Oceans Apart? China and Other Systemically Important Economies

ABSTRACT

China has been considered a systemically important economy for at least a decade. As policymakers worldwide grapple with sluggish growth there is relatively little evidence about whether the G4, which consists of the US, the Eurozone, Japan, and includes China, as a block contributes to global economic performance in a manner that is not possible when China is left out or treated exogenously. We estimate a series of panel factor and standard VARs because these are well suited to exploit cross-country links. We estimate the relative impact of domestic and global factors on these four economies. First, it is essential to treat China in a model of the G4, on a level playing field with the US, the Eurozone, and Japan to better understand how shocks among these economies interact with each other. Second, we find that domestic and global shocks can reinforce each other. Indeed, global monetary shocks explain up to 60% of variation in commodity demand and real economic conditions. We also report that there is a trade-off between domestic monetary and financial conditions. We recommend that policymakers to reexamine the potential benefits from greater policy cooperation.

KEYWORDS:

• International spillovers
• time-varying factor models
• Panel VARs
• policy coordination
• G3
• China

JEL:

• E32
• C32
• E58

Disclosure statement

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

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/1540496X.2022.2119804.

Notes

1. After research for this paper was completed a third global crisis is the COVID-19 pandemic but insufficient data prevents meaningful inclusion of its impact on our conclusions.

2. While the estimation technique used in our study has many advantages it can quickly exhaust degrees of freedom. Consequently, we exclude a ‘rest of the world’ group of economies that is, in any case, highly heterogeneous and represents a comparatively small portion of the world economy.

3. Cooperation and coordination are frequently used interchangeably. Clarida (2018, 16), however, provides a definition that can be seen as accommodating both goals, namely “ … the sharing of information regarding estimates of the unobservable inputs to policy rules such as the equilibrium real rate of interest and potential output as the considerations that would govern the timing and trajectory of a baseline policy path as well as trigger deviations from such a path..”

4. The trilemma posits that a fixed exchange rate, full capital mobility, and monetary policy independence are incompatible.

5. We omit discussion of the oft-mentioned question of the reliability of Chinese data. See, for example, Pang and Siklos (2016), and references therein. Fernald, Hsu, and Spiegel (2019) is a recent contribution which concludes that Chinese data have become more reliable over time although the record for GDP is mixed.

6. Spillbacks generally refer to shocks from the rest of the world that feed back into the advanced economies that generated the spillovers in the first place. However, they are extremely difficult to identify or require fairly restrictive models to estimate them (e.g., see Breitenlechner, Georgiadis, and Schumann 2021).

7. These are found, respectively, at https://www.nber.org/cycles.html, https://cepr.org/content/euro-area-business-cycle-dating-committee, https://www.jcer.or.jp/english/november-the-recession-indicator-rose-to-6-8.

8. Cycles for the financial sector are longer than for other conventional business cycle variables (e.g., real GDP; Borio 2012). Burnside, Eichenbaum, and Rebelo (2016), propose a moving average filter to explain deviations away from some fundamental value. We also estimated several time series, most notably for housing prices, using this filter but this did not impact our conclusions.

9. Since many of the traditional measures of recessions and expansion are categorical (i.e., a 1 indicates a recession, a 0 an expansion) this requires the use of a polychoric correlation measure to estimate the principal component. While we focus on results using PCA we do report some stylized facts using the wiring ratio technique (results not shown; also see Jordá, Schularik, and Taylor 2011).

10. A criticism is that factor models often rely on a larger number of variables that are being used. However, the total number of variables used in our study does not differ much from Stock and Watson (2018), or Hatzius et al. (2010). More importantly, even when more than 100 series are used it is typically that case that only a small number are necessary to explain most of the variation in the data.

11. This is evaluated by the uniqueness statistical measure in principal component analysis. Typically, significance is determined by retaining the eigenvalues that exceed one (also known as the Kaiser-Guttmann procedure). Moreover, as noted earlier, the choice is also motivated by the factor loadings that must be economically meaningful.

12. A consequence of the GFC is the limited usefulness of an observed policy rate as an indicator of the stance of monetary policy. Following the practice adopted by many in the literature, published shadow policy rates are used for the US, the Eurozone, and Japan, once the zero lower bound is reached. See the notes to Table 1.

13. Factor scores represent a composite measure that creates observations extracted from each factor. They are standardized and factors weights are used together with the raw data to create a score.

14. Granger-causality tests (not shown) confirm the interpretations from the variance decompositions.