Risk-averse bidders are cautious. Caution should have a predictable effect on bidding: bid more conservatively, win less often, overpay less. But “conservatively” means opposite things in different auction formats.
Note (2026-11-10): I checked the abstract of the paper (On Risk Aversion in Auctions). In first-price auctions it requires two payoff conditions (winning is never worse than the outside option, and winning with a low bid is preferable to winning only with a high bid) for greater risk aversion to make high bids more appealing; in second-price auctions it assumes a known outside option, under which greater risk aversion favors lower bids. The essay states the first condition only and omits the known outside option, and its conclusion that the preference has no inherent behavioral direction and is mediated entirely by the auction format is wider than the abstract.
In first-price auctions — where you pay what you bid — risk aversion makes high bids more attractive. Winning is never worse than the outside option, and winning with certainty at a higher price is preferable to the gamble of winning cheaply or losing entirely. A risk-averse bidder sacrifices expected profit for certainty of winning. The cautious response is to bid more.
In second-price auctions — where you pay the second-highest bid — risk aversion makes lower bids more attractive. Here, bidding more doesn't increase the price you pay when you win (the second price does that). But bidding more increases the probability of winning at unfavorable prices — you win more auctions where the second price is high. Greater risk aversion favors lower bids because higher offers increase risk exposure conditional on winning. The cautious response is to bid less.
Same psychological trait. Opposite behavioral consequences. The mechanism is the same in both cases — risk-averse agents avoid variance — but the mapping from variance to bid direction reverses with the payment rule.
The through-claim: the effect of a preference (risk aversion) on behavior (bidding) is mediated entirely by the institution (auction format). The preference has no inherent behavioral direction. It only acquires one when combined with a specific mechanism for converting bids into outcomes. This means you cannot predict how a cautious person behaves without knowing the rules they're operating under — and the same rules that make caution increase bids in one context make it decrease bids in another.