Housing Market Responses to a School Allocation Reform: Evidence from a Policy Experiment in Beijing

This project aims to exploit a school allocation reform that generated a switch from “single-school zoning” to “multi-school zoning” in a central district in Beijing to examine the housing market responses. Two unique features distinguish this school allocation reform from other settings examined in the existing literature. First, under the new school allocation policy whether a residence is subject to single-school zoning or multi- school zoning depends on the time of the property transaction. Second, the reform was announced three months before the new policy is enforced. These distinct features of this school allocation reform lead to both anticipated and unanticipated moments of discontinuity: while the announcement of the switch from single-school zoning to multi-school zoning was unanticipated, the date at which the single-school zoning would come to an end was fixed and fully anticipated following the announcement. Therefore, the reform creates a time window with an unanticipated start date but a fully anticipated end date for both potential buyers to lock in their preferred school by purchasing a residence within its catchment area under single-school zoning and potential sellers possessing such units to redeem the premiums associated with their properties before the advantage expires.

This project plans to collect the second-hand house transactions in the six central districts of Beijing over three calendar years from January 1, 2019 to December 31, 2021. Web scraping tools will be used to download all the second-hand house transactions listed by the two largest real estate brokers in Beijing, who jointly account for around 60% of the market share. To guide the empirical analysis, we will develop a dynamic model of housing market responses to a school allocation reform announced at time 0 that the single-school zoning policy will be replaced by a multi-school zoning policy for new home buyers who purchase a house after time t. The model will derive predictions on the transaction volume and price dynamics of different types of houses, defined by the quality of schools they are assigned to under single-school zoning, for both the announcement period (i.e., between time 0 and t) and the enforcement period (i.e., after time t). Event-study approaches will then be implemented to estimate the housing market effects of the unanticipated announcement of this school allocation reform to test the predictions of the model.