Abstract: Based on quantile regression models, the correlation between volumeand yield on different levels is studied.In order to eliminate the influence of yield's level to the volume, an instrumental variable and instructa IVQR model are introduced, based on which the effect of volume on yield of different levels can be analyzed objectively.Monte Carlo simulation shows that IVQR estimator has less bias and stronger robustness.In empirical analysis,it is found that yield and volume are positively correlated if the yield is on a high level, the higher the level is,the larger the correlation will be;and that yield and volume arenegativelycorrelated if the yield is on a low level, the lower the level is,the larger the correlation will be.We also find that the volume's effects on yields of most levels are similar.The outcome simultaneously shows that IVQR can dispose the endogeneity in models.

Key words: yield, volume, quantile regression models, IVQR Models