With the emergence of Google as the dominant internet search engine, its search-term usage can provide a real-time view of current public interests in numerous issues such as economics, politics, health and so forth. Thus, if large groups of people are entering certain economically-linked search terms, this could provide a clue about the general public’s financial stress levels and the onset of a recession.
Whilst there are several search-terms that effectively provided warning of the 2008 U.S recession, we need to caution that data only exists since 2004 and we are dealing with a sample set of one recession (the 2008 financial crises) as a baseline, which makes for a very small sample set. Thus the deployment of Google search trends as a recession forecasting tool on its own is highly risky and it should rather be deployed together with longer-term more proven macro-economic tools.
The one advantage we have with Google search trends is timely readings, as data is available on a daily basis as opposed to most traditional economic variables which have a 1-week or a 1-to-3 month lag.
There is currently a widely circulating post that looks at domestic searches for “coupon”, “unemployment”, “jobs”, “TV” and other media sources “YouTube” and “Netflix” to conclude that “…if the Federal Reserve looked at Google Trends they might not be so inclined to cut rates next week.” However we feel these particular search terms are not suitably robust enough as they either lag too much to recession (unemployment, jobs) or display highly seasonal annual patterns or did not provide convincing enough lead or overlap to US recession.
In fact, the 7 domestic search terms we have found as a result of an exhaustive analysis since 2015 provide quite the opposite view, namely according to the searching public, recession is imminent or already upon us!
So whilst we do not wish to dwell upon the argument if the search terms “coupon”, “TV”, “unemployment” and other media sources are suitable or not, what we do wish to demonstrate is that an equally defensible set of 7 search terms are showing quite the opposite, to at least balance out the currently circulating argument that there is nothing to worry about.
Let us get into the 7 domestic search terms we have found, that in our view, provide the best warning to the 2008 recession (our only available baseline at this point). Each search term is shown together with what percentage of its search volume contributes to the total search volume of all 7 terms, and we start with the most popular to the least popular search terms:
Now we have two ways of combining these all together – either in an equal-weighted average or a search-volume weighted average:
The volume-weighted average, with a trigger threshold of 25 and a 10-month lead to recession, implies recession is 7 months away, due in February 2020.
The equal-weighted average, with a trigger threshold of 35, and a 6-month lead to recession, implies recession is already upon us.
At this point we do not know if the search-terms are being generated by financial professionals, private investors or the general public. The most likely search-terms being generated by the general public are “recession”, “refinancing” and “debt relief”, although an argument exists that “rate cut” could also be forming part of their search repertoire if they are under financial stress and exploring the likelihood of debt-related financial relief on the horizon.
As for the advice for the Fed reviewing these search trends, I doubt very much they need to, they are evaluating more sophisticated approaches one of which we have used to show that Fed rate hikes are now two standard deviations above the norm and a rate cut is very much due.
Again, the use of Google search-terms for forecasting recession is a very new field with a sample size of 1, a very short history and many questions about who are actually generating the searches, so one needs to be very cautious about reading too much into this, and rather treat it as input to many other traditional variables such as traditional long and short leading economic indicators.
You can view our full repertoire of economic indicators used by our clients to assess stock market and economic risks on the OUR SERVICES page which has recent sample reports no older than 3 months.