I’m not going to lie, I’m addicted to the Internet, specifically social media sites like Facebook and Twitter. Every time I check my phone I make sure to see the latest updates to my photos and statuses. Some of that I can blame on being a news junkie (I primarily use twitter as news aggregate), but I also spend time socializing with friends. I’m not the outlier on this issue; it’s the norm for adults to use social networking. According to the Pew Research Internet Project, 73 percent of adults use social networking.

Some economists from the University of Michigan have taken advantage of social media to predict the number of people who will file for jobless benefits. The University of Michigan Social Media Job Loss index accumulates tweets that use common phrasing like “lost my job,” to help with this prediction. The index is in real-time and accounts for detrimental events that would affect employment, like the 2013 government shutdown.


This new algorithm predicted that 342,000 people would file for jobless benefits last week. Economists in turn predicted that 320,000 people would file, but in reality 311,000 people actually filed. Why did the Twitter algorithm predict that more than 30,000 people would file than actually did? There are a couple of possible explanations for this.

  1. People could have lost their job but failed to file for jobless benefits, this could be because of severance pay.
  2. The algorithm indexes certain phrases that are associated with the loss of a job; however, these phrases have been taken out of context, implying that someone lost their job when they did not.

I think that this new way of predicting unemployment is long overdue. Social media is an integral part of our society and I believe that economists are moving in the right direction using Twitter to predict economic factors. However, the algorithm needs to be further tested.

What do you think about social media? Are economists correct to use twitter for economic predictions or do you think there will be too many misinterpretations with the statements.