Figure 3
The horizontal axis is labeled with vertically rotated monthly ticks at roughly yearly intervals, including “2011 m 5”, “2011 m 10”, “2012 m 3”, “2012 m 8”, “2013 m 1”, “2013 m 6”, “2013 m 11”, “2014 m 4”, “2014 m 9”, “2015 m 2”, “2015 m 7”, “2015 m 12”, “2016 m 5”, “2016 m 10”, “2017 m 3”, “2017 m 8”, “2018 m 1 “2018 m 6”, “2018 m11”, “2019 m 4”, 2019 m 9”, “2020 m 2”, “2020 m 7”, “2020 m 12”, “2021 m 5”, “2021 m 10”, “2022 m 3”, “2022 m 8”, “ 2023 m 1”, “ 2023 m 6”, and “2023 m 11”. The vertical axis on the left ranges from 0.6 at the bottom to 2.0 at the top with an interval of 0.2. Four colored lines are identified in the legend centered below: a solid red line labeled “M C H I”, a blue dotted line with triangle markers labeled “G X C”, a green dashed line labeled “F X I”, and a black dashed line with open square markers labeled “C Q Q Q”. Across the graph, the region from “2020 m 2” to “2023 m 11” is shaded in gray from bottom to top. All four series start from “2012 m 8” between the values of 1.36 and 1.6, with the black curve with the square marker lying at the top and the other three curves clustered at the bottom. All the curves fluctuate with small variations until reaching 2020 m 2. After that, the curves decrease in the gray shaded region. The top black curve with a square marker ends around 1.1 in “2023 m 11”, while the other three curves end in the range of 0.86 and 0.90 at “2023 m 11”. In the lower cluster of curves, the blue curve is positioned at the top, and the red curve lies at the bottom before ending the shaded region, and then the red curve is positioned in the middle. These curves in the lower cluster follow the same trend with a small gap from left to right. Note: All numerical data values are approximated.Recursive estimation of the beta coefficients. Notes: This graph presents the recursive estimates of the beta coefficients for four Chinese ETFs over time, revealing a significant decline in all betas around January 2020, coinciding with the onset of the COVID-19 pandemic. Prior to this period, the betas remained relatively stable, indicating a greater responsiveness to the global market. Consistent with the break-regression results in this study, we have shaded the high-EPU regime (post-2020). The authors’ estimation