The Effect of Government Expenditure on Private Consumption: Evidence from China

The “investment-driven, export-led” model has long been adopted by Chinese government to promote economic growth; however, investment is a derived demand for consumption, and if there is no consumption support, investment then turns out to be ineffective. In addition, net exports also cannot sustain the demand in support of the long-term economic growth because it is influenced easily by the economic condition of international trading partners and international trade frictions [1]. On the contrary, private consumption is arguably a more stable component of economic growth and that is why it is relevant to this study.


Introduction
The "investment-driven, export-led" model has long been adopted by Chinese government to promote economic growth; however, investment is a derived demand for consumption, and if there is no consumption support, investment then turns out to be ineffective. In addition, net exports also cannot sustain the demand in support of the long-term economic growth because it is influenced easily by the economic condition of international trading partners and international trade frictions [1]. On the contrary, private consumption is arguably a more stable component of economic growth and that is why it is relevant to this study.
Since the global economic crisis of 2008-2011, the low level of private consumption in China has been seen as a factor holding back economic development. Government spending, as a method of income redistribution, has a significant effect on private consumption. Thus the main concern in this paper is whether and to what extent the government spending stimulates private consumption. We are particularly interested in studying the effect of aggregate government fiscal expenditure on private consumption in China.
We collect data from 29 provinces in China from the China Statistical Yearbook over the period 1996-2013. The issue of nonstationary is addressed by applying a panel data version of the dynamic OLS estimator developed by Stock and Watson [2]. It is shown that an increase in the aggregate government spending is expected to crowd in private consumption. And when the model is further discomposed to incorporate the three different components of government spending, such as government spending on economic construction, culture and education as well as administrative management, we find that the spending on culture and education is complementary to private consumption, whereas, both spending on economic construction and administrative management have a negative impact on private consumption. Finally, some policy suggestions are provided.

Theoretical Framework and Empirical Strategy
Following the standard Keynesian model by maximizing the household's lifetime utility subject to the budget constraint and rearranging the Euler equation, we start the analysis based on the estimable econometric equation To broaden heterogeneity among individual members and control for omitted variable bias, panel data approach is adopted to identify the parameters of concern with additional information from cross-sectional units [4,5]. Besides, according to Graham [6], the estimation of the association between government spending and private consumption could be biased if the disposable income is excluded from the model [7]. Thus, we estimate the modified model as below, Where α i0 is the intercept for each unit; α 1a is the coefficient for each independent variable; C it is the private consumption; Y d denotes the disposable income; G it represents government spending; ν it is the error term. We equally apply the panel data model above to the case including the different components of government expenditure given by, Where G 1it is government expenditure on economic construction; G 2it denotes government spending on culture and education and G 3it represents government spending on administrative management. In terms of empirical strategy, we conduct the panel unit root tests, panel co-integration tests to deal with the non-stationary time series and then successfully identify the panel dynamic OLS estimator. The main results are summarized shows in the Tables.

Concluding Remarks
The results confirm the evidence of the crowding-in effect between aggregate level of government spending and private consumption (α 1a = 0.26 in Table 1). Furthermore, we have also assessed the effects of different types of government spending on private consumption. The significant coefficients (α 1 =-0.121, α 2 =0.744 and α 3 =-0.409 in Tables 2 and 3) of the three different types of government spending imply that government spending on culture and education and private consumption are complements. Alternatively, there is a significant degree of substitutability between government expenditure on economic construction and private consumption, as well as between government expenditure on administrative management and private consumption respectively.
From a policy standpoint, the reported estimates seem to suggest that government can improve economic performance in the long run by raising the investment in education, culture and health, and in the meantime control the administrative expenditure in a more reasonable way. In addition, it will be interesting to group the 29 provinces in China into different regions and compare the varied impacts of government expenditure by region controlling the size of the government and the degree of economic growth.