Experimental Design
We run an online controlled experiment. We recruit two sets of subjects using the online research platform Prolific. The first group of participants acts as “Firms”, whereas the next group acts as the “Workers”. We plan to manipulate the formation of entitlements and feelings of deservingness by providing feedback on subjects’ relative (superior) ranks in one treatment and not in the others.
There are two parts of the experiment, and one part is randomly chosen for payment. In the first part, both the firms and the workers complete a real effort task (where they count the number of zeros in a series of matrices) within 90 seconds. Each correct answer pays the participants 5 cents in bonus. In part two, the instructions tell that each firm will be assigned two (2) workers. One of these two workers will end up getting hired by the firm. When a worker is hired by the firm, the firm gets an additional bonus of 100 cents as “hiring credit”. In addition to the hiring credit, the firm also earns an additional 10 cents for each correct answer the worker provided in stage one (so that the experimental setup is relevant to a real-life employee-employer relation, where employee performance determines the employer’s payoff). We call this firm payoff the firm profit. Of this firm profit, either 30, 40, or 50 cents (randomly determined) is suggested to be paid as a “suggested wage” to the hired worker. Non-hired workers do not get any payment from part 2.
Workers are then given a chance to negotiate for their wages. If they choose to negotiate, they are asked the amount of money (in cents) they demand from the firm. They are told that the firm may accept or reject this negotiation request. If rejected, both workers and firms lose 25 cents of their payoffs. If accepted, the worker gets the negotiated amount they asked for, and the firm gets the firm profit minus the negotiated wage.
We implement three treatments. The first treatment is the entitlement treatment. In the entitlement treatment, we match subjects acting as workers in pairs. The worker whose score is the highest in the pair is hired by the subject acting as the firm. This information is shared transparently with the hired worker and the firm. The second treatment is the random treatment. In the random treatment, one of the workers in the pair is randomly hired by the firm, and this information is made clear to both the worker and the firm. The third treatment is the unknown treatment. Here, both the workers and the firms do not learn about the hiring mechanism (which can be either random or based on performance -- as in entitlement treatment).
Our main hypothesis states that women will not initiate wage negotiations as much as men do in the control (random) treatment, where they will not know about their relative ranks, and that the fact they got hired is based on chance. This finding highlights a previously documented finding of the gender gap in willingness to negotiate. In the entitlement treatment, where both genders know about their relative ranks (that they are hired because they scored better in their pairs), we expect to evoke the senses of entitlement and deservingness, which would result in women negotiating more often and the gender negotiation gap shrinking. This defines our second hypothesis. We use the unknown treatment to test the role of beliefs in negotiation decisions. Aligned with the two previous hypotheses, we expect that female workers who believe they were hired because they performed better in the pair will negotiate more often than the women who believe it was driven by luck.