How are Google searches correlated to UK unemployment?

Posted on 11th September 2017

The volume of people searching on Google where the search term includes the word "unemployment" is correlated to the future unemployment rate.

The relative numbers of people searching Google for terms including "unemployment" is influenced by many factors. One of these factors is seasonality throughout the year. In July and August there is a 43% reduction in people searching Google for "unemployment", compared with the January to March which typically is the most common time of year for people to search for "unemployment".

A second significant influencing factor is macro-economic events, the most significant of which was the financial crisis starting in 2008. This can be seen in the search data by the distinct spike in people Googling unemployment between October 2008 and June 2009. From October 2008 there was an sharp spike as the financial crisis first unfolded followed by a prolonged period of relatively high volumes of people Googling "unemployment". The 12 month average didn't fall back to pre-crisis levels for 8 years. The volume of people Googling "unemployment" has continued to fall and is now at its lowest 12 month and 3 month rolling average of this 14 year period of analysis.


There is actually no correlation between the volume of people Googling "unemployment" in a month and the official UK unemployment rate for that month. However, if you offset the time periods so that "unemployment" search volumes leads the actual unemployment rate, a positive correlation builds.

The correlation coefficient (the degree to which two sets of data move in tandem) gradually builds until there is the strongest correlation when Google search data leads official unemployment data by 5 months. If the data sets are shifted by more than 5 months then the correlation gradually reducs back to 0 when Google search data leads official unemployment data by 11 months. The positive correlation between people Googling for search terms including "unemployment", and the official unemployment rate 5 months in the future indicates that people start searching for terms related to "unemployment" before they are actually unemployed. This is of course a very logical behaviour as people do tend to receive a lead time to their termination of employment usually of between 1 and 6 months.


Official UK unemployment statistics are not actually released until 2-3 months after the month being reported on. This means that this Google Trends data can be used to predict the actual official UK unemployment rate about 8 months in advance of the release of official figures.

As the proportion of people searching for unemployment in the UK on Google is at historic lows I would predict that in Feb-Mar 2018 the official UK unemployment rate will fall from the current, already low levels. Of course, as seen above, the correlation between Google Trends data for "unemployment" and the official UK unemployment data 5 months forward is not strong, and there are many other factors that would be taken in to account for a robust estimate of the actual unemployment rate 5-6 months in the future, therefore it is not likely that this prediction will actually be very accurate.

Google Trends search terms for "unemployment" searches versus "claim benefits" searches