Pandemic Shock Rewired Housing Price Connectedness Across Chinese Cities
The COVID-19 pandemic did more than disrupt commerce and mobility in China. It altered the way housing markets in major cities transmitted risk to one another, changing both the strength and direction of cross-regional spillovers in second-hand home prices.
Using the second-hand housing price index from 38 major Chinese cities, investigators applied a large time-varying parameter vector autoregression model to measure how price volatility moved through the urban network over time. They then built a time-varying complex network of housing price volatility spillover to track how inter-city connectedness evolved under the pandemic shock. The result was a market structure that did not stay fixed for long. Early in the outbreak, the shock weakened overall connectedness; roughly 40 weeks later, the effect had shifted and became positive.
The market did not simply recover. It reorganized.
Volatility Spillover Fell First, Then Turned Positive
The most striking finding lies in the trajectory of the overall connectedness measure. The pandemic initially exerted a negative influence on the housing market network, reducing the degree to which price volatility spread across cities. In practical terms, shocks in one city were less likely to propagate quickly and forcefully into others during the first phase of the public health crisis.
That pattern fits a market under stress from restrictions on movement, lower transaction activity, and heightened uncertainty. When population mobility falls, inter-city housing demand and capital flows weaken as well. With fewer buyers traveling, fewer relocations, and more hesitation among households and investors, local price signals have less opportunity to transmit across regional markets. The network becomes less tightly coupled, even if underlying uncertainty remains high.
After approximately 40 weeks, the direction changed. The connectedness effect became positive, suggesting that the system had moved beyond simple suppression of activity and into a phase where market linkages began to strengthen again. A housing market under prolonged disturbance can accumulate delayed adjustments: pent-up demand, policy responses, changing expectations, and reallocation of household preferences can all re-engage cross-city pricing relationships. In that setting, the network may not return to its pre-shock shape; it may settle into a new configuration with different channels of contagion and different nodes of influence.
Cross-Regional Spillover Declined Under the Shock
The pandemic also significantly reduced the cross-regional spillover level of housing price volatility. This is a narrower result than the connectedness estimate, but it points in the same direction. Volatility did not travel as efficiently across city boundaries during the pandemic period, which implies a dampening of systemic transmission in the short run.
For policymakers, that kind of reduction can look reassuring at first glance. Lower spillover means less immediate synchronized turbulence across the urban system. Yet that temporary calm can be misleading. A weaker spillover network during an external shock does not eliminate risk; it can simply redistribute it, allowing local imbalances to accumulate before the system re-links and transmits stress again. Housing markets often react slowly to large macro shocks, especially when households delay transactions and developers, lenders, and investors reassess expectations simultaneously.
The analysis of the city network is important because real estate is not a set of isolated local markets. Prices in one city can influence perceptions, capital allocation, and policy expectations in another. When that transmission mechanism weakens or strengthens, the entire system changes shape.
Population Mobility and Behavior Help Explain the Network Shift
Mechanism analysis linked the pandemic’s influence on city housing price volatility mainly to urban population mobility and the psychological and behavioral responses of market participants. Both channels are plausible and mutually reinforcing.
Mobility is the structural channel. Travel restrictions, fewer business trips, reduced commuting, and constrained migration all directly reduce the flow of households across cities. With fewer people moving between urban centers, the demand side of housing becomes less integrated. Price discovery slows, and inter-city spillovers soften.
Behavior is the second channel, and it can be just as powerful. During a public health emergency, households often defer purchases, delay moves, and recalibrate what they want from a home. Investors may become more risk averse or more selective. Sellers may hold back listings if they believe conditions will improve. These shifts are not just emotional reactions; they alter transaction volumes, pricing expectations, and the speed at which one city’s market is able to influence another’s.
A pandemic therefore works through both physical restrictions and market psychology. The network consequence is a housing market that first fragments and then, over time, starts to reconnect under altered conditions.
Why Systemic Risk Monitoring Has to Track Networks, Not Just Prices
The study frames its findings around systemic risk monitoring during major public health events. That is the central policy implication. Housing markets can appear stable if one focuses only on average prices or single-city snapshots, yet the network underneath may be changing rapidly. Connectedness, spillover, and volatility transmission provide a more sensitive view of how shock waves move through the real estate system.
For China’s housing market, this matters because the market is large, interconnected, and exposed to both economic and behavioral disturbances. A public health shock can disrupt transactions, but it can also alter the pattern of regional dependence that links city markets together. Once those links change, the risk profile changes with them.
The paper argues for better supervision and prevention of systemic risks induced by city housing price connectedness during major public health events. That call is grounded in the observed shift from weakened spillover to renewed positive connectedness after several months. Waiting for prices alone to stabilize would miss the broader problem: network structure itself can evolve after the initial shock has passed.
Aggregate Data Offer a Broad View, Not a Household Portrait
The study relied on pre-existing, non-identifiable aggregate data at the city level. It used housing-market indices, population-mobility indices, search-attention indices, and publicly reported COVID-19 statistics. No individual participants were recruited, interviewed, or directly observed, which means the work is best read as a systems-level analysis rather than a patient- or household-level study.
That matters because the conclusions apply to market connectivity, not to individual buyer behavior in isolation. The model captures how cities move together, where volatility originates, and when transmission weakens or strengthens. It does not identify the specific households, developers, or lenders behind each move. The strength of the design lies in that breadth: a crisis of this scale demands a network lens.
The remaining question is not whether housing markets were affected by the pandemic. They were. The more actionable issue is how long the disruption suppressed spillovers, when connectedness returned, and which channels were responsible for the turn. This analysis places a marker at roughly 40 weeks, but the exact sequence of policy responses, mobility changes, and expectation shifts will need further examination in future work.