The economic cost of “decoupling” from China
Economists have argued that, just as there were gains from trade and investment links with China, there will be a cost to decoupling. Some analysts have built complicated econometric models to predict the possible economic consequences based on scenario analysis. However, in view of the complexity of US–China economic relations, cost–benefit analysis is very subjective. I have recently tried to provide an alternative approach to estimating the potential cost of decoupling by examining the responses of stock markets.
My article in the new issue of Economic Affairs uses Google Trends search results to measure investor sentiment towards decoupling from China. A high-frequency weekly dataset of these results was created, running from 5 January 2020 to 20 June 2021. It was then linked to the Dow Jones Industrial Average (consisting of 30 prominent companies listed on stock exchanges in the US), the S&P500 (the top 500 largest companies listed on stock exchanges in the US), and the NASDAQ Composite (mainly the information technology sector), by employing generalised autoregressive conditional heteroskedasticity (GARCH) models. Because of the generally high efficiency of stock markets in reflecting available information, their responses to the US–China decoupling are important indicators of economic impacts.
A word on Google Trends: this is a Google product that looks at the popularity of Google search queries in various nations and languages. Its properties include anonymity, topic categorisation, and aggregation. It has been applied through hundreds of studies in various fields such as information systems and computer science, health care, political science and international relations, as well as economics, business, and finance.
In particular in finance, it is widely accepted that Google Trends data can be used to predict stock returns. However, how to interpret the data varies. Most of these studies interpreted Google Trends as a measurement of investor sentiment, that is, the behavioural factor. A few others interpreted Google Trends as a combination of fundamental and behavioural factors.
From the viewpoint of communications, Google Trends can be interpreted as a measurement of the public agenda, that is, those issues that the public thinks are the most important.
My study created a weekly index in the US using the key phrase ‘decoupling China.’ This index shows the normalised volume of the decoupling-from-China narrative in the US and is used as a proxy to measure investor sentiment. The term “investor sentiment” is primarily interpreted as a fundamental variable – i.e., based on real economic costs – rather than a behavioural indicator. The index shows that the largest peak in the decoupling-from-China narrative occurred in August 2020, when then-President Trump talked publicly about a possible full decoupling from China.
The results based on GARCH models show that all the composite indices responded negatively to the decoupling index. While the Dow Jones Industrial Average represents only a very small number of US firms, the S&P 500 and NASDAQ indices are more representative of the US economy. The implication is that the negative effect on the US economy from decoupling from China is likely to be very considerable. Results also show that decoupling can cause higher volatility in composite stock indices. As to the industry level, the same conclusions generally hold. Robustness tests using daily Google Trends data support these results.
Stock markets are efficient in the sense that they almost always reflect all accessible data. While most complicated models may not account for all variables, the power of stock markets is that they provide an expected value for all possible scenarios. My study provides an alternative approach to estimating the potential cost of US-China decoupling and contributes to the ongoing debate in the USA.
Kerry Liu’s article ‘The economic impact of America’s decoupling from China: a Perspective from Stock Markets’ appears in the latest issue of Economic Affairs. Journal subscriptions are available at https://ordering.onlinelibrary.wiley.com/Lite/Subs.aspx?doi=10.1111/(ISSN)1468-0270&ref=1468-0270