交互效应面板分位数回归模型的迭代估计

Iterative Estimation of Panel Quantile Regression Model for Interaction Effects

文章基于创新性交互效应面板分位数模型的设定,提出了一种新的迭代算法,蒙特卡洛模拟结果表明,与其他估计方法相比,该迭代算法有更好的有限样本性质,偏误较低且方差较小。由于个体效应及交互效应均允许与自变量相关,个体与时间的异质性得以更加灵活地表达,因此使模型具有更强的适用性。最后,将模型应用于我国房价的实证分析,并用迭代算法来对参数进行估计,回归结果表明,房价决定因素的效应存在异质性,且高房价地区的房价主要受到投资的影响,而低房价地区的房价的关键影响因素为人民生活水平和经济发展水平。

Based on the setting of an innovative panel quantile model of interactive effects,this paper proposes a new iterative algorithm.The Monte Carlo simulation shows that compared with other estimation methods,this iterative algorithm has satisfactory finite sample performance with low bias and small variance.Since both individual effects and interaction effects are allowed to be correlated with independent variables,the heterogeneity of individuals and time can be expressed more flexibly,which makes the model more applicable.Finally,the paper applies the model to the empirical analysis of housing price in China,and uses the iterative algorithm to estimate the parameters.The regression results show that there is heterogeneity in the effects of housing price determinants,and that the prices in areas with high housing prices are mainly affected by investment,while the key influencing factors for prices in low housing price areas are income and the level of economic development.