Abstract: Clarifying the connectivity of the Phillips curve is of great significance for early warning of local economic risks, preventing cross-region risk transmission, and promoting regional coordinated development. Based on the provincial data from China from 2005 to 2024, an empirical study is conducted to estimate the regional Phillips curve and identify its connectivity features by using quantile time-frequency connectivity methods. The results show that, firstly, the Phillips curve in China exhibits multiple features, including a flat eastern region, a reverse sloping central western region, and a conventional shape in the Northeast region. Secondly, in terms of the time dimension, the connectivity between curves exhibits a typical time-varying characteristic of being higher during crisis periods and lower during stable periods. After 2014, the eastern, central and western regions have become net risk transmitters. Thirdly, in terms of the frequency dimension, the short-term connectivity of Phillips curve has always dominated, but after the epidemic of COVID-19 infection, the contribution of long-term connectivity has been improved. Fourthly, in terms of the quantile dimension, the tail of the Phillips curve has higher risk, and the central region is a net transmitter of shocks at all quantile levels. In the coming period, it is necessary to focus on the short-term volatility spillover risks under low output and low inflation conditions. Therefore, it is urgent to establish a differentiated regional risk monitoring and early warning system, and continuously improve and innovate the risk management toolbox to ensure the steady recovery of the economy in a relatively safe state.

Key words: regional Phillips Curve, time connectivity, frequency connectivity, quantile connectivity