In recent years, the frequent occurrence of non-standard defaults of local government financing vehicle (LGFV) has aggravated the risk of local government implicit debt.

The key from statistical inferring cause and effect inference is to effectively solve endogenous problems. Limited by data acquisition, a large number of research problems in the field of management still need to be carried out in non -experimental data, so endogenous problems are particularly obvious.

On the basis of combing the traditional policy effect evaluation methods such as fixed effect model, instrumental variable, difference in differences, regression discontinuity design and synthetic control method, this paper investigates policy evaluation methods under multidimensional spillover effects and emphasizes the unique advantages of spatial network models in addressing the challenges of multidimensional spillover effects.