OLD BIASES DIE HARD

Programming while black will cost you $10,000 in salary

Racial discrimination may be alive and well in the technology industry.

The online hiring platform Hired released its 2017 State of Salaries report today (Feb. 9). The San Francisco-based company has a database of 280,000 interview requests and job offers for 45,000 engineers and technologists on its platform. Hired lets candidates post their preferred salaries so that companies can interview them and extend offers. Because of that, the company says, it has unique insights into salary expectations and actual offers.

Hired found striking differences among racial groups.

First, African-Americans were 49% more likely to get job offers compared to the average white candidate. Latinos and Asians, on the other hand, were 26% and 45% less likely, respectively, than white candidates were to receive offers.

Yet when it came to salaries, blacks were at a greater disadvantage. At first, the data looks promising. On average, blacks were offered $2,000 more than the salary preference stated on their profile. The reverse relationship held true for other racial groups, especially whites, who received about $1,600 less than their preference on average.

But the primary reason appears to be that blacks’ salary expectations are far lower than peers in different racial groups. Blacks on average asked for salaries of about $113,000 compared to $126,000 for whites. Ultimately, whites received the highest annual compensation offers ($124,900) compared to other racial groups, especially blacks ($114,800).

Race Preferred salary Offered salary Difference offered vs. average salary
White $126,600 $124,900 3.6%
Asian $122,600 $122,600 1.7%
Latino $121,100 $120,100 -0.4%
Black $112,800 $114,800 -4.8%

Even after Hired controlled for experience and age, the same pattern appeared (since the sample sizes for these subgroups were not large enough to be statistically significant, they were not included in the report). The company used a dataset of 1,400 job candidates in San Francisco and New York to minimize geographic and other variables for its racial analysis.

Hired is not certain about the cause for this discrepancy, but its data scientist Jessica Kirkpatrick has a hunch:

“The question is why would a candidate with the same years of experience at the same job ask for less money? We think this [lower] preferred salary is a sign mostly of historical bias and discrimination against people. The preferred salaries actually bake in the historical bias people have felt [from] years of discrimination… Companies are taking advantage of that.”

Hired found a similar phenomenon among women. “Women undervalued themselves for preferred salaries compared to male counterparts,” said Kirkpatrick. “They were asking for less money, and getting higher salaries than they asked for, but still lower than their male counterparts.”

Indeed, not much has changed from 2014, when the American Institute for Economic Research found that Hispanics, Asians and blacks all made less than their white counterparts (although Hispanics were worst off in this study, earning $16,353 less annually on average). The study controlled for education, occupation, age, geography, gender, citizenship status, marital status, and children.

Fixing this problem is not straightforward. Employers may not be discriminating explicitly on race (all personal data on race and age is hidden on Hired, and only available voluntarily from candidates). Lower salary expectations and discomfort toward negotiating could put specific groups at a disadvantage, as well as a history of lower salaries that would persist over time as candidates based expectations on their personal past.

 “Merit-based practices may trigger bias against women and ethnic minorities.” Emilio Castilla, a researcher at the MIT Sloan School of Management, has found that even merit-based bonuses can exacerbate bias. His research at a large private company found bonuses for women, ethnic minorities, and employees born outside the US were lower than white men with virtually identical performance reports, jobs, units, and managers. “The lesson is not that companies shouldn’t adopt merit-based practices but that the pursuit of meritocracy is more difficult than it first appears,” Castilla told USA Today. “If not designed and implemented carefully, merit-based practices may trigger bias against women and ethnic minorities.”

One approach taken by companies such as Salesforce, Facebook, and Microsoft has been to equalize gender pay gaps by analyzing the entire company’s payroll, and adjusting pay to eliminate gender discrepancies.

In 2015, Salesforce adjusted the pay of 6% of its 17,000 salaried employees after a comprehensive review. CEO Marc Benioff said the cost of eliminating the gender pay gap was about $3 million. The $52-billion company says it will repeat the process this year after acquiring more than a dozen new companies in 2016. None of the companies, so far, has revealed detailed data about how they conducted the analyses based on job, location, seniority, and other factors.

Hired started tackling the problem itself in late 2016 by giving candidates access to market rates for salaries when they set their own preferences (it will be assessing the results later this year). Internally, the company has also opted to use objective salary guidelines such as education and experience rather than relying on negotiating ability or pay history. Kirkpatrick said Hired might lose candidates negotiating higher salaries, but fairness better served the long-term health of the organization.

home our picks popular latest obsessions search