Abstract: This paper constructs a nonlinear autoregressive distribution lag model (NARDL) and a time-varying parameter factor expansion vector autoregressive model with random volatility (SV-TVP-FAVAR), explores the long-term asymmetric effect of China's macro tax burden on economic growth, and examines the short-term dynamic effect of tax policy on economic growth from aggregate and structure perspectives. The study found that there is a significant long-term equilibrium relationship and asymmetric effect between macro tax burden and economic growth in China, and the positive pulling effect of the negative tax fluctuation on economic growth is weaker than the restraining effect brought by its positive fluctuation; from the perspective of total amount, the increase of tax in each period has a significant negative inhibitory effect on economic growth; from the perspective of structure, the increase of commodity tax has an inhibitory effect 0n economic growth in the period of economic crisis and economic recovery, and has a promoting effect in the period of economic stability Increasing income tax in each period can significantly promote economic growth, but increasing property tax has a negative inhibitory effect on economic growth The double round regulation tax policy of total amount driving and structural optimization is more conducive to promoting high-quality economic growth in China.

Keywords: Tax Policy, Economic Growth; Non-linear Effect