ABSTRACT
This study explores the non-linear effects of economic policy uncertainty, bilateral trade intensity, and capital flow on China’s financial cycle spillover when institutional distance changes over the period 1997Q1-2017Q4. Main findings indicate that there is a linear effect of these influential factors on China’s financial cycle spillover during the overall sample period and a non-linear effect during the normal and crisis periods. The transition function exhibits a smooth and gradual change trend during the normal period and a double-threshold effect during the crisis one. Furthermore, these influential factors present differences with regard to facilitating and restraining effect in different periods. These results have important implications for policymakers to make macroprudential policies.
Conflicts of Interest
The authors declare no conflict of interest.
Supplementary Material
Supplemental data for this article can be accessed http://afr.gzhu.edu.cn/.
Notes
1. For more details please refer to Diebold and Yilmaz (Citation2012).
2. For robustness purposes, we have also experimented using alternative 28- and 36-quarter rolling window length and our conclusions reached have not been affected.
3. When we calculate the rolling window regression of the spillover index, we use the data of 1992Q1-1996Q4 as the rolling window, that is, using the 20-quarter (5 years) as the window length. So the spillover index lost the data of 1992Q1-1996Q4.
4. About the sample countries, we have done a lot of work on data collection and model fitting. When choosing developed country as the sample, our initial consideration was the G7 (United States, United Kingdom, Germany, France, Japan, Italy, and Canada). But the financial cycles of Italy and Canada are characterized by strong collinearity with other countries when fitting the vector autoregression (VAR) shown below. As a result, we have generated a singular matrix in the financial cycle when calculating VAR, and the financial cycle matrix cannot converge. With the applicability and availability of the data, we finally determined the developed countries samples as the United States, United Kingdom, Germany, France, and Japan.
5. This performs a proprietary local quadratic interpolation of the low-frequency data to fill in the high observations.