Experimental Design
University lab subjects will be recruited to participate in experiments in the Texas A&M Economic Research Laboratory. In each session, subjects will be randomly assigned to the role of buyer or seller, and will be assigned to negotiate over a trading price with randomly matched participants in the opposite role. There will be two treatments assigned as the session level: Voice and Modified Voice. Half of the sessions will require negotiation via Zoom audio without chat; the other half will use the same interface but with pitch-modified audio to obscure the voice (and thus, gender) of participants.
Participants are randomly assigned to the role of buyer or seller and keep the same role for all four rounds of negotiation. Each negotiation is with a new, randomly-matched partner of the opposite role. In each negotiation round, the buyer is assigned an integer-valued private value, drawn randomly from the range [45, 46, 47, . . . , 55], and the seller is assigned an integer-valued private cost, drawn randomly from the range [10, 11, 12, . . . , 20]. In each round, the buyer and seller negotiate over the trading price. The buyer’s profit is equal to the private assigned value minus the agreed-upon trading price; the seller’s profit is equal to the agreed-upon trading price minus the private cost. Participants are given five minutes to negotiate with their partners. If five minutes pass without agreement, then both earn a profit of zero. Costs of delay are introduced as follows: buyers’ private values are reduced by 1 and sellers’ private costs are increased by 1 every 20 seconds beginning at the 3-minute mark until the end of the negotiation. All participants will also respond to a post-experiment survey.
We predict that the gender gap in earnings will be smaller when voice is modified. We also predict that sellers will, on average, earn less than buyers.