When the economy is booming, investment demand is strong, housing price fluctuations increase, and housing price fluctuations attract attention. Most studies focus on the impact of external factors on housing price fluctuations. There are few studies on how the time factor of house age affects housing price fluctuations, which creates research motivation. This study aims to explore the relationship between short-term error correction and long-term equilibrium of house prices in different stages. This study will use methods such as VAR vector autoregression model, Johansen co-integration test, Granger causality test, error correction model (vector error correction model (VECM) for short) impact response analysis and variation decomposition. We found that there is a complementary relationship between pre-sale houses, new houses and old houses in the long-term equilibrium relationship between house prices, and that there is a substitution relationship between new and second-hand houses. The short-term error correction lead and lag is a one-way cyclic relationship of pre-sold houses → old houses → new mid-century houses → new houses → pre-sold houses, while mid-century houses lead pre-sold houses and new houses. There is a significant difference in the impact response at different stages. The explanation of the variation of ancient houses and old houses does not reach 50%, while the explanations of the other stages exceed 60%.