In this paper, regional real estate evaluation indexes, real estate input and output efficiencies are sorted out. Under the condition of constant returns to scale, a three-stage DEA method is used to equalize the relative efficiencies of decision-making units in multiple-input and multiple-output problems, and the three-stage DEA method is combined with the Malmquist index method to form a Malmquist-DEA dynamic evaluation model, which in turn reflects the changes in total factor productivity over successive periods. The spatial autocorrelation method and Markov chain are used to explore the patterns and trends in the distribution of geospatial data variables. The results show that the overall total factor productivity index and technical efficiency increased by 2.77% and 3.05% on average, while the technical progress index decreased by 0.27%. The increase in the technical efficiency index is the main reason for the increase in total factor productivity in China’s real estate industry. The total factor productivity index as a whole shows that eastern > central > western regions, similar to the economic development situation. The total factor productivity of different provinces shows strong regional differences, and real estate development is affected by neighboring cities in the short term, making it difficult to achieve transfer across stage levels, and the degree of risk of transferring to a low level state is low.
