Abstract: China has undergone an institutional transition from welfare-oriented to market-oriented housing allocation system since the year of 1978. Housing reform was closely related to social change and economic reform. Thus, it has significant influences on socioeconomic conditions. Housing resource is an important indicator of social stratification. In this paper, Xiamen City is selected as a research area. The city incorporates a variety of residential patterns in different periods including transitional communities with local characters, post-reform communities, commercial communities and city villages. The housing resource data was obtained by spatial sampling and survey. We conducted a field study with 1075 face to face questionnaires. More than ten years after the housing allocation system was stopped, housing ownership was characterized with diversity in terms of income, education, family size and occupation. The empirical results showed that people have various sources to obtain houses. Commercial house in real estate was the primary housing source, accounting for 56% of the residents with house ownership. Public house was ranked as the second source, accounting for 33.4%, which was still an important source after the housing reform due to path dependence. The logistic regression results showed family income, education and family size, multiplied, increased the probability of housing ownership which has positive feedback to self-identification of social status. Income and occupation have positive feedback to housing resource. For non-public economy unit job holders and high income group, they have more opportunities to access the public houses at a lower price, and gained more profits in the real estate market. This paper explored the driving mechanism of the housing stratification by using time series and cross-section data. The housing reform process was divided into four stages. The cluster analysis indicated the housing expenditure gap among low income, median income and high income groups, increased during the housing reform process. But the time series data didn't verify the cause-and-effect relationship. Structural equation model using cross-section data was further used to verify the relationship between housing ownership and its influencing factors. Family income has strong effect on the probability of housing ownership, followed by education, while occupation and family size have least effect. Income and housing ownership in turn have direct effect on self-identification. The interaction between income and occupation indicated that housing resource stratification resulted from economic and institutional factors. After housing reform, the house distribution system should consider the housing gap induced by economic status and occupational prestige and expand the coverage of welfare housing. Current housing resource was still influenced by the previous distribution system. And residents have flexible choices on residential location and housing resources, which would accelerate the complexity of urban changes. This research quantified the current situation of housing resource stratification, and then further analyzed its mechanism and effects. The research in this paper has enriched the theory of social stratification, and provided scientific evidence for the sustainable housing reform.

Key words: housing ownership, housing reform, social stratification, structural equation model, Xiamen City