我国财政政策的私人投资需求非线性效应研究

An Empirical Investigation on the Nonlinear Effect of Fiscal Policy on Private Investment in China

本文采用MS-VAR模型实证检验了我国财政政策对私人投资需求的影响是否存在非线性特征,并采用区制依赖型脉冲响应方法(Regime-Dependent IRF)分析了财政政策对私人投资影响的方向和持续时间。实证结果显示,财政政策存在非线性的私人投资效应;在投资需求不足时期,扩大财政支出显著促进私人投资增长,持续时间约20个月;在投资需求旺盛时期,增加财政支出显著抑制了私人投资增长,持续时间约10个月。

Before 1998, the existing studies assumed that the effect of fiscal policy on investment demand was linear. Alesina et al. （1998） reveal that there might be a nonlinear effect of fiscal policy on private investment for the first time. This might be due to the effects of fiscal policy through interest rate and labor market during the expansion and recession periods. The researchers have focused on the nonlinear effects of fiscal policy on the private investment since 1998. For last decades, China has implemented fiscal policy to stimulate private investment. But the effects are not as good as expected. This might be explained by the nonlinear effects. Therefore, it is necessary to conduct an empirical investigation on the nonlinear effects based on the data from China. We utilize MS-VAR and Regime-Dependent IRF to test whether there is a nonlinear effect of fiscal policy on private investment and the extent of the effects. This empirical test includes following steps： Step 1： Variable selection. Assume thatyt = （△piz ,△et ,△rt）, where △pit is growth rate of private investment, a proxy for demand of private investment, △rt is growth rate of government tax, a proxy for government revenue, △et is growth rate of government expenditure, a proxy for government expenditure. Step 2 ： LR test. We use LR Linear Test derived by Lo and Zivot （2001） to test the covariances between linear VAR and MS-VAR： where ∑^2 is the covariance of MS-VAR and ∑^1, is the covariance of linear VAR. If the null hypothesis is rejected, it means that there is nonlinear relationship between the variables. The MS-VAR is used in this study. Step 3 ： Determination of lag order of VAR based on AIC and HQ. We employ AIC and HQ to derive the order of VAR. Step 4 ： Estimation of MS-VAR. We test whether there exists a nonlinear effect of fiscal policy on private investment. Thus, we compare the coefficients and residuals under different regimes. If there are significant differences between the regimes, it means that there is a nonlinear effect. Step 5 Dynamic relationships among variables by utilizing the Regime-Dependent IRF. We investigate the direction, extent and duration of the nonlinear effects of fiscal policy on the private investment. The data employed for this study includes monthly data from January 1990 to November 2013 and is from Zhongjing Statistics Database. We find that ： （i） LR test shows that there is a nonlinear effect of fiscal policy on private investment as LR = 212. 344 ,the null hypothesis is rejected at 1% significance level. The comparisons between the regression residual graphs also provide the similar results. （ii） Government expenditure has a positive impact on private investment under low investment demand regime, lasting for 20 months. （iii） On the contract, government expenditure has a negative impact on private investment under high investment demand regime, lasting for 10 months. The implication for this study is that it is important to identify the periods of investment demand. Government should use the expenditure as a policy tool for private investment during the period of lack of demand. On the contrast, government should not use expenditure during the period of excessive demand.