I estimate the annual local government spending multiplier to range from 2.22 to 0.72 and the accumulated multiplier, which is the total increase of local output per 1 RMB increase of local government spending during 2001-2019, to be 11.8 at the prefecture city level in China, where government spending is known to have a significant impact on the supply side of the local economy. To achieve identification, I construct a novel instrument for local government spending: the fraction of unoccupied raw land in the downtown area in 2000. After 2000, as local governments increasingly rely on land sale revenues to finance expenditures, a higher fraction of unoccupied urban land is associated with less land requisition, which requires compensation to displaced occupants, and hence higher profits from land sales for local governments. Moreover, the fraction of raw land is orthogonal to a rich set of city fundamentals in 2000 thanks to the inefficient utilization of urban land before 2000. The multiplier is found to be significantly larger in cities where the local governments have smaller debt capacity and where the private capital is more mobile. Quantitatively, the crowding-in of private inputs following government spending - such as labor and capital - can explain over 70% of the output increase. At the firm level, government spending is found to improve local productivity and market access, and attract more firm entry. I also find significant and positive spillover effects of local government spending within each city. Overall, the paper highlights the macroeconomic importance of government spending targeting the supply side of the economy.