Bias isn’t the only problem with credit scores—and no, AI can’t help

Link: https://www.technologyreview.com/2021/06/17/1026519/racial-bias-noisy-data-credit-scores-mortgage-loans-fairness-machine-learning/

Excerpt:

But in the biggest ever study of real-world mortgage data, economists Laura Blattner at Stanford University and Scott Nelson at the University of Chicago show that differences in mortgage approval between minority and majority groups is not just down to bias, but to the fact that minority and low-income groups have less data in their credit histories.

This means that when this data is used to calculate a credit score and this credit score used to make a prediction on loan default, then that prediction will be less precise. It is this lack of precision that leads to inequality, not just bias.

…..

But Blattner and Nelson show that adjusting for bias had no effect. They found that a minority applicant’s score of 620 was indeed a poor proxy for her creditworthiness but that this was because the error could go both ways: a 620 might be 625, or it might be 615.

Author(s): Will Douglas Heaven

Publication Date: 17 June 2021

Publication Site: MIT Tech Review

Which Groups Are Still Dying of Covid in the U.S.?

Link: https://www.nytimes.com/interactive/2021/06/10/us/covid-death-patterns.html

Graphics:

image

Excerpt:

“Previously, at the start of the pandemic, we were seeing people who were over the age of 60, who have numerous comorbidities,” said Dr. Krutika Kuppalli, an infectious disease expert at the Medical University of South Carolina. “I’m not seeing that as much anymore.” Instead, she said, hospitalizations have lately been skewing toward “people who are younger, people who have not been vaccinated.”

More than 80 percent of those 65 and older have received at least one dose of a Covid-19 vaccine, compared with about half of those aged 25 to 64 who have received one dose. Data collected by the C.D.C. on so-called breakthrough infections — those that happen to vaccinated people — suggest an exceedingly low rate of death among people who had received a Covid-19 vaccine.

Author(s): Denise Lu

Publication Date: 10 June 2021

Publication Site: New York Times

Update to Special Reports on Traffic Safety During the COVID-19 Public Health Emergency: Fourth Quarter Data

Link: https://www.nhtsa.gov/sites/nhtsa.gov/files/2021-06/Update_Traffic%20Safety%20During%20COVID-19_4thQtr-060121-web.pdf

Graphic:

Excerpt:

The NEMSIS data include metrics on crash severity. For people treated at the scenes of motor vehicle crashes, EMS professionals use an injury scoring system called the Revised Trauma Score (RTS) to determine the level of care needed to save the lives of the injured. Under
RTS, patients who present with a probability of survival of 36.1% or less are considered severely injured and are often transported to Level 1 or Level 2 trauma centers that provide higher levels of critical care to the most severely injured. Figure 4 shows the percentage of
patients in crashes whose probability of survival was in this range for 2019 and 2020. Beginning in Week 12 of 2020, the percentage of those injured with a probability of survival of 36.1% or less never dropped below 1%, suggesting an increase in the severity of crashes.

Publication Date: June 2021

Publication Site: NHTSA

Early Estimates of Motor Vehicle Traffic Fatalities and Fatality Rate by Sub-Categories in 2020

Link: https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/813118

Graphic:

Excerpt:

Total fatality rate per 100 million VMT [vehicle miles traveled] is broken down by roadway function class: rural versus urban interstate, arterial, local/collector/street. The results shown in Figure 2 indicate that the increased trend of the total fatality rate per 100 million VMT from March to December 2020, was mainly driven by the fatality rate per 100 million VMT on the rural local/collector/street, rural and urban arterial roadways.

Publication Date: May 2021

Publication Site: NHTSA

2020 Fatality Data Show Increased Traffic Fatalities During Pandemic

Link: https://www.nhtsa.gov/press-releases/2020-fatality-data-show-increased-traffic-fatalities-during-pandemic

Excerpt:

Preliminary finding show that traffic fatalities rose in most major categories over 2019: 

Passenger vehicle occupants (23,395, up 5%)

Pedestrians (6,205, flat from 2019)

Motorcyclists (5,015, up 9%)

Pedalcyclists (people on bikes) (846, up 5%)

Crash factors and demographics reviewed by NHTSA that showed the largest increases in 2020 as compared to 2019 included: 

non-Hispanic Black people (up 23%); 

occupant ejection (up 20%);

unrestrained occupants of passenger vehicles (up 15%);

on urban interstates (up 15%);

on urban local/collector roads (up 12%);

in speeding-related crashes (up 11%);

on rural local/collector roads (up 11%); 

during nighttime (up 11%); 

during the weekend (up 9%); 

in rollover crashes (up 9%); 

in single-vehicle crashes (up 9%) and; 

in police-reported alcohol involvement crashes (up 9%).

There are a few categories that are projected to have decreases in fatalities in 2020. Fatalities in crashes involving a large truck (commercial or non-commercial use) are projected to decline marginally (down 2%).  Fatalities among older persons (65+ years of age) are projected to decline by about 9 percent.

Publication Date: 3 June 2021

Publication Site: National Highway Traffic Safety Administration

Which US vaccine plans actually helped hard-hit communities?

Link: https://www.technologyreview.com/2021/06/07/1025824/us-vaccine-equity-success-story/

Excerpt:

One success story took place in Philadelphia, thanks to an effective collaboration between two health systems and Black community leaders. Recognizing that the largely online signup process was hard for older people or those without internet access, Penn Medicine and Mercy Catholic Medical Center created a text-message-based signup system as well as a 24/7 interactive voice recording option that could be used from a land line, with doctors answering patients’ questions before appointments. Working with community leaders, the program held its first clinic at a church and vaccinated 550 people.

….

In Alabama, for example, National Guard mobile vaccination units were set up with the ultra-cold freezers needed to transport and store mRNA-based covid-19 vaccines. “Why not, when this particular push is over, leave those freezer units with the federally qualified health centers that are already in those communities?” McClure says. “You’re starting to build the infrastructure for being able to deliver vaccination on a consistent basis.”

Author(s): Mia Sato

Publication Date: 7 June 2021

Publication Site: MIT Technology Review

Latest Data on COVID-19 Vaccinations Race/Ethnicity

Graphic:

Excerpt:

This week’s (May 10 to May 17, 2021) pace of vaccination remained similar to last week across racial/ethnic groups. Across reporting states, vaccination rates increased by 1.3 percentage points for White people, from 40.3% to 41.6%, and by 1.2 percentage points for Black people, from 26.6% to 27.8%, maintaining the gap in rates between these groups (Figure 4). The rate for Hispanic people increased by 1.6 percentage points from 28.8% to 30.4%, while the rate for Asian people increased by 1.9 percentage points, from 50.2% to 52.1%.

Author(s): Nambi Ndugga, Olivia Pham , Latoya Hill, Samantha Artiga, Raisa Alam , Noah Parker

Publication Date: 19 May 2021

Publication Site: Kaiser Family Foundation

Over the past century, African-American life expectancy and education levels have soared

Link: https://www.economist.com/graphic-detail/2021/05/20/over-the-past-century-african-american-life-expectancy-and-education-levels-have-soared

Graphic:

Excerpt:

AT THE turn of the twentieth century, a newborn white American could expect to live for around 48 years. That was 15 years longer than a newborn African-American could expect. Improvements in hygiene, medicine and other public-health measures led those numbers to rise dramatically. By mid-century, life expectancy for African-Americans had nearly doubled, to 61 years, while for white Americans it rose to 69. By 2017 the gap had narrowed further, to three and a half years: 75.3 for African-Americans, 78.8 for whites. But Hispanic Americans outlive them both, to an average of 81.8 years. In other words, both races have progressed significantly, but gaps remain. This same pattern exists across a number of metrics.

The most disturbing aspect of this pattern is not just the enduring gap in outcomes between black and white Americans, though it has narrowed markedly. It is that, as the work of Anne Case and Angus Deaton, both economists at Princeton, has shown, life expectancy fell for all demographic groups of Americans between 2014 and 2017 for the first time since 1993. The rise in mortality rates has been especially stark for whites without college degrees, owing to what they call “deaths of despair”: drug overdoses, suicide and diseases caused by heavy drinking.

Publication Date: 20 May 2021

Publication Site: The Economist

Banks resist racial-equity audits, but they shouldn’t

Link: https://www.startribune.com/banks-resist-racial-equity-audits-but-they-shouldn-t/600049751/

Excerpt:

By lunchtime Tuesday we should know whether the Wells Fargo & Co. shareholders adopted a proposal to have the company conduct a racial-equity audit, an idea championed by a pension fund shareholder affiliated with the Service Employees International Union.

Wells Fargo and other big banks have recommended shareholders vote down these racial-equity audit proposals, a feature of this year’s annual shareholder meeting season.

The banks are likely to have the votes, but hopefully they don’t put the whole idea into a file and forget about it.

Author(s): Lee Schafer

Publication Date: 24 April 2021

Publication Site: Star Tribune

Actuaries Find Race and Ethnicity Disclosure Gap

Link: https://www.thinkadvisor.com/2021/04/06/actuaries-find-race-and-ethnicity-disclosure-gap/

Excerpt:

About 12% of U.S. residents are Black, and 15% are Hispanic or Latino.

In the insurance industry, 12% of employees are Black and 12% are Hispanic or Latino.

Just 4.4% of new SOA members are Black, and 3.7% of new members are Hispanic or Latino.

Author(s): Allison Bell

Publication Date: 6 April 2021

Publication Site: Think Advisor

Provisional Mortality Data — United States, 2020

Link: https://www.cdc.gov/mmwr/volumes/70/wr/mm7014e1.htm

Graphic:

Excerpt:

During January–December 2020, the estimated 2020 age-adjusted death rate increased for the first time since 2017, with an increase of 15.9% compared with 2019, from 715.2 to 828.7 deaths per 100,000 population. COVID-19 was the underlying or a contributing cause of 377,883 deaths (91.5 deaths per 100,000). COVID-19 death rates were highest among males, older adults, and AI/AN and Hispanic persons. The highest numbers of overall deaths and COVID-19 deaths occurred during April and December. COVID-19 was the third leading underlying cause of death in 2020, replacing suicide as one of the top 10 leading causes of death (6).

The findings in this report are subject to at least four limitations. First, data are provisional, and numbers and rates might change as additional information is received. Second, timeliness of death certificate submission can vary by jurisdiction. As a result, the national distribution of deaths might be affected by the distribution of deaths from jurisdictions reporting later, which might differ from those in the United States overall. Third, certain categories of race (i.e., AI/AN and Asian) and Hispanic ethnicity reported on death certificates might have been misclassified (7), possibly resulting in underestimates of death rates for some groups. Finally, the cause of death for certain persons might have been misclassified. Limited availability of testing for SARS-CoV-2, the virus that causes COVID-19, at the beginning of the COVID-19 pandemic might have resulted in an underestimation of COVID-19–associated deaths.

This report provides an overview of provisional U.S. mortality data for 2020. Provisional death estimates can give researchers and policymakers an early indication of shifts in mortality trends and provide actionable information sooner than the final mortality data that are released approximately 11 months after the end of the data year. These data can guide public health policies and interventions aimed at reducing numbers of deaths that are directly or indirectly associated with the COVID-19 pandemic and among persons most affected, including those who are older, male, or from disproportionately affected racial/ethnic minority groups.

Author(s): Farida B. Ahmad, Jodi A. Cisewski, Arialdi Miniño, Robert N. Anderson

Publication Date: 31 March 2021

Publication Site: CDC