In financial betting markets, where experts and traders might have incentives to lie and mislead others in order to capitalize on their mistakes, proper incentives become especially important for accurate results, Chen said.
"If [experts] incur costs to get better information, then providing incentives is a good way for them to be motivated to make more accurate predictions," Chen said.
A "proper scoring rule" helps determine how much compensation a consultant ought to receive based on various factors such as the accuracy of the prediction, Chen said.
To explain her application of a "proper scoring rule" formula, Chen used the example of a meteorologist determining the probability of a hurricane striking Florida in any given year.
Chen's formula would calculate the reward for the meteorologist based on the prediction's accuracy whether the hurricane ultimately occurred. The formula would encourage the meteorologist to report his best estimate because it would maximize his payment, according to Chen.
"If we are rewarding the expert according to this formula, if the expert is rational, he will report truthfully," Chen said.
Chen also presented a hypothetical flu prediction market in which individuals would place bets on the number of people who would get the flu in a given year to demonstrate the potential for market manipulation. Because the Centers for Disease Control use the flu market predictions to purchase vaccines, a flu vaccine producer would have a financial incentive to falsely report its predictions for the incidence of illness in order to inflate market prices, Chen said.
"My personal take is that it's probably impossible to design a specific mechanism to consider information elicitation in the context of decision-making," Chen said. "If that's the case, then the more important question is whether we can think about a mechanism to limit the manipulation."
Proper scoring rules must also take into account the actual outcome of an event, according to Chen. In a company's search to hire a chief financial officer, for example, an expert could trick the system by recommending that a corporation hire worse candidates in order to receive a larger payoff, Chen said. For this reason, the formula must take into account all possible outcomes of the situation at hand.
Proper scoring rules could also incorporate the predictions of multiple experts by paying them based on how much they improve an aggregate prediction.
In turn, each expert would pay another expert based on the probability that the first expert's predictions are wrong, Chen said.
Chen received a master's degree in economics from Tsinghua University in Beijing, China, and a PhD in information sciences and technology from Pennsylvania State University. She received a Faculty Early Career Development Award from the National Science Foundation for her work in prediction markets.
