Skip to main content

Robot Auction Bidders Are Such Buzzkills


Competing against automated agents induces auction bidders to skip risky behaviour

Robot Auction Bidders Are Such Buzzkills

The genteel wave of a bid paddle, the rat-tat-tat of auctioneer voices, the crack of a gavel — that’s all so yesterday. Electronic financial markets and sophisticated online platforms such as eBay are the new norm, bringing with them automated bidders. These days, the chances of bidding against a computer are higher than against another human; in fact, computerized traders are responsible for 70 percent of the volume on U.S. stock markets.

That trend got Ryan Riordan asking: How does the move towards automated bidding affect the behaviour of human bidders in online markets?

“When trading in financial markets, I found that most of what we were doing ended up being electronic,” says Riordan, Distinguished Professor of Finance at Smith School of Business. “When I was presenting results from previous work, people always seemed to be appalled by the thought they were trading against a computer and not with another human being.”

Riordan, and colleagues Timm Teubner (Karlsruhe Institute of Technology) and Marc T. P. Adam (University of Newcastle, Australia), designed an intriguing study to learn more about the effect of computerized agents in auction settings. Working out of the Karlsruhe Institute of Technology, the researchers tested the real-time emotions of 120 bidders in two settings: where the participants competed against two human bidders (referred to as a high-agency setting); and where they competed against two computerized agents (a low-agency setting). The low-agency experiment took place two weeks after the high-agency experiment, and computerized agents used bids previously submitted by human counterparts. The only difference between the two events was the participants’ perception of whether they were competing against humans or computers.   

Throughout the experiment, the participants’ heart and skin conductance rates were monitored to track their levels of overall arousal, along with the intensity of their emotions at three distinct moments: just before participants submitted their bids; the moment they saw an information screen; and upon learning of the final outcome of the auction.

Robots Dampen Emotions

As it turns out, the study showed that both bidding behaviour and overall emotions are affected by the belief a person has about whether they are interacting with a sentient being or not. Past research has shown that such beliefs can influence emotions, especially when it comes to competitive tasks – meaning that people tend to get more riled when they’re beaten at chess by a friend than by a computer.  

In this study, researchers found that emotional intensity spiked significantly when people knew they were competing against other humans, especially in high-stake moments with higher arousal, such as when auction outcomes were revealed. Winning an auction against computerized agents, however, still triggered stronger emotional reactions than losing in the high-agency setting.

Bidding behaviour was also affected by the presence of computerized agents. Often, part of the reason people take part in auctions is for the thrill of winning against other people. Past research has shown that, at times, the desire to win outweighs even potential monetary gains. This new study did, indeed, reveal that participants would place larger bids in high-agency settings.

The researchers also uncovered a twist. High arousal tends to lead to risky behaviour, and the researchers noticed exactly that among those competing against other humans. But the risky bets vanished once the participants played against computer agents. “In other words, lower levels of agency yield less emotional behaviour,” says Riordan.

“Be aware of the environment you’re operating in, and try to be aware of how your emotions may affect your bids.”

With social competition significantly shaping bidding behaviour, auction platform operators can effectively manipulate agency levels by ramping up or decreasing peer visibility online. To foster the elevated emotions that come with competition, operators can display profile pictures or usernames.

“If you’re a seller and want to achieve the highest price possible,” Riordan says, “it’d be best for your consumers to know they’re competing against both computers and humans, without knowing precisely which they’re directly competing against.”

And the takeaway for bidders? “Be aware of the environment you’re operating in, and try to be aware of how your emotions may affect your bids.”

In a larger context, the research raises questions about the speed at which we embrace technology that automates jobs previously held by people. “We don’t know how human behaviour changes when they start interacting with a machine rather than another human,” Riordan says.

“The more advanced markets get, the more grey area we’re going to see. How does our behaviour change when we interact with Siri (on the Apple iPhone) versus interacting with a human operator? We don’t yet know the effect these interactions have on people and how those interactions will change because of it. With everything changing so quickly, it’s worth looking into.