Is investor sentiment (a-a-achoo!) contagious?

Does investor sentiment in one market influence sentiment in another?[1]. It’s a good question and in this post we investigate whether Axioma’s ROOF scores give any credence to the old adage, “When Wall Street sneezes, all markets catch a cold.” 

US and Japan

The table below shows the yearly average correlation and r-squared of the ROOF scores for the US All Cap portfolio (US-LMS) at t-1, with those of Japan All Caps (JP-LMS) the next day for the last four years. A few observations:

  • 2019 looks like 2016, in terms of the influence of US scores on Japan’s the next day
  • During 2016 and 2019, the US scores ‘explained’ 43% and 49% of Japan’s scores, respectively
  • In 2017, the year with the least influence from US scores, the Japanese market had its best performance

In a previous post on the recent US market rout[2], we introduced the idea of different institutional investor types: “Institutional investors fall into three temporal categories: short, medium, and long-term. Short-term investors tend to rebalance intra-week on any given day. Medium-term investors tend to rebalance monthly, and long-term investors quarterly. We match the ROOF scores to each investor type by looking at a 5-, 20-, and 60-day moving average of our daily sentiment scores.” The table below shows the correlation and explanatory power of the US ROOF scores at t-1, with that of Japan the next day for the three investment horizons: short, medium, and long. 


A few observations: 

  • As with the daily data, 2019 looks more like 2016 for this relationship
  • The correlation with and explanatory power of the US ROOF scores increases with the investment horizon (exception is 20D vs 5D in 2017)
  • 2017 was a year of independence for Japanese investors

US and UK

What about markets that overlap in their time zones[3]? In particular, we are interested in investigating whether the influence of US investors over their UK counterparts has changed with Brexit. Below is a chart showing the daily ROOF scores for the US All Caps at t-1 and the UK All Caps the next day from 2006 to 2019 (YTD). On average, the US scores at t-1 explain about a third (0.34) of the UK ones the next day[4] during that period[5]. But what about since Brexit?

The table below shows the correlation and r-squared of the US All Cap ROOF scores and their two components (risk tolerance and risk aversion) with their UK counterparts for 2016-2019. Some observations:

  • The US influence over the UK scores was also high in 2016
  • Since 2016, the explanatory power of US scores has more than halved from their historical level
  • In all four years, there is a stronger link between the risk aversion scores of both countries than their risk tolerance ones (i.e., they seem to get scared together, but greedy separately).


These results point to Brexit weighing on UK investor sentiment far more than what goes on in the US market. 

US and China

What about the other elephant in the room? China has historically been a pretty isolated and uncorrelated market with the rest of the world and the US in particular. Has the US-China trade war changed that? The chart below shows the daily ROOF scores for the US and China from 2003-2018.

For the past 15 years, there has been no ‘contagion’ between US investor sentiment and that of Chinese investors, but that all changed as the US-China trade war escalated. The chart below shows the correlation between the US All Cap ROOF scores and sub-components (risk tolerance (RT) and risk aversion (RA)) at t-1, with those of the China All Cap market portfolio the next day from 2015-2019 (YTD).

Some observations:

  • Correlations between the US and China scores went from essentially nothing to 0.38 in a single year
  • The link seems to come mostly from the risk tolerance sub-score, which has a correlation of 0.40 versus 0.31 for risk aversion (i.e., good news about the trade war has the same effect on both investor types, whereas bad news is received slightly more differently)
  • As with other markets, the relationship seems to get stronger with a longer investment horizon, although it peaks at 20-day instead of 60-day for others.

Conclusion

We know that local investor sentiment, proxied by risk appetite, influences market returns. We also know that global market returns are somewhat correlated with each other, especially in those markets where time zones overlap. Using the Axioma ROOF scores, we can also build a correlation matrix of investor sentiment and use it to predict how negative/positive sentiment in one market impacts a market that opens after it.

The table below is a correlation matrix (bottom diagonal) and associated R-squared (top diagonal) for the 9x9 relationships in 2019 (YTD). Note that only the US market is taken at t-1, all others are contemporaneous relationships. Notice the strong influence of Asia ex-Japan sentiment on Global EM. We also note that the US’ ROOF Score influence on global developed markets has declined in recent years and is much now much lower (0.43) than its 60% weight would suggest. The bottom line is that sentiment does influence risk appetite across markets, just like volatility and correlation do. Investors should take note of the supply and demand for risk assets driven by the balance of risk tolerance versus risk aversion in other markets when making risk-budgeting decisions in their own markets.


[1] Thank you, YiYang Guo, product specialist in Axioma’s HK office for asking it.

[2] Post available here.

[3] The UK has about a two-hour overlap with the US market before closing at 16:30.

[4] The r-squared for same day ROOF scores is about 0.40.

[5] The r-squared excluding 2008 is still 0.31.

Olivier d'Assier

Olivier d'Assier is Head of Applied Research, APAC, for Axioma and is responsible for generating unique regional insights into risk trends by leveraging and analyzing Axioma's vast data on market and portfolio risk. d'Assier's research helps clients and prospects better understand and adapt to the evolving risk environment in the Asia Pacific.