This Economic Policy Uncertainty Index, Using Baker, Bloom and Davis methods to calculate:
figure 1: Hong Kong Economic Policy Uncertainty Index (blue). source: Luke et al. pass policyuncertainty.com.
Given the high degree of uncertainty regarding the direction of political control in Hong Kong, the recent downward trend seems strange to me. On the other hand, perhaps economic policy is now considered static (in Beijing), so there is little uncertainty about the direction of economic regulation and the environment. Or the uncertainty of economic policy is highly correlated with the uncertainty of the economy, and it decreases as economic conditions and business conditions improve. (A good primer on the correlation between economic policy uncertainty and the economic uncertainty index is Ferrara, Lhuissier and Tripien (2017).)
figure 2: Hong Kong GDP (Ch.2019$) growth rate, y/y,% (red, left scale), and expected changes in business conditions (blue bar, right scale), from Chart 1 The Government of the Hong Kong Special Administrative RegionThe orange point for the second quarter of 2021 is an advance estimate of 7.5% y/y, as stated in the report Hong Kong Monetary Authority (August 2, 2021).
Another explanation is that as the sample of newspapers has shrunk (now Apple Daily is one of the source newspapers), the reporting behavior of other newspapers may have changed. Originally scheduled to be: Wen Wei Po, The Star, Ming Pao, Oriental Daily, Hong Kong Economic Daily, Sing Tao Daily, Hong Kong Economic Times, Apple Daily, Hong Kong Commercial Daily, Ta Kung Pao.
There is no real way to easily test any of these hypotheses, and I can use the available data. What I can answer is whether and when the relationship between the uncertainty index of China and Hong Kong has changed.
image 3: The Hong Kong Economic Policy Uncertainty Index (blue) and China’s Economic Policy Uncertainty Index come from Davis and Liu Hesheng (red), both on a logarithmic scale. source: Luke et al. pass policyuncertainty.com, and Davis et al. pass policyuncertainty.com.
To this end, I estimate the regression of log HK EPU to log China EPU and use the one-step advanced Chow test of recursive residuals to check for structural fractures. This method uses a regression to the YYYY.MM period, and then checks whether the prediction error of YYYY.MM+1 may have occurred randomly (roughly speaking). The results showed that there was structural fracture in 2017M07 (significance level is 0.0135). This is a short time after the Hong Kong Chief Executive election held in M03 in 2017.
image 3: The recursive residual (black) and the marginal probability of one-step advance Chow test (blue circle). The linear regression residual of log HK EPU to log Chinese EPU. Source: Author’s calculations.
Alternatively, you can view the n-step advanced Chow test to see where possible structural fractures are. This is the case of performing a series of Chow tests on all possible breakpoints in the sample.
Figure 4: Recursive residual (black) and marginal probability of n-step Chow test (blue circle). The linear regression residual of log HK EPU to log Chinese EPU. Source: Author’s calculations.
These results show that 2017M01 has structural fracture (significance level is 0.0116). This is slightly earlier than the Hong Kong Chief Executive election held at M03 in 2017.
This data is consistent with the interruption of correlation between the China and Hong Kong series, which occurred before and after the election of the Hong Kong Chief Executive or when the Hong Kong National Security Law was implemented. It can also be seen that-using the correlation between China and Hong Kong’s uncertainty index-the reported Hong Kong EPU has been particularly low in the past few months (although it is only significant when it is below the 5% significance level).
Obviously, the above is an informal analysis. Another approach is to link the HK EPU to some macro and/or financial variables and test whether these relationships are structurally broken.
