Based on the panel data of 35 large and medium-sized cities in China from 2008 to 2016, Tobit model and input-oriented data envelopment analysis (DEA) method was used to explore the impact of urban housing sales price on urban total factor productivity. It is found that in the overall dimension, the urban housing price and urban productivity present an inverted U-shaped curve, which increases first and then decreases. The urban housing price level exceeds the maximum critical value of "Houses are for living in, not for speculation", and the inhibiting effect of housing price rise on urban total factor productivity is significant. The effect of housing price on the productivity of cities with different population sizes is heterogeneous. When the city population size reaches the standard of the very large-sized city, the housing price and urban productivity show an inverted U-shaped feature, and when the city population size reaches the standard of the super large-sized city, the inverted U-shaped feature of housing price and urban productivity disappears. The inverted U-shaped relationship between housing prices and urban productivity in different types of cities and the heterogeneity of cities indicate that China's large and medium-sized cities should carry out the fine adjustment of housing prices, and the policy positioning of "Houses are for living in, not for speculation" has a solid theoretical basis and practical necessity.