It is a capital mistake to theorise before one has data, opined Sherlock Holmes. But which clues are relevant? Investors track internet searches as part of their effort to understand the coronavirus. The pipe-smoking detective might not have approved.

Goldman Sachs has been monitoring Baidu and Google search volume on virus-related keywords such as coronavirus, Sars, Wuhan and surgical mask. They also track measures such as air pollution, property sales and traffic congestion.

The internet search terms seem a good gauge of the market mood. As the chart shows, the surge in searches in late February correlates well with the spasms of a jittery market, after the disease spread to northern Italy. But it does not give investors much of an edge. Last week Goldman warned investors they were potentially underestimating the impact of the coronavirus. Only a few days earlier, it had noted that Google searches for the virus were tailing off.

Chart showing how the markets are reacting to the fear of the Coronavirus. MSCI ACWI World Index versus Global google searches on coronavirus

Big Data has a patchy record when it comes to health. Starting in 2008, Google used inquiries about flu-like symptoms and remedies to track the disease. Its predictive power initially seemed impressive. But the search data overstated a severe outbreak in late 2012 by a factor of nearly two. Google Flu Trends was withdrawn in 2015.

That should not surprise. As any cyberchondriac knows, typing symptoms into Google is not necessarily a sign of illness. That is especially the case when the media is full of news of a severe outbreak. In the case of a new disease like Covid-19, interest in the topic extends far beyond those who fear they have been exposed.

But researchers report encouraging results using more complex models. Moreover, data can be supplemented from other sources. Infections cause an elevated resting heart rate and disrupted sleep. That can be monitored by the devices like those sold by Fitbit, now part of Google. A study using anonymised data found it significantly improves the ability to predict flu, the Lancet reported last month.

Exploiting such information raises privacy concerns. Investors should not count on getting more valuable data. Health authorities are another matter. As the struggle to contain Covid-19 shows, the case for better tracking is open-and-shut.

Our popular newsletter for premium subscribers Best of Lex is published twice weekly. Please sign up here.



Please enter your comment!
Please enter your name here