Nowcasting the Bear with Google Trends

  • -
  • J. Bughin

Dr Jacques Bughin, UN consultant, Solvay Business School ULB, Portulans Institute and G20Y, former Director McKinsey Global Institute, and senior partner McKinsey & Company.

Table of Contents:

  1. March 16

1. March 16

Back to Table of Contents

The stock market has taken a major downbeat again on this day, possibly as the result from fear mounting that COVID-19 is more difficult than imagine to contain in short time, further leading to inevitable sanitary risks, and breakdown of aggregate demand. Looking like I did in my previous article on Google trends, five points are becoming clear:

  1. “Stock market crash” and “recession terms” online searches are strongly and positively correlated, and have spiked five times in the last 5 years, e.g. end of Aug 2016, Early November 2016, Early Feb 2018, last three weeks of Dec 2018, and now recently from Feb 23. Those spikes are at the same time the stock markets were in sharply down territory, and the change in intensity of searches (especially stock market crash) is closely linked to the size of the drop, e.g. minus 1,400 points of the Dow Jones Industrial in Aug 2016 (for a spike of 2.6 times average crash term search intensity), minus 2,300 in Feb 2018 (for a spike of 4.5 times average search intensity), mid February fall of 6,000 points (and a spike of 6.7 times average search intensity for market crash).

  2. We find a strong correlation between stock market crash search intensity and Google search intensity for Coronavirus, lately. In fact, there is a positive correlation in level of about 65%, and in the range of 60-70% in the window (\(t-3\) days , \(t+3\) days). Typically 10 points increase in searches for cororonavirus leads to 6 points increase in stock market crash searches.

  3. Combining both 1 and 2, and making possibly heroic and simplistic maths, increase of coronavirus searches to peak worldwide has been more or less associated with a 10% points loss in the Dow Jones Industrial, or about 2,500 points.

  4. The stock market reduction is already twice higher than implied by online searches. The optimistic view will be that the market is over-reacting. The efficiency hypothesis side (‘market knows it all’) may rather suggest that search intensity for pandemic is uniquely high, and might have more disproportionate effects than what we found by simple linear extrapolation. Looking since 2004 (the earliest we got data on Google trends), market crash terms intensity levels of this current level for COVID-19 were only found by 2007 and 2009 at the previous crisis - as is the search for the term “recession”.

  5. This recession may be caused by supply factors (travel stop, disruption in value chains, deleverage affecting the company prospects and liquidity), but may be driven by a perception of damage on the consumption side, requiring a budget expansion fix if this shock is permanent.

In fact, we had looked in previous research as to how different category searches may affect sales. In effect, we have looked at how non-food shopping could nowcast retail sales, or aggregate private consumption. We also did so on how automobile searches could affect car sales, etc.1 The findings were that those category search intensity changes were able to nowcast next quarter spending changes, and in such a way, that the dynamic of searches we are witnessing now linked to COVID-19, may mean a drop of consumer spending, of rather large significance, possibly in the range of between 5 to 10%. If our analysis, by 2011, was anywhere right - where we managed to find a cointegration between searches intensity and retail -, it may also mean that retail spending might be affected structurally (read permanently).

We might thus be in a clear demand shock in the making here. Monetary policy might be short of a full fix; we need a New Deal plan perhaps, here. Should we then push the “pseudo-excuse” to invest budget a) in a more comprehensive and agile infrastructure for global fit in healthcare worldwide (so as to get prepared to next pandemics), as well as b) in rebuilding habitats for animals away from our cities, (so as to avoid the inevitable rise of scary zoonoses)?

© Jacques Bughin. Written March 15. Comments more than welcome. All errors are mine. References listed as they are found in the text.

Learning from the curve

An open source research project on COVID19 and economics. A collaboration between academics to reach out to policy makers and the general public.

Website design by Alessandro Gallina