地理科学进展 ›› 2013, Vol. 32 ›› Issue (3): 372-380.doi: 10.11820/dlkxjz.2013.03.006

• 城市地理 • 上一篇    下一篇

基于分位数回归的江苏省城市化动力因子分析

牛品一, 陆玉麒, 彭倩   

  1. 南京师范大学地理科学学院,南京210023
  • 收稿日期:2012-07-01 修回日期:2012-11-01 出版日期:2013-03-25 发布日期:2013-03-25
  • 通讯作者: 陆玉麒(1963-),男,教授,博士生导师。E-mail:luyuqi@263.net E-mail:luyuqi@263.net
  • 作者简介:牛品一(1988-),男,硕士研究生,主要研究方向为空间规划与区域经济。E-mail:niupinyi@126.com
  • 基金资助:
    国家自然科学基金项目(41071084)。

Driving factors of urbanization in Jiangsu Province based on quantile regression

NIU Pinyi, LU Yuqi, PENG Qian   

  1. School of Geography Science, Nanjing Normal University, Nanjing 210023, China
  • Received:2012-07-01 Revised:2012-11-01 Online:2013-03-25 Published:2013-03-25

摘要: 识别城市化动力机制对城市化持续、合理发展有着至关重要的作用。文章对国内外研究进行综述发现,以往对城市化动力机制的计量研究多采用最小二乘法(OLS),对城市化的发展动力判断过于粗略的问题。文章首先通过变异系数、核密度估计方法,描绘出江苏省城市化发展分布状况,发现江苏省历年城市化相对差异基本稳定,城市化高速发展的同时,呈现出“双峰趋同”的态势。分位数回归方法能精确地描述解释变量对被解释变量在特定分位点的边际效应,因而可使用分位数回归与OLS回归对比分析影响城市化发展的动力因子。以2005年为界,将研究时段分为两个阶段,前后城市化动力因子存在明显差异OLS回归发现,行政力对城市化的影响作用越来越小,而市场力量、外向力等因素对城市化的影响相对增大。分位数回归结果表明,行政力量对城市化高的地区影响较弱,外向力对城市化水平高的地区的影响要大于城市化水平低的地区,市场力对城市化的影响随着城市化水平的提高则经历了一个先增大后减小的过程,产业结构的影响复杂多变。由此发现,江苏省城市化在不同阶段发展路径有不同的侧重点。

关键词: 动力因子, 分位数回归, 核密度, 江苏, 人口城市化

Abstract: Study of the dynamic mechanism of urbanization in China is very important for the sustainable and rational progress of urbanization. The most commonly used statistical method in quantitative studies of urbanization abroad and at home is OLS regression (ordinary least squares), which only provides a rough assessment of driving forces of the dynamic process of urbanization. In this paper, by calculating coefficients of variation and using kernel density estimation method, we first describe the distribution of the counties in terms of urbanization levels, based on the statistics of urbanized population during 2000-2009 in Jiangsu Province. The results indicate that the differences in urbanization level among the counties remain unchanged throughout the years. With the overall rapid process of urbanization, the distribution starts to exhibit a“two peaks co-existing”pattern, on the low end and high end of the urbanization level. Quantile regression allows one to obtain a more comprehensive picture of the effect of the predictor variable on the response variable, thus we conduct a comparative analysis on the factors driving urbanization, using both quantile regression and OLS regression. In the study, the process of urbanization has been divided into two phases, before and after 2005, and OLS analysis has found that administrative power has decreasing effect on urbanization while the effects of market power and centrifugal force are relatively increasing. Quantile regression analysis provides more reliable and more detailed information, which shows that administrative power has less effect on the areas with higher urbanization level; centrifugal force has higher effect on the areas with higher urbanization level than the ones with lower urbanization level; the effect of market power has gone through an increasing phase and then a decreasing phase as the urbanization level continuously increases; industrial structure has complex and ever-changing effects; city level has stronger effect on the areas with higher urbanization level. Over all, urbanization of the counties in Jiangsu Province is a dynamic process showing different characteristics at different stages.

Key words: driving factors, Jiangsu Province, kernel density, quantile regression, urbanization of population