The size and destination of China’s portfolio outflows

ABSTRACT

China’s large financial system is relatively closed, raising concerns that liberalization of China’s capital account could have disruptive effects on the global financial system. We estimate a portfolio allocation model to estimate an economy’s foreign portfolio investment, using panel data for 39 economies. We estimate the model separately for equity and debt securities holdings. We find that portfolio allocation to a foreign economy depends on the destination economy’s market size, gravity variables, governance indicators and capital controls in source and destination economies. We then construct a counterfactual scenario of China’s overseas portfolio investment allocations in 2015 if China had liberalized capital outflows. The analysis indicates that China’s holdings of overseas portfolio assets would have been large at 13% to 29% of Chinese GDP, or 5 to 12 times its actual levels. These asset holdings would have been predominantly from the world’s deepest financial markets: the United States, euro area and Japan. Emerging-market economies would have received relatively little additional portfolio inflows from China, suggesting that liberalization of China’s portfolio outflows may not prove disruptive to the global financial system.

KEYWORDS:

• International capital flows
• portfolio allocation
• china
• balance of payments

JEL CLASSIFICATION:

• G15
• F32
• F02

Acknowledgments

We thank Mark Kruger, Gurnain Pasricha and Walter Steingress for their constructive comments and suggestions. We would also like to thank Alexander Lam for his excellent research assistance with this project.

Data availability statement

The authors confirm that the data supporting the findings of this study are available within the article and its supplementary materials.

Disclosure statement

The views expressed in this paper are our own and do not necessarily represent those of the Bank of Canada or the Reserve Bank of Australia.

Notes

¹ We use the definition of debt securities from the IMF CPIS database which includes both long-term debt securities and short-term debt securities. Long-term debt securities cover instruments such as bonds, debentures, and notes. Short-term debt securities cover treasury bills, negotiable certificates of deposit, commercial paper, and bankers’ acceptances.

² The 10-year US treasury yield is included only in the equity regressions.

³ All six World Bank governance indicators are also individually highly correlated with capital controls.

⁴ Random effects estimation results are shown in Appendix B and GEE results are available upon request from the authors.

⁵ The results for these robustness tests are shown in Appendix B and they are very similar to our preferred specifications. The correlation of the equity instrument with equity market cap as a share of GDP is 43%, while its correlation with the dependent variable is low at −7%. First-stage regressions generate highly significant coefficient estimates of 0.33 and 0.35, for source and destination economies, respectively. For debt markets, the correlation of the instrument variable with total debt outstanding as a share of GDP is 33%, while its correlation with the dependent variable is just 13%. First-stage regressions generate highly significant coefficient estimates of 0.36 and 0.56, for source and destination economies, respectively.

⁶ Results from the GEE regressions are available from the authors upon request.

⁷ 2015 is the last year for which we have data for most explanatory variables. Capital control measures reported by Fernández et al. (2015) were available only through 2013, and we simply assume they remained unchanged from their latest values.

⁸ As another approach to develop appropriate governance measure values, we also estimate an ordered probit model to predict the thresholds of our measure of governance associated with the three capital account openness classifications. The model estimates a threshold of around −0.4 for switching between a capital account classification of 1 (most restrictive) to 0.5 (half-restricted), and a threshold of around 0.4 for switching between 0.5 and 0 (the exact threshold varies depending on equity and debt restrictions and the treatment of outliers). While the coefficient on governance is statistically significant, the model struggled to predict the economies in the half-restricted classification, likely because that classification has the widest variance in terms of governance.

⁹ Our assumptions on governance in the equity and debt scenario analysis are based on the average score according to the level of equity outflow restrictions. Results do not change materially if we assume governance to take on the average score according to the level of debt outflow restrictions.

¹⁰ Our models predict that in 2015 China would not have had any international portfolio allocation with its actual level of portfolio restrictions and governance measures, reflecting China’s relatively poor score on both of these indicators.

¹¹ The results from the FGLS estimation are also quite comparable to those we obtain using the predictions from the GEE estimation. The GEE predictions indicate China’s total outbound equity investment would be $1.8 trillion (16% of GDP) upon full liberalization and substantially better governance, compared with a fitted value of $0.7 trillion for 2015 (7% of GDP). We interpret this as additional equity outflow of $1.1 trillion (10% of GDP) due to liberalization. The GEE predictions indicate China’s total outbound debt investment would be $2.3 trillion (20% of GDP) upon full liberalization and substantially better governance, compared with a fitted value of $0.2 trillion for 2015 (2% of GDP). We interpret this as additional debt outflow of $2.0 trillion (18% of GDP) due to liberalization. In total, the GEE model predicts additional portfolio outflows of $3.1 trillion (28% of GDP) at the upper bound, which is in line with the FGLS results shown in Tables 2 and 3. For the sake of brevity, we do not present the GEE estimation and prediction results here, but they are available from the authors upon request.

¹² While financial centres such as Hong Kong and Singapore are not included in our model regression, we can apply the model coefficients to estimate the amount of Chinese portfolio flows they would receive.

¹³ See Hatzvi and Nixon (2015) for a description of China’s portfolio outflow programs.

¹⁴ This excludes Luxembourg.

¹⁵ See Agbloyor et al. (2014); Cerutti and Puy (2015); Choong et al. (2010); Durham (2004); Eichengreen, Gupta, and Masetti (2017); Ghosh and Qureshi (2016); Hoggarth, Jung, and Reinhardt (2016); Igan, Kutan, and Mirzaei (2016); Pagliari and Hannan (2017); and Prasad and Subramanian (2007).

¹⁶ The model’s estimate for gross investment into China should be treated with caution, given China’s combination of having large financial markets and little overseas investment. In addition, the economic significance of the inflow capital control variable is low. Nevertheless, the projection for significant increases in investment outflows and inflows is consistent with earlier work by Kruger and Pasricha (2016).

¹⁷ The trilemma is a well-known policy challenge whereby a country can only choose two of the following three options: an independent monetary policy, a fixed exchange rate and an open capital account.