中文摘要：

考虑到货币需求与其影响因素之间复杂的关系，基于支持向量分位数回归（support vector quantileregression, SVQR） 模型，文章研究了货币需求及其影响因素之间的非线性依赖关系，给出了货币需求条件密度预测方法，并将其与传统的线性分位数回归模型进行了比较。选取中国2004年1月至2014年12月期间工业增加值、消费物价指数（consumerpriceindex，CPI）、利率与M1的月度数据进行实证研究，结果表明SVQR模型不仅能够很好地拟合货币需求，而且能够给出准确的概率密度预测结果。

英文摘要：

Considering the complex relationship between money demand and its influence factors, the nonlinear dependence between money demfind and its influence factors is discussed based on the sup- port vector quantile regression（SVQR） model, and a method of money demand conditional density prediction is proposed. The SVQR model is also compared to the traditional linear quantile regression （LQR） model. The empirical study is conducted based on the monthly data of three factors including industrial added value, consumer price index（CPI）, interest rate and the monthly data of M1 from January 2004 to December 2014. The results show that the SVQR model not only can fit well money demand, but also can give accurate conditional density prediction results.