Experimental Design
We run an RCT with active financial advisors. The experiment investigates whether and how prior beliefs regarding gender affect the quality of advice to male and female clients. Are advisors using gender as a proxy for financial literacy and, in turn, offer higher prices to those perceived to be less literate? Do they update their prior beliefs if they learn about the client's financial literacy?
In the first round of the experiment, all advisors are confronted with a randomized sequence of the same ten bank client profiles. Profile information comprises age, marital status, education, wealth at the bank, and risk tolerance, and is modeled as closely as possible in accordance with actual profiles from our administrative bank data. The ten profiles consist of five pairs of profiles for which this information is almost identical, but differs by gender. However, only those advisors that are randomly assigned to the treatment group obtain information on the client's gender. In order to make this information salient but not suggestive of our research question, we reveal gender through photographic portrait pictures. For each of the ten profiles, advisors are asked to rate individuals' financial literacy, consideration of costs and fees, independence in securities decisions, and then also their own preferred front-end load, and their own confidence in these assessments using a scale ranging from 1 (very poor/not at all/very low) to 7 (very good/very strongly/very high/very confident).
Before the second round of the experiment, we inform advisors in both the treatment and control groups that we will again present them with a selection of six out of the ten customer profiles from the previous session. The six profiles, three pairs of female and male clients, are the same across all advisors, but the sequence of appearance is randomized. We opt for only six profiles to prevent survey fatigue, and we pick those three pairs that offered the greatest variability in terms of client attributes. We also inform advisors that those six profiles will contain two additional pieces of information compared to round one, namely the number of correct answers (rated from 0 to 7) that the client achieved in the financial knowledge quiz, and the client's confidence level in making private financial decisions as reported in an earlier client survey (rated from 1 to 7). We then task the advisors with reevaluating the five questions per client in the first round. In the second round, advisors can see their prior answers under each question. Participants must re-enter values in an empty answer pane to ensure that they do not stick to their earlier answers solely for convenience.