Abstract: This paper makes a stationary VAR model for the amount of loans and the real GDP series in China, and an AR model with a trend term and a constant term for the contribution series, which measures the contribution of the loans to the real GDP in China. The result of the first model shows a significant interaction between the increase rates of loans and the real GDP, and the second shows that the linear relationship between loans and the real GDP has a significant downward trend and a long memory property，both of which mean that, in the short run, increasing loans is effective on the real economic growth, while in the long run, it induces the inflation additionally. These conclusions give a good support on the effectiveness of changing loans as a monetary policy tool, and also an empirical proof on testing the currency-neutral in China indirectly.

摘要： 本文以全国金融机构各项贷款余额的季度时点数据，以及经季节调整、并利用不变价格国内生产总值指数全面剔除价格影响的实际季度GDP数据为观察序列，通过对二者建立滞后两期的稳定VAR模型，得到二者增长率之间的短期动态影响关系；同时，使用带有趋势和漂移项的AR模型来拟合贷款余额总量对实际GDP的贡献率序列，得到二者之间的长期线性关系是带有逐渐下降趋势并具有长记忆性时变特征的结论。该结论表明，在短期内贷款总量对实际经济增长具有刺激作用，但长期内这种刺激作用会引致通货膨胀的发生。本文结论为验证通过缩放信贷规模来调节经济的政策手段的有效性提供了有力支持，也为间接验证货币中性理论在中国的成立性提供了实证依据。