Since quantile regression estimation has advantages compared with other estimation methods in some respects, such as the overall’s distribution yield without specific assumptions and better describing the leptokurtosis and other data characteristics. In terms of financial risk measurement, VaR is essentially the value of rate of return quantile function, hence the quantile regression has a natural advantage to estimate VaR. This paper uses quantile regression and ARCH models to estimate VaR, and obtains the relevant results of model parameter estimation, proposing some prospect of the currently popular GARCH model. Volatility is one of the most important features of financial time series,financial time series exists non-stationary and serious heteroscedasticity, the traditional linear model can not explain financial data’s leptokurtosis, volatility clustering and leverage, and various types of conditional heteroskedasticity model can overcome these problems. Therefore, this article choose to estimate VaR by Quantile regression methods and GARCH Models, and obtaining the relevant results of model parameter estimation with the Shanghai Composite Index.The results show that the number of students last year and per capita household income significantly affect the number of students basicly. And quantile estimation can find at the low level of quantiles, per capita household income index, which does not has a significant impact on the number of students at the high level of quantiles, has a significant impact on the number of students. Quantile regression can dig out more informations than the least-squares. |