Wednesday, September 28, 2011

How Good Are We at Predicting the Future (like the 2012 Presidential Election)?

As many of you know, every four years I throw out my prediction for who is going to win the Presidential election. So far, I've only been "wrong" once when in 2004, I "punted" and didn't declare a winner (however, I did try to place a bet with "Intrade" on "W", but I couldn't get the site to take my credit card). It's a bit early to predict next year's election
(however, if the economy doesn't pick up between now and next summer, Republicans will control the House, Senate, and Mitt Romney will be sitting in the Oval Office come January 2013),
but evidently I'm not alone in my propensity to predict as a recent Freakonomics podcast noted ("The Folly of Prediction"). Specifically, the podcast noted that
  • Human beings love to predict the future
  • We're not very good at it, and
  • Because bad predictions are rarely punished, this situation is unlikely to change
This podcast explores why humans are so bad at predicting. One individual interviewed, Philip Tetlock, documents just how bad we are. Tetlock is a psychology professor and author of Expert Political Judgment. He conducted an experiment that asked nearly 300 political "experts" to make a variety of forecasts about dozens of countries around the world. After tracking the accuracy of about 80,000 predictions over the course of 20 years, he found that they weren't a whole lot better than the average person on the street. So much for being an expert.

The podcast also notes how the predictions we remember tend to be the ones which were wildly unexpected and then came true. Steven Levitt, one of the Freakonomics co-authors, notes how every year his mother predicted that there would be a stock market crash. She has only been right twice, but she conveniently forgets the numerous times she's been wrong and only talks about the two times she's been right.

Levitt's mother is not the only one. The podcast notes that if you look at all the people, the economists, who talked about the financial crisis ahead of time, those guys constantly remind anyone who will listen: “I was right, I was right, I was right.” But when people are wrong, there’s no person on the other side of the transaction who draws any real benefit from embarrassing you by highlighting the bad predictions. In other words, because there is no market mechanism or incentive for keeping the prediction makers honest, there’s lots of incentive to go out and make wild predictions.

Imagine, however, if every time a pundit appeared on TV, his or her "batting average" appeared next to their name, letting viewers know just how good they are at predicting. That would be an incentive not to make bad predictions, or at least not opining in areas in which you have little or no expertise.

There is at least one arena where bad predictions are punished and are aptly called "prediction markets." These are speculative markets where people "bet" on various outcomes (e.g., who will win the next Presidential election, whether the economy will go into a recession by December 2012, whether there will be a "successful" WMD attack by December 2013, whether the Yankees will win the World Series, etc.) where people are rewarded for making accurate predictions and punished for making inaccurate ones, which is why advocates of prediction markets argue that they are better at predicting outcomes than most experts.

The most popular prediction market is probably "Intrade" although the Iowa Electronic Market is the oldest (and from my limited experience less volatile and more accurate than Intrade). The way it works is that if you buy 10 shares of, say, President Obama winning the 2012 election for $4.74 per share (the current price on Intrade), then if he wins, you will receive $10.00 per share or $100.00 and have made $52.26. The current price of a share is interpreted as the probability that the outcome will occur. That's right, currently Intrade is predicting that there is a 47.4% chance that President Obama will be reelected (the Iowa Electronic Markets currently have Obama at a 47.6% chance of winning), which largely reflects the current state of the world economy. If it picks up, then so will Obama's chances. If it doesn't, then he will probably be a one-term President. We will see what we will see.

If you want to learn more about "The Folly of Prediction" listen to the Freakonomics podcast. Or, if you don't want to listen, you can also access the transcript of the podcast: "The Folly of Prediction: Full Transcript."

No comments:

Post a Comment