A three bedroom, two bath home with a fireplace, featuring a large backyard with a fence and a neighborhood pool - - the American dream.
That dream isn’t feasible for all. Earlier this year, an investigation into home loans showed that the algorithms being used systematically discriminate against qualified minority applicants. These algorithms are based on user data and are all profit driven. This occurs all while nonprofit corporations or public health agencies are handicapped by data privacy laws that hinder them from using similar data – think credit and financial information.
The United States data protection laws often permit the use of data for profit but are much more restrictive in socially beneficial use cases.
Taking a deeper look at the mortgage industry, a recent Northwestern University study found that racial disparities in the mortgage market really have not declined in the last four decades when it comes to discrimination in loan acceptance and interest rates. The study indicates that Black and Latino mortgage applicants are more likely to be declined than white applicants, and if they are approved, they are more likely to be offered high-cost loans.
A report by the Urban Institute showed that there was also a significant gap in homeownership rates between whites and blacks — 72% of white Americans owned homes compared to 42% of Black Americans, a jarring statistic.
In fact, the percentage is so jarring because since the Fair Housing Act was passed in 1968, there has been negative progress on home ownership rates. The gap between black and white homeowners has only widened with time and black Americans saw an even more dramatic drop in homeownership rates after the recession than white Americans.
While home lending decisions are officially made by loan officers at each institution, they are largely driven by software. Most lenders lean on the classic or traditional FICO score, which is an important note because the FICO score rewards ‘traditional credit,’ which can be detrimental to people of color because they have had less access to it than while Americans. Unlike other scoring methods, it doesn’t consider things like paying bills on time but will penalize you for making any type of late payment on a bill.
The emergence of digital mortgage platforms, which automate the process of seeking loan approval, has also raised the question of whether they reduce discrimination or add to it. That question is sticky. While technology and scoring take color out of the loan decision-making process, it doesn’t mean that it doesn’t have flaws. Briefly put, what data sets you put into the technology will result in what the algorithm decides, so if you are putting in biased data, then you are going to get biased data on the backend.
In the United States, homeownership accounts for as much as two-thirds of the wealth of a typical household. The mortgage approval process, therefore, represents a large potential stumbling block on the road to wealth and stability. The mortgage and lending industry should be keeping a close eye and adjusting their algorithms frequently to account for any unforeseen bias, and loan officers should also ensure they are putting in the effort to call out flaws.
While the algorithms and loan officers have the final say if you are approved or denied, there are ways to protect yourself from discrimination:
1.) Get familiar with the lending process
Education = Empowerment- before you begin the homebuying process take some time and learn more about the process. Understand how your credit is accessed, research what the national average is for fees, interest rates, and loan terms for someone in a similar financial situation as yours.
2.) Have your documentation ready
3.) Know your credit score and what it means
Mortgage lenders rely mostly on your credit report so get familiar with what your score is and if it needs improvement prior to submitting your application.
4.) Shop Around
It’s important to do a little shopping- don’t forget to check out local community banks and credit unions as they typical offer much more flexible lending standards and greater access to different credit options.
Here at Extract, we’ve developed machine learning capabilities to better capture discrete data fields documents – then redact, classify, index, or extract that data.
If you’d like to learn more about how we capture data, and continuously improve our systems, please reach out today. Or if you would like to learn more about how Extract focus on fairness and best practices in AI and Machine Learning, check out this blog.
Source:
About the Author: Taylor Genter
Taylor is a Marketing Manager at Extract with experience in data analytics, graphic design, and both digital and social media marketing. She earned her Bachelor of Business Administration degree in Marketing at the University of Wisconsin- Whitewater. Taylor enjoys analyzing people’s behaviors and attitudes to find out what motivates them, and then curating better ways to communicate with them.