Gender-differentiated Credit Scores: Bridging the Gender Gap in Access to Credit

This post was written by Sean Higgins, Postdoctoral Fellow at CEGA. Sean is currently studying digital payments, savings, and gender-differentiated credit scoring algorithms in the Dominican Republic and Mexico.

Focus group discussion in the Dominican Republic (Photo Credit: Única Mendez)

When it comes to accessing credit, low-income women in developing countries find themselves stuck in a perpetual data void.

Although evidence suggests that women default less than men do on loans (Karlan and Zinman, 2009; D’Espallier et al., 2011), low-income women disproportionately lack credit histories, property rights, and formal earnings — all factors that restrict their access to formal credit and prevent them from building the credit histories they need to access productive assets and achieve financial stability. In other words, a data gap exists that prevents lenders from being able to assess the creditworthiness of low-income women, and from lending to those who would be most likely to repay and who would most benefit from credit.

Credit algorithms using big data from mobile phone call detail records (CDR) provide an opportunity to overcome the gender gap in access to credit. In recent years, digital credit — credit products that use nontraditional data to make instant, automated, and remote loan decisions and disbursements — have expanded rapidly in developing countries (CGAP, 2016). By the end of 2014, Kenya’s M-Shwari (a partnership between the mobile network operator Safaricom and the Commercial Bank of Africa) had provided over 20 million loans to 3 million unique borrowers (Cook and McKay, 2015), while Tanzania’s M-Pawa (Vodacom and Central Bank of Africa) has lent to 5 million borrowers (Aglionby, 2016). Because these products include instantaneous loan approval and disbursement, they are particularly useful to help the poor cope with income shocks (Bharadwaj, Jack, and Suri 2019).

To learn more about the experiences of low-income women without credit histories, we conducted focus groups in three distinct regions of the Dominican Republic: the east (La Romana), the west (Santiago), and the north (Puerto Plata). A number of insights from the perspective of these women emerged from the focus groups:

1. Low-income women perceive that without a credit history, it is nearly impossible to access credit, except in cases where the loan applicant has a formal job and receives a payroll invoice. Since many low-income women do not have formal, steady jobs, this excludes them from credit access.
2. Taking loans from informal lenders at extremely high interest rates is a nearly universal practice among low-income women excluded from credit markets due to their lack of a credit history. These women reported taking these expensive, informal loans when facing health-related emergencies or food insecurity.
3. When asked about differences in access to credit between men and women, low-income women without credit histories noted that because men were more likely to hold a formal job in the household, they were more likely to be able to access credit. In addition, they pointed to the sexist machista culture as a source of bias against women reinforced by lenders. When the women applied for credit, for example, they were almost always asked more about their partner’s credit profile, employment status, and savings habits than their own. A number of women reported not being allowed by their partners to manage household finances even though they would like to; furthermore, they felt that access to credit would provide them the bargaining power to have a say over household finances.

Because algorithms based on CDR data do not require a formal labor earnings history, we expect these algorithms to be less biased against women. Our findings from the focus groups also provide preliminary qualitative evidence about the potential impacts of using gender-differentiated credit scoring algorithms based on call detail records to bridge the gender gap in access to credit.

In a large-scale study funded by CEGA’s Digital Credit Observatory, UN Foundation, USAID, and World Bank, we will construct a gender-differentiated credit scoring model using CDR data, and then evaluate the impact of loans for women identified as “good borrowers” by CDR-based algorithms (but who would not otherwise have access to credit using traditional credit scoring methods) using a randomized control trial. Based on the focus group findings, we hypothesize that loans provided to these women will increase financial stability and increase their intra-household bargaining power.