Abstract
This paper classifies government expenditures into three types and constructs a theoretical framework to explore the influences of government expenditures and corruption on total factor productivity. It then establishes a dynamic spatial autoregressive model and a panel threshold model to test the theoretical hypotheses based on provincial panel data from 2007 to 2014. The results illustrate that “U” shape curve relationships exist between government expenditures of administrative service, investment development, safeguard governance and total factor productivity, and that the relationship between the government expenditure structure and total factor productivity follows an “inverted U” shape curve. In addition, the findings suggest that increased corruption levels can directly reduce regional total factor productivity and that the effects of the proportions of administrative service expenditures, investment development expenditures, and safeguard governance expenditures on total factor productivity have a single corruption threshold. Increasing the proportions of government expenditures could improve the total factor productivity up until the threshold value.