The Application Of Quantile Regression Model

Quantile Regression is the idea of Koenker and Bassett proposed in1978, which isin order to make up for the ordinary least squares (OLS) in regression analysis ofdefects. It provides for between conditional quantile regression relationship and thevariables of a basic estimation method. Its essence is that it through the quantile takesbetween0and1any value, regulating of regression plane position and the steering.This can make independent estimation of different quantiles of the dependent variable.It not only can be represent of all the information in the data, but also focus more onspecific regions of the data, such as extreme position data. Quantile regression is a semi-parametric model estimation. It does not need the data distribution type prior to makecertain assumptions. It has very good properties. The paper mainly uses the quantileregression method respectively in the commodity market economy, securities andfutures in the financial field of all relevant data for analysis research.Firstly, based on the semi-parametric theory of quantile regression andnonparametric model, which is based on the theory of local linear regression androbust regression LOWESS are compared. Changed the window width and variabledbandwidth local linear estimation and its properties are introduced. The robustestimation--LOWESS estimation of the specific steps in detail are also introduced. Andbased on our country of the resident consumption price index as the main indicator of inflation and the customs commodity import and export volume of the three regressionmodels are analyzed and compared. Analyzed the results of graphics, which isindicated the quantile regression estimation more robust.Secondly, based on the risk measure method of VaR model calculation, it usedsemi-parametric quantile regression model and parameter method of TGARCH modelof the Kupiec likelihood ratio tests for comparing failure rate. Based on the empiricalanalysis of China Growth Fund of the above two models are analyzed and compared.Drawn by Quantile Regression Prediction of VaR in the given significance level caneffectively measure the risk in securities market.Again, introducted to futures market in the price volume relation theory to theanalysis of status, according to the “Karpoff” asymmetric Volume-Price Relationhypothesis on Dalian Commodity Futures Exchange futures contracts on soybeanyield and volume of the corresponding processing, and uses the quantile regressionmodel on the yield and the logarithm of the volume analyzes the relationship. In the21branches site coefficient estimation value and the corresponding quantile regressioncurve graphic as a whole on the basis of the analysis, confirmed the futures market inthe presence of the quantity price " Qi Yang" and" price volume also rises"phenomenon.Finally, the conclusion of this paper was given, and proposed further researchdirection.