Abstract: On the basis of considering spatial heterogeneity and narrowing the scale of research areas, and taking the 21 cities of Guangdong province as research unit, the paper utilizes the method of ESDA to prove the existence of significant spatial correlation of income level in Guangdong province from 1980 to 2015: from the perspective of local autocorrelation, there is still a large gap between cities of Guangdong province, the high-high income cluster is far away from the low-low income cluster, and the situation has changed little over the years; high income group has a certain effect on low-income group, but has no obvious influence on low-income cluster. Through the geographical weighted regression, the paper concludes that globalization, marketization location and decentralization are the main reasons for income inequality in Guangdong. In a single variable GWR, marketization and decentralization are inverted U-shaped curve rolling in the income level, and urbanization intensifies regional income gap; in the multivariate GWR,the inverted U-shaped trend of marketization and decentralization is weaken, while urbanization shows U-shaped curve and its effect on regional income levels is strengthened.

Key words: Guangdong, income distribution, ESDA, GWR, influence factor