Reducing gender discrimination in MSME lending by informing loan officers of their own implicit bias

This project will examine the causal impact of informing loan officers about their potential gender biases on their lending behavior and loan portfolio performance. We will assess this by first administering an Implicit Association Test (IAT) to loan officers, and then randomly varying the timing of score disclosure. Some officers will receive immediate feedback on their IAT score, while others will receive the feedback later. This will allow us to evaluate whether loan officers alter their behavior after becoming aware of their potential biases.

Loan officers participating in the study will be asked to complete an online tool that includes the following modules:

Module I. Implicit association test: The module looks at the strength of associations between concepts (e.g., men, women) and stereotypes (e.g., business, family) with the Implicit Association Test (IAT). The test consists of a series of words that are presented on the screen that the respondent must classify, for example, as associated with the categories "family" or "business" (in some sub-blocks) or "male" or "female" (in other sub-blocks). In the concluding sub-blocks, the four categories are paired into two groups: one "congruent" with usual social associations (female-family and male-business), and the other "incongruent" with unusual social associations (female-business and male-family). Respondents are required to categorize words (i.e. "children", linked to "family") into the appropriate categories (business and family) within 5 seconds. The logic behind the test is that response speed is unconsciously influenced by irrelevant ranking categories. In simpler terms, responding becomes easier when closely linked items share the same response key (Greenwald, McGhee, & Schwartz, 1998). For example, individuals that have gender biases may respond more quickly to identifying the word "child" ("sales") when the correct category "family" ("business") is presented on the same side as the category "female" ("male"). However, their response may be slower when the correct category "family" ("business") is positioned on the same side as the category "male" ("female"). The module consists of two test sub-blocks, each with 5 repetitions, to familiarize respondents with the activities. It is then followed by four evaluation sub-blocks, with 25 and 50 repetitions each, of the IAT.

Experiment: After bank officers complete this module, those assigned to the treatment group will receive additional information on their IAT results, including some suggested practices to limit the influence their automatic associations have on their workplace decisions and behaviors. Those in the control group will receive this information at a later time.

Module II. Short demographic survey. Respondents will also answer a short survey asking them about their household composition, years working in the sector, household income, and whether they have hold a loan themselves in the past years.

2024-10-02

2024-12-20

Loan rejection rates, loan size, interest rates, guarantor requirements, borrower repayment behavior

The impact will be assessed by randomly varying the timing of disclosure of loan officers' potential gender stereotypes. To begin, loan officers' gender biases will be measured using an Implicit Association Test (IAT). After the test, half of the loan officers will receive immediate feedback on their IAT score (the treatment group), while the other half will receive their feedback six months later (the control group).

Loan officer behavior and portfolio performance will be tracked and evaluated over the six-month period following the test.

Randomization done in office by a computer.

Individual level randomization

No

Not a clustered design

160 individuals

80 treated and 80 control