EM bonds, with a larger allocation to Chinese bonds, offer significant diversification benefits to a global portfolio of financial assets. We analyse the quantum of this diversification benefit using frequentist and Bayesian data science modeling in Python, under the classic mean-variance portfolio optimization framework. The expected returns and expected volatility were taken from Blackrock and JPM Asset Management publications in early 2021.
Frequentist tests on equality of variances mostly rejected the null hypothesis that increased EM bond exposure did not reduce the portfolio's expected volatility. Bayesian modeling then allowed for a more precise estimate of the volatility reduction generated by an increased allocation to EM bonds.
A description of the research methodology and the statistical findings are presented in https://towardsdatascience.com/portfolio-diversification-with-emerging-market-bonds-ef1ec966531a?sk=133b7b4d2c1629d5f9a37282bdc08801