Chinese liquidity effects on the Australian macroeconomy, 2002–2017

ABSTRACT

China’s growing importance to the Australian economy has been well recognized in policy circles but remained relatively untested in formal empirical analysis. This paper examines the reactions of Australian macroeconomic variables to Chinese money growth and inflation over the post-2002 period using VAR estimation, historical decompositions and long-run cointegration models. The consistent impact of Chinese money growth on Australian inflation and on the exchange rate seen in the VAR analysis is supplemented by evidence of cointegrating relationships between the Australian variables and both Chinese money growth and Chinese inflation.

KEYWORDS:

• International transmission
• Australia
• China
• liquidity
• VAR/VECM

JEL CLASSIFICATION:

• F42
• F36

Acknowledgments

The authors thank Mardi Dungey and Farrokh Langdana for helpful comments and are grateful to Gu Jinlin, Hu Xinyi and Zhang Wenhao for valuable research assistance.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ See also Burdekin and Siklos (2012) for evidence of widespread cointegration between the Shanghai market and other Asian markets as well as Australia and the United States.

² See also Levieuge (2017) on the relative performance of VAR models relative to VECMs in a case where the cointegrating relationships may have been subject to shocks.

³ Our focus upon money supply rather than interest rates is primarily driven by the People’s Bank of China’s emphasis on money supply rather than interest rate targeting as well as the complications arising from China’s ongoing transition from administratively determined rates over our sample period.

⁴ Although the money supply is not used as a policy variable by the Australian Reserve Bank, and M2 data are no longer collected, Australian Treasury Secretary John Fraser recently acknowledged the ongoing relevance of the M3 money supply data: ‘[O]ver the longer term, Friedman made a case that there was a link between M3 and prices … [so] fiddle with money supply at your peril’ (as quoted in Roberts and Nelson 2017).

⁵ Ironically, rising Chinese resource demand may have helped contribute to this manufacturing decline via a ‘Dutch disease’ effect insofar as Chinese demand shocks elicit the movement of factors of production from Australia’s non-resource sector to the resource sector (Dungey, Fry-McKibbin, and Linehan 2014). Meanwhile, a secondary service sector boom has further reduced the relative importance of the Australian manufacturing sector (Tyers and Walker 2016).

⁶ Alternative interpolation methods also considered include Denton (1971) and Litterman (1983). In addition, we tried using monthly industrial production as an indicator variable contributing to the monthly movement pattern of the interpolated GDP series. The resulting series are almost identical. Due to the fact that the Australian industrial production series was discontinued in March 2016, we chose the Chow-Lin method without an indicator variable.

⁷ Only the indicated initial effect on Australian reserves is significant with Chinese CPI – and this comes with a counterintuitive negative sign.

⁸ As a robustness check, we also constructed IRFs identified by recursive methods based on standard unrestricted VARs. In order to measure the contemporaneous effects of policy variables on economic variables, the policy variables should be ordered last. Therefore, the ordering of the variables for the IRFs is as follows: Australian GDP, Australian reserves, the Australian exchange rate, Australian CPI, Australian M3, Chinese M2 (CPI). The IRFs are reported in Appendix Figures A2 and A3.

⁹ The main stimulus package in November 2008 was accompanied by a sequence of measures targeting specific industries in early 2009 as well as rapid expansion in bank lending. Soaring liquidity helped fuel a housing price run up along with stock market gains that were concentrated primarily in the property, construction and building materials sectors (cf, Burdekin and Weidenmier 2015).

¹⁰ Rodriguez and Rowe (2007) apply similar modifications where $D_t$ represents the possible deterministic components of the process.

¹¹ Only significant coefficients are included in the equations.