Predicting Traffic Jams with Google Trends

My new IZA discussion paper on predicting traffic jams in Germany is out. Using hourly Google Trends data on “stau” searches we can predict the number of ADAC traffic jams reports. I am collecting the traffic jam reports from the ADAC website and use it to construct a crude proxy for road conditions which I then predict using Google search. The main point is this:

At 7:00 am there is a peak of Google search for “stau” and at 9:00 am there is a peak of ADAC traffic jam reports. In the afternoon the Google searches peak at 16:00 hrs whereas the ADAC traffic jam reports peak at 18:00 hrs. Google searches enable you to predict road conditions two hours in advance.

How important a problem is the phenomenon of a traffic jam you ask?

Adverse road conditions caused by traffic jams contribute to a host of undesired side effects such as increased carbon emissions, energy waste, additional transportation and production costs, waste of labor, delays in product deliveries and they also contribute to worsening health conditions as well as to more accidents and even road rage. If we thought of a city, a region, a country or any other social unit as a large living organism road traffic would be one of its circadian rhythms and traffic jams would be an obstruction to its entrainment. It is hence not surprising that obstructing the smooth flow of traffic sends ripple effects deep into many aspects of socioeconomic life.

According to the German automobile club ADAC, in 2014 there have been 475,000 traffic jams on German highways which amounted to 960,000 kilometers of jammed traffic. These numbers represent an increase of 14.4% and 15.6% respectively compared to the year before. This is just an instance of a many year trend which is only expected to get worse even though the report of the ADAC estimates that most of the current increases are due to progress in the method of documentation of traffic jams.  According to the ADAC these traffic jams amounted to 285,000 lost hours which is  32 years.

According to INVENT, a consortium of German automobile manufacturers, on a daily basis traffic congestion accounts for:

  • 33 million litres of wasted fuel,
  • 13 million hours of delay and
  • an economic damage of 250 million euros.

Understanding and forecasting road conditions is hence an important socioeconomic problem.

After controlling for day of the week and hour of day fixed effects I show that searches for “stau” improve the forecasting of the number of ADAC traffic jam reports significantly explaining well over 80% of its variation: a 1% increase in stau searches now implies .4% increase in traffic jam reports two hours later.

Finally more often than not these searches are accompanied by a geographic attribute which may be a city (e.g. “Stau Berlin”), a region (e.g “stau NRW”) or a highway (e.g. “stau A3”). This means that knowing where these searches emanate from and which highway they are interested in we obtain the origin and path of an upcoming itinerary. Traffic planners should use Google searches for traffic congestion, forecasting and prevention.

This entry was posted in data, economics, technology and tagged , , , , . Bookmark the permalink.