Inequality, Discrimination, and the Financial System, Spring 2021

Giacoletti, Heimer, and Yu develop empirical tests for discrimination that use high-frequency evaluations to address the problem of unobserved heterogeneity in a conventional benchmarking test. They approach to identifying discrimination requires two conditions: (1) the subject pool is time-invariant in a short time horizon and (2) there is high-frequency variation in the extent to which evaluators can rely on their subjective assessments. The researchers bring our approach to the residential mortgage market, using data on the near-universe of U.S. mortgage applications from 1994 to 2018. Monthly volume quotas reduce how much subjectivity loan officers apply to loans they process at the end of the month. As a result, the volume of new originations increases by 150% at the end of the month, while application volume and applicants' quality are constant within the month. Owing to within-month variation in loan officers' subjectivity, Giacoletti, Heimer, and Yu estimate that Black mortgage applicants have 3.5% to 5% lower approval rates, which explains at least half of the observed approval gap for Blacks. When they use this approach to evaluate policies, Giacoletti, Heimer, and Yu find that market concentration and FinTech lending have had no effect on lending discrimination, but that shadow banking has reduced discrimination presumably by having a larger presence in under-served communities.

This paper uses a new administrative data set of auto loans to test the predictions of multiple models of economic discrimination. Auto loans are often subject to discretionary markup, where a dealer increases the rate paid by the borrower in order to receive additional compensation from the lender. Lanning first shows this practice manifests in an average disparity between Black and White customers. He then uses attitudinal measures in conjunction with administrative data containing millions of records from multiple lenders to explicitly test the Becker model of discrimination by utilizing key percentiles in the distribution of prejudice attitudes. The researcher is also able to execute credible tests of simple models of statistical discrimination and search with discrimination. Ultimately Lanning finds results that are highly consistent with the predictions of a standard Becker model of discrimination, implying that prejudice has a large impact on racial gap in markup. Additionally, he finds that neither search nor statistical discrimination appear to meaningfully contribute to racial disparity observed in this market.

Annan constructs a census of the market for mobile banking in village Ghana and estimate that 1 out of every 5 mobile money transactions is overcharged relative to mandated rates. In an experiment, the researcher randomizes the matches between vendors and customers, finding strong evidence of "gender misconduct gap": female vendors are 40% more likely to commit such misconduct relative to male vendors. Misconduct is asymmetric: female customers are 89-96% relatively more likely to suffer misconduct, and while female vendors discriminate against customers of their gender, male vendors favor their gender. Differences in empowerment and beliefs about gender are relevant mechanisms.

Exploiting a 20-year sample of leveraged buyouts matched to French administrative data, Fang, Goldman, and Roulet document that, relative to a control group of firms, target firms experience after the buyout a reduction in pay gaps together with an increase in profitability. The wage difference between men and women reduces by 6.5%, that between managers and non-managers by 3.3% and that between senior (above 50 years old) and younger workers by 18.1%, relative to their respective means. Composition effects drive these results. Post-buyout, target firms separate from expensive employees in the high-pay categories (men, managers, older employees) and replace them with cheaper employees. At the same time, men and young employees who stay at the firm experience small pay increases. Together, the results are consistent with the notion that, in seeking to improve target firms' efficiency, private equity investors reduce wage inequalities inside target firms by cutting highly-paid employees' rents and fostering their separation from the firm.

Increasing homeownership has long been a major policy goal in the U.S. Kulkarni and Malmendier argue that two primary policy tools since the 1990s, (i) eased access to mortgage financing and (ii) targeting underserved neighborhoods, have increased racial segregation and hampered upward mobility for Black families. First, while mortgage policies were effective in increasing overall homeownership, only Black homeownership increased in geographically targeted neighborhoods, while white homeowners decreased in these neighborhoods. This result is strongest in cities (commuting zones) with improved access to mortgages, making it easier for white families to move to non-targeted neighborhoods. Second, easier access to mortgage financing predicts reduced upward mobility among low-income Black families. For low-income white families, the researchers estimate an adverse effect for those remaining in the targeted neighborhood but little or no adverse effect overall. Kulkarni and Malmendier show that changes to the neighborhood (locational effects) and not selection or direct effects of homeowning drive the negative impact on upward mobility. Local mechanisms include a reduction in education spending and an increase in crime. The researchers provide evidence for a main channel driving the decline in upward mobility: house prices in targeted neighborhoods decline, and the resulting decline in property tax revenue reduces education spending and lowers school quality in targeted neighborhoods. Further, stringent land-use housing restrictions perpetuate these segregation effects even when the Black children are adult by preventing them from moving to better neighborhoods.

Undurraga randomly assigned consumer loan requests (of random amount and term) to gender-balanced prospective borrowers who then randomly submitted them to a representative sample of loan-officers from Chilean banks, for whom Undurraga elicited their gender preferences prior to the experiment. He finds that loan requests submitted by women are 18.3% less likely to be approved, with most of the gender effect coming from gender-biased officers, particularly males, suggesting taste-based discrimination. Undurraga further tests for statistical discrimination through an information experiment where he randomly informed some officers about official statistics indicating that women have higher repayment rates than men. The researcher finds the information treatment was not only ineffective to reduce gender discrimination, but the gender-biased officers in the treatment-group discriminated more against women relative to their control counterparts, reinforcing the taste-based hypothesis.

Gertsberg, Mollerstrom, and Pagel study shareholder support for corporate board nominees in the context of the California gender quota, which was passed in 2018. Using hand-collected data for approximately 600 firms, they show that, prior to the quota, female nominees received greater shareholder support than their male counterparts. This is consistent with a pre-quota environment in which female board nominees were held to a higher standard than male nominees. Second, the researchers show that incumbent female directors in the post-quota environment receive greater support than incumbent men, while support for new (mandated) female nominees decreases to the level of support for new male nominees. This indicates that the quota led to a conversion in the bar for men and women to become board nominees, and that it did not lead to new female board nominees being of lower quality than male nominees. Gertsberg, Mollerstrom, and Pagel likewise challenge the notion that the negative stock price reaction to the quota reflects value destruction due to an insufficient supply of female directors. Instead, they provide evidence that dysfunctional board dynamics are driving the reaction, in the sense that stock prices reacted negatively to entrenched boards who failed to turn over the least supported directors when adjusting their boards to comply with the new law.