Is the labor market turning? Latest data from Google shows a pick up in search interest for the team “unemployment benefits”. We’ll need a couple more months of data before we can begin to feel confident a trend is under way. (1/13) 
If you’re new reader, this thread takes a look at whether Google search interest data is a useful predictor of US recession timing, and seeks to make an educated guess on when the next recession will arrive based on the data, which comes from trends.google.com. (2/13)
Google assigns an interest value to search terms based on a scale of 0 to 100, explaining that “a value of 100 is the peak popularity for the term. A value of 50 means that the term is half as popular. A score of 0 means there was not enough data for this term.” (3/13)
It’s important to mention that search interest data from Google Trends is sampled, so values can fluctuate even within the course of a day. You can read more about data sampling here: support.google.com/trends/answer/… (4/13)
This chart shows how elevated search interest for the term “unemployment benefits” was associated with a risk of recession during the last business cycle. From 2006 until 2009, as the economy moved closer toward recession, interest grew in a non-linear way. (5/13)
The last 4 years or so of search data show that interest in the term “unemployment benefits” again appears to be growing in a non-linear way. Does that mean a recession is imminent, and if so, when can we expect its arrival? (6/13)
History has that when the unemployment rate rises by .5 percent over the previous year, the US typically enters a period of recession. (7/13)
This scatter plot shows the association between the change in search interest for the term “unemployment benefits” and the change in unemployment rate over a 1-year period. (8/13)
The trendline associates a 50% yoy increase in search interest for the term “unemployment benefits” with a .5 percent lift in the unemployment rate over the previous year, a jump that has corresponded with recession starts. (9/13)
Note that the association between the two variables is weaker around the 50% mark, so if we want more confidence, we may favor 100% as more accurate benchmark for a recession signal. (10/13)
If we extrapolate the current trendline, and assume that a recession start around the time business conditions decline to a point where we see the 4-month average search interest for unemployment benefits reach 50%.. (11/13)
...then we might expect an NBER-sanctioned recession to start some time in late Q2 of 2020, perhaps by May. This anticipates an earlier onset than the last time we estimated its arrival. (12/13)
For those who prefer benchmarking search interest growth to 100%, then we would expect a recession by November of next year. Either way, the model is predicting a recession beginning in year 2020. (13/13)
