Predicting the outcome of an election is easier said than done – as amply demonstrated by the last U.S. presidential election and the 2013 B.C. provincial election.
The pitfalls of polls – among them a decline in landline phones, potential biases in online polling and widespread distrust of unknown callers – are fairly well known. And while that doesn’t doom pollsters to get it wrong, it raises the question of whether there exist other tools for getting it right.
Werner Antweiler, an associate professor at the University of British Columbia’s (UBC) Sauder School of Business, is quick to challenge the idea that prediction markets are superior to polls in determining electoral outcomes. But where polling may fall short, trading may have answers.
“Markets are aggregators of information. When we see stocks trading, that’s essentially about predicting the future earnings of companies,” said Antweiler, who serves as market maker for the school’s three 2017 provincial election markets. The markets, which opened in March, are non-profit and will remain open to investors until May 8.
“What is different in terms of markets versus polls is that markets are looking at new information and can act very quickly to new information that arrives,” he said.
“Polls are kind of backward-looking. They ask the question, ‘Who are you going to vote for?’ And that’s already two or three days old. Markets ask a different question, which is, ‘Who do you think is going to win the election?’”
Prediction markets aren’t new. They exist around the world commercially, accepting real money for futures contracts on upcoming events. UBC has been running prediction markets for two decades, and Antweiler said their success in predicting provincial and federal elections has been varied.
“In 2013, the polls were off, and so were our traders, except those traders who took a contrarian view at the time,” he said.
The platform allows investors to participate in three different online markets, making bets related to the popular vote, legislative assembly seat share and majority government.
In the latter, for example, investors can pay $0.53 for a BC Liberal majority contract. If the Liberals win with a majority on May 9, the trader’s contract will pay out $1.
As of the April 26 market close, bids for NDP majority contracts were $0.31, and bids for any other outcome were $0.155. These contracts, too, will pay out $1 if correct; the rest will lose. The prices indicate how much investors are willing to pay for a contract, in theory based on which party they think will win and whether current market prices accurately reflect that.
“Futures markets are very good at processing information and helping predict future states of the world,” Antweiler said. “Market prices reveal essentially what is the likelihood of one event or another.”
He noted that in 2013, some traders paid $0.10 on Liberal majority contracts to later earn back 10 times their investment.
Unlike polls, which aggregate voter answers to specific questions, markets produce data reflecting what participants believe will occur. However, as was illustrated in 2013, the markets are tied to polls, as well as news and other information sources and even trader intuition. But market participants are motivated to sideline personal preference, said Antweiler, who will be watching the prediction markets and the polls leading up to the election.
“By putting money where your mouth is, you have a strong incentive to get right who is likely to win the election, no matter who you think should be winning. It’s just adding one more data point to our predictions – and choosing a different channel. What we need is robustness in predictions, and basically getting it from many different sources is probably better than just relying on one source.” •
