Forecasting macroeconomy using Granger-causality network connectedness

ABSTRACT

The connectedness among financial institutions reflects potential channels for risk contagion and the amplification of risk to the financial system that can also propagate into the real economy. This study investigates the predictive power of financial network connectedness for macroeconomy. We highlight the connectedness by quantifying the effects of risk transmission among financial institutions in Granger-causality networks. The aggregate macroeconomy is viewed as a proxy for economic activity and is extracted from several monthly single macroeconomic variables by principal component analysis. We use the n-month-ahead multivariate predictive regressions to explore the predictive power of the connectedness and test whether the predictive ability is robust. The results show that after controlling for a number of factors, an increase in network connectedness among Chinese financial institutions strongly and stably predicts higher Chinese economic activity about four to twelve (except for five) months into the future.

KEYWORDS:

• Financial institutions
• network connectedness
• predictive regression
• macroeconomy

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ With the improved market structures, legal systems and institutional arrangements, Chinese stock market has had the most flexible market mechanism in Chinese economy and has achieved some form of market efficiency (Luo, Ren, and Wang 2015; Liu, Xia, and Xiao 2020).

² The larger value means the better the performance of all aspects of the macroeconomy.

³ Considering the factors such as the frequency of macroeconomic indicators and the availability of data.

⁴ Wind Info is the market leader in China’s financial data services industry.

⁵ We also set the criteria for Granger-causality connectedness measure as 5%. However, note that the network connectedness is weak because of few edges and the elements. This sparse network cannot capture the mutually beneficial business relationships between Chinese financial institutions. Therefore, our article adopts the criteria 10% in the full text.

⁶ According to the following formula in Newey and West (1987): number of lags $q=floor(4\times (\frac{T-n}{100}{)}^{(\frac{2}{9})})$, where n equals 12, corresponding to the number of month lags (denoted n in Equation(6)), and T is 131, corresponding to the 131 months between January 2008 to January 2019, and floor represents the floor function.

Additional information

Funding

This work was supported by the National Natural Science Foundation of China (No.71771042), the Fundamental Research Funds for the Central Universities (N180614004) and the Project of Humanities and Social Science of Ministry of Education of China (Grant No.18YJCZH224).