Does China’s monetary policy framework incorporate financial stability?

Abstract

This article investigates the response of monetary policy to financial instability in China. We estimate a forward-looking Taylor rule model with a constructed comprehensive financial stress index using the time-varying coefficient method. Empirical results suggest that financial stability has always been a main concern for China’s monetary authorities even in periods with low financial pressure. Moreover, China’s central bank tends to lower the policy interest rate in response to financial instability, but the size of policy responses varies substantially over time. Although the proportion of policy interest rate change due to financial stability concern is less relative to developed countries, financial stability is increasing in importance for monetary policymaking in China. We also find that banking stress and stock-market stress are two main concerns for China’s central bank, while little evidence supports that exchange-market stress can drive the reaction of China’s central bank.

Keywords:

• Forward-looking Taylor rule
• financial stability
• monetary policy
• time-varying coefficient model

JEL codes:

• E44 E52 E58

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Time-varying coefficient (TVC) model is first proposed by Schlicht and Ludsteck (2006).

2 In October 2012, Xiaochuan Zhou, the president of China’s central bank, gave a talk in the Lecture hosted by 2012 annual Per Jacobson foundation during the conference of International Monetary Fund and World Bank in Japan, and set forth that Chinese government always takes the financial stability seriously and views it as the premise of stable price.

3 Source: China’s Monetary Policy Execution Report from the last quarters of years 2012 and 2013.

4 To compare with the literatures, we also do the estimations for k = 1 and 3, and the results are quite similar.

5 State-space models have some limitations for the empirical work. For example, it is hard to attain accurate estimated parameters since the results are sensitive to their initial values. Moreover, the log likelihood function is highly non-linear, leading to failure in optimization of the log likelihood (Baxa et al., 2013).

6 See http://www.pbc.gov.cn.

7 We use the package for data frequency transformation in EVIEWS to do the frequency transformation and also exclude the seasonal factor in GDP by X-12 seasonal adjustment.

8 Real GDP = (nominal monthly GDP/CPI in the current month) × 100, where monthly CPI is with year 2000 as the base year.

9 We adopt Hodrick–Prescott filter to estimate the potential GDP and the output gap.

10 We exclude sovereign risk since China holds the largest amount of foreign reserve in the world, its composition of foreign debts is reasonable and it maintains a low level of sovereign risk during the sample periods.

11 Treasury bond market in China is relatively small and there is no three-month bond traded. Besides, the issue of Treasury bond is not regular. Hence, we employ the three-month time deposit rates as the proxy for yield to maturity of three-month Treasury bond.

12 Subjective weighting method heavily relies on experts' experience. Instead objective weighting method is widely recognized by academia. For the detailed description about CRITIC method, readers can refer to Diakoulaki et al. (1995). We also take the variance-equal weighting approach to construct FSI index and the resulting key estimations remain unchanged.

13 Several classic unit root tests include ADF test, Phillips-Perron test and KPSS test. The results are similar, except that the null for the KPSS test is rejected at the 5% significance level for the policy interest rate. This is due to a break point in the policy rate series in Oct. 1997. We employ a breakpoint DF unit root test to take account of the break and the result show that the unit root in the policy rate can be rejected at the 1% significance level.

14 The high inflation during this period is largely caused by China’s 4 trillion RMB Stimulus Plan to counteract the negative effect of global financial crisis.

15 Here, a monetary policy rule is stable because it could maintain the actual GDP around the potential GDP.

16 As argued in Baxa et al. (2013), the positive effect of financial instability on the policy interest rate might result from scaling the financial instability index. Thus, we do not care about the positive impact of FSI unless it is caused by the positive and significant regression coefficient of FSI, which is not the case based on Figure 6.

17 The RMB exchange rate regime switched to a managed float with reference to a basket of currencies on July. 2015.

Additional information

Funding

We acknowledge the financial support by the China Postdoctoral Science Foundation (grant no. 2016M600132), Natural Science Foundation of Zhejiang province, China (grant no. LQ17G030008), and National Social Science Fund of China (grant no.18BGL224). All responsibility for errors and omissions lies, however, with the authors.