政府消费的生产性——基于生产网络模型的刻画、分解和检验

The Productiveness of Government Consumption in Production Networks Model: Theoretical and Empirical Explorations

在经济形势错综复杂且减税降费规模扩大背景下,政府迫切需要优化财政支出结构,而其基本前提是对各类财政支出如何影响经济做出研判。本文就政府消费是否通过生产率渠道作用于经济以及如何作用进行理论分析和实证检验。通过将政府消费引入带有生产网络的多部门一般均衡框架,本文实现了政府消费生产性效应和需求侧效应的刻画和分解,明晰了政府消费冲击借助生产率渠道在生产网络中放大和扩散的传导机制。基于以上理论分解,本文利用工具变量方法进行实证检验,发现政府消费具有显著的生产性效应,该效应作用于行业生产率进而影响产出的过程具有4年时滞。这些研究发现对财政政策制定,特别是财政支出的结构优化具有借鉴意义。

As a market-based redistribution activity of the government, fiscal expenditure including government consumption and public investment has a great direct impact on social production and employment and a strong ability to allocate resources. Therefore, it is usually regarded as an important means of regulation and control by the government during the recession. However, in the traditional sense, government expenditure such as infrastructure is called public investment, while expenditure on science, education, culture, health and public administration is called government consumption. In order to clarify the understanding of how government expenditures affect the economy and provide a theoretical basis for how financial departments arrange expenditures, it is necessary to carry out theoretical analysis and empirical test on the productivity mechanism of government consumption affecting output, which is also the interest of this paper.Referring to recent research on production network, such as Acemoglu et al.(2016), Liu(2019) and Baqaee and Farhi(2019), this paper constructs a multi-sector general equilibrium model with production network and clearly describes the mechanism of government consumption shock. The influence of government consumption on output is clearly decomposed into the productive effect formed by affecting the productivity and the demand-side effect formed by affecting the purchase of final products. Based on the theoretical model, this paper uses the samples of 47 industries in China from 2000 to 2014 to conduct an empirical analysis on whether government consumption influences output by acting on firm productivity, that is, the existence of government consumption productivity channel. Firstly, the transmission mechanism and final impact of productivity shock described in the model are used to estimate the elasticity of substitution of important parameters in the model. Secondly, the difference between the final account and budget of total government consumption expenditure is used to construct policy impact. The regression in this empirical model is carried out at the industry level, so the total government consumption shock can be reasonably considered to be orthogonal to the productivity shock and other exogenous shocks at the industry level. Thirdly, the productive effect of government consumption shock is constructed based on the decomposition results of the model. Since the productivity effect is the core independent variable of regression, the network effect reflecting the spillover between industries must be exogenous while ensuring the exogenous impact itself. In this paper, method in Nakamura and Steinsson(2014) is used to fit network effect changes with total government consumption shocks, and the part explained by total government consumption shocks is separated out as instrumental variables. Finally, the possible time lag of government consumption shock is noted.Compared with the existing literature, this paper may have the following two aspects of marginal contributions:one is in the theoretical modeling and analysis of the impact of fiscal expenditure. What this paper has done at the margin is to introduce this setting into a multi-sector general equilibrium model. The advantage is that it can find the analytical expression for the static model comparison and decompose the impact of fiscal expenditure from the demand side and the production side. This decomposition also lays the foundation for the empirical analysis, which makes it possible to test by reduced-form regression. The second refers to the empirical analysis of the impact of fiscal expenditure. This paper attempts to test whether government consumption should enter the production function,that is, have an impact by means of productive effect. The advantage of the test in this paper is that it is based on a structure model to reduce the possible errors in the estimation.