristics of leptokurtic and heavy-tailed.So it is difficult for the financial and economic data to fit with Gauss distribution.In recent years,the research of heavy-tailed series has been one of the hot issues in the field of statistics.The asymptotic distribution of the cointegration test statistic for the heavy-tailed process contains an incalculable indexα.To solve this problem,this paper provided bootstrap sampling algorithm and block-bootstrap sampling algorithm to calculate the critical value of the cointegration test statistic for heavy tailed process without estimating the heavy-tailed index.This paper includes six chapters:Chapter one briefly introduced the research background of the cointegration test and the heavy-tailed series.Chapter two mainly introduced the heavy tailed distribution,the cointegration process and the related knowledge.Chapter three introduced the bootstrap sampling algorithm of the cointegration test for the heavy-tailed process;proved the rationality of the algorithm;demonstrated the effectiveness of the method by Monte Carlo simulation.Chapter four constructed the block bootstrap sampling algorithm of the cointegration test for the heavy-tailed process;proved the rationality of the algorithm;demonstrated the effectiveness of the method by Monte Carlo simulation.Chapter five used two kinds of methods to test the cointegration relationship between stock index and economic growth in china.Chapter six summarized the main contents of this paper;proposed the shortcomings of the two algorithms;described the content needed to be improved. |
