Figure 1: Relationship Between Credit Score and Total Bankcard Limit
by Gender
This figure describes the relationship between total bankcard limit and credit score (bin). Credit score is the Equifax Risk Score. The dashed line displays the average values for female cardholders and the solid line shows the results for male cardholders. For the entire credit score distribution (<350 through >=800), the solid line is above the dashed line. For very low credit scores, the average total bankcard limits are around $14,000 to $16,500, then they decrease and the values for both men and women are lowest among those with a credit score in the 500 – 549 range. After that, total bankcard limits increase. Source: Authors’ calculations using Home Mortgage Disclosure Act (HMDA) data, Black Knight McDash loan servicing data, and Equifax Credit Risk Insight Servicing data. Credit score is the Equifax Risk Score. Demographic information comes from the HMDA data.
Figure 2: Relationship Between Income and Total Bankcard Limit by
Gender
This figure describes the relationship between total bankcard limit and income (bin). The dashed line displays the average values for female cardholders and the solid line shows the results for male cardholders. For incomes below $100,000, the dashed line is above the solid line, meaning that women have greater values for average bankcard limit than men do. For income ranges above $100,000 the lines switch and the solid line (Males) is above the dashed line (Females) for the majority of income ranges. Both lines slope upwards, as the total bankcard limit increases, on average, with higher incomes for both men and women. Source: Authors’ calculations using Home Mortgage Disclosure Act (HMDA) data, Black Knight McDash loan servicing data, and Equifax Credit Risk Insight Servicing data. Credit score is the Equifax Risk Score. Demographic information comes from the HMDA data.
Figure 3: Bankcard Differences over Time by Gender
Panel A. Total Bankcard Credit Limit
Panel B. Unconditional Differences in Total Bankcard Limit by Year
and Limit Decile
Notes: Authors’ calculations using Home Mortgage Disclosure Act
(HMDA) data, ICE McDash loan servicing data, and Equifax Credit Risk
Insight Servicing data. Demographic information comes from the HMDA
data.
This two-panel figure illustrates the total bankcard credit limit (Panel A), and the unconditional differences in total bankcard limit (Panel B) over time from January 2006 through December 2016. In Panel A, the solid dark blue line plots the values for female bankcard holders and the dashed red line plots the values for males. The solid dark blue line is above the dashed red line for the entire series, meaning that men had higher limits, on average, than women, from 2006 through 2016. Panel B is a heatmap displaying the gender difference in credit limits by year and by limit decile. Positive values are shaded green, and the color becomes darker with larger values. Negative values (meaning that women have an advantage in the levels of their endowments relative to men) are shaded blue. Values around 0 dollars (representing similar values of the endowment effect among men and women by decile and year) are white. The heatmap predominantly displays various shades of green, indicating that males generally have higher credit limits than females across most years and deciles. The darkest green areas appear in the highest deciles (80-90) across most years, with particularly strong male advantages in 2006-2008 and 2013-2014, where differences exceed $5,000. The gender advantage gradually decreases in the middle and lower deciles, showing lighter green shades with differences between $375 and $2,750. The heatmap reveals that gender differences in credit limits vary not only by limit decile but also change over time, with the general male advantage persisting but showing some reduction in specific segments in later years. Source: Authors' calculations using Home Mortgage Disclosure Act (HMDA) data, ICE McDash loan servicing data, and Equifax Credit Risk Insight Servicing data. Demographic information comes from the HMDA data.
Figure 4: Total Bankcard Credit Limit by Year of Birth
Panel A. Difference by Birth Year
Panel B. Difference by Birth Year, by Limit Decile
Notes: Authors’ calculations using Home Mortgage Disclosure Act
(HMDA) data, ICE McDash loan servicing data, and Equifax Credit Risk
Insight Servicing data. Credit Score is the Equifax Risk Score.
Demographic information comes from the HMDA data. Birth year is
winsorized at 1985 and 1930 to avoid outliers.
This figure consists of two panels showing credit limit differences by year of birth and gender. Panel A is a line chart showing the difference in total bankcard credit limit by age. Two lines are plotted: a solid blue line representing females and a dashed red line representing males. Both genders show highest credit limits for those born between 1935-1950, with limits steadily declining for younger cohorts born after 1950. Credit limits converge to around $12,000 for those born in 1985. For all birth years, males have higher credit limits than females. Panel B is a heatmap displaying the gender difference in credit limits by year and by limit decile. Positive values are shaded green, and the color becomes darker with larger values. Negative values (meaning that women have an advantage in the levels of their endowments relative to men) are shaded blue. Values around 0 dollars (representing similar values of the endowment effect among men and women by decile and year) are white. Most of the heatmap displays various shades of green, indicating higher credit limits for males across most birth years and deciles. The brightest green areas appear in the upper deciles (70-90) for birth years 1930-1945, showing the largest gender differences. Small areas of blue appear in the earliest birth years (1930-1935) in the lower deciles (10-20), indicating instances where females have higher limits than males in these segments. Source: Authors' calculations using Home Mortgage Disclosure Act (HMDA) data, ICE McDash loan servicing data, and Equifax Credit Risk Insight Servicing data. Credit Score is the Equifax Risk Score. Demographic information comes from the HMDA data. Birth year is winsorized at 1965 and 1990 to avoid outliers.
Figure 5: Total Bankcard Credit Limit by Age
Panel A. Difference by Age
Panel B. Difference by Age, by Limit Decile
Notes: Authors’ calculations using Home Mortgage Disclosure Act
(HMDA) data, ICE McDash loan servicing data, and Equifax Credit Risk
Insight Servicing data. Credit Score is the Equifax Risk Score.
Demographic information comes from the HMDA data. Age is winsorized at
25 and 80 to avoid outliers.
This figure consists of two panels showing credit limit differences by age and gender. Panel A is a line chart showing the difference in total bankcard credit limit by age. Two lines are plotted: a solid blue line representing females and a dashed red line representing males. Both genders show credit limits starting around $10,000 at age 25, then steadily increasing with age. Credit limits peak around age 70-75, with males reaching approximately $37,000 and females reaching about $33,000. After age 75, both lines decline slightly. For most of the age range, males have higher credit limits than females, with the gap widening around age 35. Panel B is a heatmap displaying the gender difference in credit limits by age and by limit decile. Positive values are shaded green, and the color becomes darker with larger values. Negative values (meaning that women have an advantage in the levels of their endowments relative to men) are shaded blue. Values around 0 dollars (representing similar values of the endowment effect among men and women by decile and year) are white. Most of the heatmap shows green shades, indicating higher credit limits for males across most age groups and deciles. The intensity of green increases in the upper age ranges, particularly between ages 60-75 and the two highest deciles. Small areas of blue appear in the oldest age categories (75-80) in the lower deciles, indicating some instances where females have higher limits than males. Notes: Authors' calculations using Home Mortgage Disclosure Act (HMDA) data, ICE McDash loan servicing data, and Equifax Credit Risk Insight Servicing and McDash Credit Score is the Equifax Risk Score. Demographic information comes from the HMDA data. Age is winsorized at 25 and 80 to avoid outliers.
Figure 6: Average Marginal Effects
Panel A. AMEs Using Cross-Sectional Data
Panel B. AMEs Using Panel Data
Notes: Authors’ calculations using Home Mortgage Disclosure Act
(HMDA) data, ICE McDash loan servicing data, and Equifax Credit Risk
Insight Servicing data. All Credit Score is the Equifax Risk Score.
Income in the cross-sectional data sample is the HMDA income, reported
at the time of mortgage application. Income in the panel data sample is
an income estimate from the CRISM data. Demographic information comes
from the HMDA data.
This figure contains two horizontal bar charts showing Average Marginal Effects (AMEs) across different model specifications. All bars in both panels are navy blue. Panel A displays average marginal effects in dollars using the cross-sectional data. Eight model specifications are listed on the y-axis: Base Model, Full Model, No Age, No Cards, No Credit Score, No Income, No Race, and No State. Most specifications show negative marginal effects of approximately -$1,500. "No Cards" shows a slight positive effect of approximately $200. "No Income" shows the most negative effect at approximately -$3,200 dollars. Panel B displays average marginal effects using panel data. The same eight model specifications appear on the y-axis. All specifications show negative effects. Most models show effects of approximately -$1,800. "No Income" again shows the most negative effect at approximately -$3,200 dollars. "No Cards" shows the least negative effect at approximately -$200. Source: Authors' calculations using Home Mortgage Disclosure Act (HMDA) data, ICE McDash loan servicing data, and Equifax Credit Risk Insight Servicing data. All Credit Score is the Equifax Risk Score. Income in the cross-sectional data sample is the HMDA income, reported at the time of mortgage application. Income in the panel data sample is an income estimate from the CRISM data. Demographic information comes from the HMDA data.
Figure 7: Average Marginal Effects by County Employment Status
Panel A: Employment Status
Panel B: Mass Layoff Status
Notes: Authors’ calculations using Home Mortgage Disclosure Act
(HMDA) data, ICE McDash loan servicing data, and Equifax Credit Risk
Insight Servicing data. Demographic information comes from the HMDA
data. Results for mass layoff analyses use data from 2006 to 2011, while
results for the baseline and unemployment rate analyses use data from
2006 to 2016.
This two-panel figure the average marginal effects (of the female dummy variable) by employment status (Panel A) and mass layoff status (Panel B). The coefficient plot in Panel A plots the AME for the following five county-level employment categories: "Baseline," "No Mass Layoff," "Mass Layoff," "Low Unemployment," and "High Unemployment." Each category displays a dark blue point estimate with vertical blue lines representing confidence intervals. All point estimates except for "Mass Layoff" are below zero, with values ranging from approximately -$1,500 to -$250. The "Mass Layoff" category has a point estimate near -$250 dollars but shows the widest confidence interval, illustrating this negative AME value is not significant. The coefficient plot in Panel B displays dark blue point estimates with vertical blue lines representing conference intervals for the following categories: "Baseline," "Mass Layoff: Overall," "Mass Layoff: Men>Women," and "Mass Layoff: Women>Men." The "Baseline" point estimate is approximately -$1500 dollars (signifying male borrowers have larger average credit limits) with a narrow confidence interval. All mass layoff categories have point estimates with wide confidence intervals extending from about -$1,500 to $1,000 dollars. Source: Authors' calculations using Home Mortgage Disclosure Act (HMDA) data, ICE McDash loan servicing data, and Equifax Credit Risk Insight Servicing data. Demographic information comes from the HMDA data. Results for mass layoff analyses use data from 2006 to 2011, while results for the baseline and unemployment rate analyses use data from 2006 to 2016.
Figure 8: Unconditional Quantile Regression Results: Gender
Differences ($) Across Time by Limit Decile
This figure shows a heatmap of the annual gender differences from 2006 through 2016 for the total bankcard limit for each decile of the limit variable. The y-axis shows the Bankcard Limit Decile ranging from 10^(th) percentile to the 90^(th) percentile, and the x-axis shows year. These results are derived from the decompositions described in section 5.3 and are illustrated by a 9x11 box grid, with the first decile (10th percentile) on the first row and the 90^(th) decile on the top row. Positive values are shaded green, and the color becomes darker with larger values. Negative values (meaning that women have an advantage in the levels of their endowments relative to men) are shaded blue. Values around 0 dollars (representing similar values of the endowment effect among men and women by decile and year) are white. The two highest limit deciles consistently show the largest positive gender differences (green), indicating higher limits for males. Beginning in 2011, the middle deciles (30^(th) - 60^(th)) show a pattern of negative gender differences (blue), indicating higher limits for female borrowers. Source: Authors' calculations using Home Mortgage Disclosure Act (HMDA) data, ICE McDash loan servicing data, and Equifax Credit Risk Insight Servicing data. Demographic information comes from the HMDA data. Reported Z-axis values are the midpoint of each bin.
Figure 9: Relationship Between Credit Score and Bankcard Mail Offers
Received by Gender
This figure describes the relationship between Credit Score (bin) and total promotional bankcard mail offers. The dashed blue line displays the average number of offers for women and the solid red line shows the results for men. For Credit Scores above 600, the solid red line is above the dashed blue line, meaning that men receive more promotional bankcard mailers than women do. Both lines slope upwards until they reach a peak among adults with credit scores between 650 and 750, then both lines trend downwards. Source: Authors’ calculations using Mintel/TransUnion data with sample weights provided by the vendor. Demographic information comes from the Mintel data. Includes years 2009 – 2017. Credit score is transformed version of VantageScore 2.0.
Figure A1: Bankcard Differences over Time by Gender
Panel A. Number of Bankcard Accounts
Panel B. Average Bankcard Credit Limit
Notes: Authors’ calculations using Home Mortgage Disclosure Act
(HMDA) data, ICE McDash loan servicing data, and Equifax Credit Risk
Insight Servicing data. Demographic information comes from the HMDA
data.
This two-panel figure illustrates the number of bankcard accounts (Panel A) and the average bankcard credit limit (Panel B) by gender over time from January 2006 through December 2017 (at monthly increments). The solid dark blue line plots the values for female bankcard holders and the dashed red line plots the values for males. In Panel A, the solid dark blue line is above the dashed red line for the entire series, meaning that men had more bankcard accounts, on average, than women, from 2006 through 2017. In Panel B, the dashed red line is above the solid dark blue line for the entire series, as men have larger average bankcard credit limits, on average, than women. Source: Authors’ calculations using Home Mortgage Disclosure Act (HMDA) data, Black Knight McDash loan servicing data, and Equifax Credit Risk Insight Servicing data. Credit score is the Equifax Risk Score. Demographic information comes from the HMDA data.
Figure A2: Bankcard Differences by Credit Score Category
Panel A. (Male-Female) Difference in the Number of Bankcard
Accounts
Panel B. (Male-Female) Difference in Total Bankcard Credit Limit
Panel C. (Male-Female) Difference in Average Bankcard Credit
Limit
Notes: Authors’ calculations using Home Mortgage Disclosure Act
(HMDA) data, ICE McDash loan servicing data, and Equifax Credit Risk
Insight Servicing data. Credit Score is the Equifax Risk Score.
Demographic information comes from the HMDA data.
This three-panel figure illustrates the estimated (male – female) difference in the number of bankcard accounts (Panel A), the total bankcard credit limit (Panel B), and the average bankcard credit limit (Panel C) from January 2006 through December 2017 (at monthly increments). Therefore, positive values identify time periods (months) where the estimates for men are larger than the estimates for women. Conversely, any negative values mean that the value for female bankcard holders is greater than the value for male bankcard holders. The lines plot the gender differences among individuals in five Equifax Risk Score cohorts: <= 579 (blue), 580-669 (red), 670-739 (green), 740-799 (yellow), and >=800 (gray). Source: Authors’ calculations using Home Mortgage Disclosure Act (HMDA) data, Black Knight McDash loan servicing data, and Equifax Credit Risk Insight Servicing data. Credit score is the Equifax Risk Score. Demographic information comes from the HMDA data.
Figure A3: Age by Birth Year Cohort Analysis
Panel A. Summary Stats
Panel B. Regression Analysis
Notes: Authors’ calculations using Home Mortgage Disclosure Act
(HMDA) data, ICE McDash loan servicing data, and Equifax Credit Risk
Insight Servicing data. Credit Score is the Equifax Risk Score.
Demographic information comes from the HMDA data.
This figure contains two panels analyzing credit limits across age groups by birth cohorts. Panel A (Summary Stats) shows two scatter plots displaying dollars on the y-axis (ranging from $2,000 to $8,000) and age on the x-axis (68, 69, 70). The left plot represents the 1938-1940 birth cohort, showing three data points around the $6,000 level. The right plot represents the 1945-1947 birth cohort, displaying three data points around the $4,000 level. Panel B (Regression Analysis) displays regression results with confidence intervals across ages 68, 69, and 70 on the x-axis and dollars (ranging from -$8,000 to -$2,000) on the y-axis. Two sets of data points with error bars are shown: the 1938-1940 cohort (represented by blue dots) positioned around the -$6,000 the positive dollar range, and the 1945-1947 cohort (represented by red dots) positioned around -$2,000 to -$4,000. Source: Authors' calculations using Home Mortgage Disclosure Act (HMDA) data, ICE McDash loan servicing data, and Equifax Credit Risk Insight Servicing data. Credit Score is the Equifax Risk Score. Demographic information comes from the HMDA data.
Figure A4: Average Marginal Effects for Credit Score and Income, by Category
Panel A: Credit Score Category
Panel B: Income Category
Notes: Authors’ calculations using Home Mortgage Disclosure Act
(HMDA) data, ICE McDash loan servicing data, and Equifax Credit Risk
Insight Servicing data. Credit Score is the Equifax Risk Score. Income
is the HMDA income, reported at the time of mortgage application.
Demographic information comes from the HMDA data. Bars represent 95%
confidence intervals.
This figure plots the estimated average marginal effects for the female dummy variable at each credit score category (Panel A) and income category (Panel B) on total bankcard limits. These are estimated while holding the values of the other covariates at their mean values. The blue dots in both panels plot the regression coefficients, Γ and Θ, from Equation 2 in Section 4.1, along with the 95 percent confidence intervals. Source: Authors’ calculations using Home Mortgage Disclosure Act (HMDA) data, Black Knight McDash loan servicing data, and Equifax Credit Risk Insight Servicing data. Credit score is the Equifax Risk Score. Demographic information comes from the HMDA data.
Figure A5: Robustness Checks: Average Marginal Effects
This is a horizontal bar chart showing average marginal effects in dollars across seven different model specifications. The x-axis shows marginal effect values in dollars ranging from -1,500 to 0, with tick marks at -1,500, -1,000, -500, and 0. The y-axis lists seven model specifications from top to bottom: "Full Model," "Financial Distress," "Financial Distress Interaction," "Restricted To No Financial Distress," "Omit 2006," "No Time Effects," and "No Time by State Effects." Each specification is represented by a dark blue horizontal bar extending from approximately -$1,200 to -$1,500. The bars are relatively uniform in length, all showing negative marginal effects of similar magnitude across the different model specifications, indicating consistency in results across robustness checks. Source: Authors' calculations using Home Mortgage Disclosure Act (HMDA) data, ICE McDash loan servicing data, and Equifax Credit Risk Insight Servicing data. Credit score is the Equifax Risk Score. Demographic information comes from the HMDA data.
Figure A6: Average Marginal Effects by Census Region
This figure is a scatter plot showing average marginal effects (in dollars) across the four U.S. census regions. The y-axis displays values ranging from -$1,800 to -$800. The x-axis shows four census regions: Northeast, Midwest, South, and West. Each region is represented by a data point with vertical confidence interval bars. All marginal effects are negative across all regions, with values ranging between -$1,600 and -$1,100 dollars. The Northeast shows the lowest (most negative) marginal effect, and the Midwest shows the highest (least negative) marginal effect. The South and West regions show similar marginal effects. All four regions display confidence intervals of varying lengths. Source: Authors' calculations using Home Mortgage Disclosure Act (HMDA) data, ICE McDash loan servicing data, and Equifax Credit Risk Insight Servicing data. Credit score is the Equifax Risk Score. Demographic information comes from the HMDA data.
Figure A7: Gender Differences Across Time by Decile, as a Percentage of
the Male Total Bankcard Limit
This figure shows a heatmap of the annual gender differences from 2006 through 2016 for the total bankcard limit for each decile of the limit variable. The difference is measured as a percentage of the total bankcard limit. These results are derived from the decompositions described in section 5.3 and are illustrated by a 9x11 box grid, with the first decile (10th percentile) on the first row and the 90th decile on the top row. Positive values are shaded green, and the color becomes darker with larger values. Negative values (meaning women have larger total bankcard limits than men, on average) are shaded blue. Values around 0 percent (representing similar values between men and women by decile and year) are white. The top two rows of the Panel A are solidly green, meaning that men have larger bankcard limits than women among those above the 80th percentile. Source: Authors’ calculations using Home Mortgage Disclosure Act (HMDA) data, Black Knight McDash loan servicing data, and Equifax Credit Risk Insight Servicing data. Demographic information comes from the HMDA data. Reported Z-axis values are the midpoint of each bin.
Figure A8: Heat Maps of the Endowment Effect Across Time by Decile
Panel A. Difference due to the Endowment
Effect, Measured in Dollars ($)
Panel B. Difference, As a Percentage of the Male Total Bankcard
Limit
Notes: Authors’ calculations using Home Mortgage Disclosure Act
(HMDA) data, ICE McDash loan servicing data, and Equifax Credit Risk
Insight Servicing data. Demographic information comes from the HMDA
data. Reported \(Z\)-axis values are
the midpoint of each bin.
This two-panel figure shows heatmaps depicting the share of the gender difference due to the endowment effect. Annual estimates for each decile of the total bankcard limit variable from 2006 through 2016 are displayed. Both the magnitude of the endowment effect, measured in dollars (Panel A) and the difference as a percentage of the average male total bankcard limit (Panel B), are displayed. These results are derived from the decompositions described in section 7 and are illustrated by a 9x11 box grid, with the first decile (10th percentile) on the first row and the 90th decile on the top row. Positive values are shaded green, and the color becomes darker with larger values. Negative values (meaning that women have an advantage in the levels of their endowments relative to men) are shaded blue. Values around 0 dollars (representing similar values of the endowment effect among men and women by decile and year) are white. From 2006 through 2009 the endowment effect is positive for all deciles before decreasing in 2010 and becoming negative for all deciles in 2011. Source: Authors’ calculations using Home Mortgage Disclosure Act (HMDA) data, Black Knight McDash loan servicing data, and Equifax Credit Risk Insight Servicing data. Demographic information comes from the HMDA data. Reported Z-axis values are the midpoint of each bin.
Figure A9: Heat Maps of the Coefficient Effect Across Time by Decile
Panel A. Difference due to the Coefficient
Effect, Measured in Dollars ($)
Panel B. Difference, As a Percentage of the Male Total Bankcard
Limit
Notes: Authors’ calculations using Home Mortgage Disclosure Act
(HMDA) data, ICE McDash loan servicing data, and Equifax Credit Risk
Insight Servicing data. Demographic information comes from the HMDA
data. Reported \(Z\)-axis values are
the midpoint of each bin.
This two-panel figure shows heatmaps depicting the share of the gender difference due to the coefficient effect. Annual estimates for each decile of the total bankcard limit variable from 2006 through 2016 are displayed. Both the magnitude of the coefficient effect, measured in dollars (Panel A) and the difference as a percentage of the average male total bankcard limit (Panel B), are displayed. These results are derived from the decompositions described in section 5.2 and are illustrated by a 9x11 box grid, with the first decile (10th percentile) on the first row and the 90th decile on the top row. Positive values are shaded green, and the color becomes darker with larger values. Negative values (meaning that women receive higher returns on their observed characteristics than men) are shaded blue. Values around 0 dollars (representing similar values of the coefficient effect among men and women by decile and year) are white. For almost all years in the sample, the coefficient effect is negative for the deciles below the median of the total limit distribution. Source: Authors’ calculations using Home Mortgage Disclosure Act (HMDA) data, Black Knight McDash loan servicing data, and Equifax Credit Risk Insight Servicing data. Demographic information comes from the HMDA data. Reported Z-axis values are the midpoint of each bin.
Figure A10: Example KOB Decomposition
Notes: This example produces a KOB decomposition where the interaction and endowment effects are negative, but the coefficient effect is positive. To generate this result, female endowments are greater than male endowments, while men have greater average values for the Y variable than women. If the relationship for men is given by Y_(M) = α_(M) + β_(M) X_(M) and the relationship for women is given by Y_(W) = α_(W) + β_(W) X_(W), then the coefficient effect is (β_(M) - β_(W))X_(M), the endowment effect is given by β_(W)(X_(M) - X_(W)), and the interaction effect is given by (β_(M) - β_(W))(X_(M) - X_(W)). For a canonical example in this format, see Jones and Kelley (1984). This diagram schematically displays the decomposition of group difference in mean predicted outcome from the perspective of Women, when Men have been considered the reference group. In this abstract diagram endowments are on the X-axis, and Credit Limit is on the Y-axis. Two solid, upward (positive) sloping lines represent the relationship between endowment levels and total credit limit. However the line for men has a steeper slope, meaning that as men accrue higher levels of endowments, they receive greater gains to their total credit limit than women. A dashed line originating from the same point on the y-axis as the men’s outcome, but with a slope identical to women’s, illustrates the counterfactual. This example produces a KOB decomposition where the interaction and endowment effects are negative, but the coefficient effect is positive. To generate this result, female endowments are greater than male endowments, while men have greater average values for the Y variable than women.
Figure A11: KOB Decomposition by Life Cycle Period
This figure is a grouped bar chart showing KOB decomposition results across five age categories: <=34, 35-44, 45-54, 55-64, and 65+. The y-axis displays dollars ranging from $0 to $4,000. The four decomposition components are shown for each age category, represented by different colored bars: difference (solid blue), endowment effect (solid pink), coefficient effect (dotted green), and interaction effect (solid light orange). For all age groups, the blue bars (raw difference) are tallest, as they represent the sum of the three other components. The interaction effect is the smallest effect for all age groups and therefore it’s bars are shortest. The middle age groups (35-44, 45-54, 55-64) show similar patterns, with the most dramatic changes appearing in the oldest (65+) age group. The difference (blue bars) increases with age, starting at approximately $1,200 for the <=34 age group and reaching about $4,100 for the 65+ age group. Source: Authors' calculations using Home Mortgage Disclosure Act (HMDA) data, ICE McDash loan servicing data, and Equifax Credit Risk Insight Servicing data. Credit Score is the Equifax Risk Score. Demographic information comes from the HMDA data.