As part of our commitment to a fair workplace, Redfin has published data on how we pay women and men since 2015, with a pledge to discuss and remedy any significant gaps. In our December 2017 diversity report, we announced that we would shift the calendar for publishing this data to April, which let us separate the pay report from the diversity report, exploring each topic in more detail.
Here is the data synthesizing pay for every title at Redfin where there are at least two men and two women with that title. As in past years, we do not have enough data across a large number of departments to make the same comparisons based on race, ethnicity and non-binary gender.
|Employee Group||Women % of Average Salary||Men % of Average Salary|
|Managers of Agent Teams||98.80%||101.00%|
|Licensed Support Staff||101.20%||97.80%|
|Unlicensed Support Staff||99.90%||100.10%|
|Programmers, Testers, Product Managers, Seattle||99.80%||100.10%|
|Programmers, Testers, Product Managers, San Francisco||101.90%||99.00%|
|Real Estate and Business Operations Recruiters, Seattle||97.80%||103.60%|
|Customer Experience Coordinators, Partner Program, Seattle||100.20%||99.70%|
|IT Helpdesk Analysts, Seattle||99.90%||100.10%|
Gender-Based Differences in February 2018 Salaries for Groups of Redfin Employees
The gaps in salary between women and men are mostly small, in large part because we try to be rigorous about paying employees based on objective guidelines. Each gap is nonetheless worth exploring, so we can identify and correct sources of bias as they emerge.
The gaps among our agents and team managers surprised us the most, because the real estate teams are where compensation plans at Redfin are most regimented. We found that these gaps are the result of men having higher tenure, as agents earn modest increases to their base salaries over time. The opposite effect is why women make more money as licensed support staff such as transaction coordinators, as women in these roles have more tenure than men.
How We Account for Competing Offers
The other noteworthy gap is in recruiting, which just barely cleared the minimum number of men and women in the same role to support a comparison. Some of this gap is because recruiters with the same title are still at different levels of seniority, with some being closer to a promotion than others.
Like any company, Redfin sometimes finds itself in situations when a new hire demands more money than his or her peers. Our ideal process would be that we’d change our pay ranges on a regular schedule for both current employees and new hires when market-research data shows that pay for a particular role has gone up. In reality, we often learn of changes in market rates when candidates choose to share that information with us, or when current employees show us data about friends in similar roles at other companies, or even go so far as to get an offer somewhere else.
How we respond in these situations has an impact on pay equity. We often refuse to match the compensation requests of a candidate outside of our pay ranges. In cases when we pay above the range to get a candidate, we then have to evaluate whether to increase the pay of others in the same role, often using market-research data, and comparing differences in skills, performance and experience. Any pay increases to bring others to parity may not happen right away, but instead through regular performance and compensation reviews.
In a cost-conscious business that tries to charge the lowest-possible fees to our customers, it’s unrealistic to make the high-water mark for employee pay the average across the board. It results in an endless ratcheting up of pay. We understand that, as U.S. wages rise, this process is unavoidable and even healthy, but we’d still prefer that it happen in a deliberate way. We’re planning to watch our pay exceptions more closely in the year ahead to see what we can learn.
We’ve also updated our recruiting process to ensure we don’t ask candidates about their pay history. We made this change in part to comply with new laws in places like San Francisco that prohibit these questions, but rolled it out nationally to avoid perpetuating biases in how women and men are paid. This sometimes means we end up paying people with lower salary expectations more than we could get away with, but that’s what’s fair.
The pay comparisons for agents, team managers, and licensed support staff are limited to markets where we have at least two men and two women with the same title. Twenty-six such markets qualified for the pay comparison of buyers’ agents; 12 for sellers’ agents.
Calculating the Weighted Average
For each row in the table, we make several comparisons between women’s and men’s pay , one comparison for each role or level of seniority. For example, we employ entry-level, mid-level and senior-level software programmers in Seattle. We average women’s and men’s pay for each level, and do the same for testers and product managers.
To calculate a weighted average for each row, we take the average of those averages, weighting each comparison according to how many employees are being compared. If, hypothetically, there are 24 entry-level programmers but only eight senior programmers, we assign three times more weight to the pay comparison for entry-level programmers when calculating the overall gender gap for technical employees. Others in the industry, like Microsoft and Glassdoor, have similarly weighted or adjusted their averages for gender-based pay comparisons.
Accounting for Differences in Seniority Between Men and Women
This is the best way to compare people within a given level, and then to synthesize that for many different levels within a group. But it obscures one reason men make more money than women: there are often more men in senior positions than women. Just extend the hypothetical example above to imagine that only two of those eight senior programmers are women, whereas 18 of the 24 entry-level programmers are.
If, hypothetically, all the senior programmers earned $150,000 per year and all the entry-level programmers earned $75,000, the table above would show that men and women were paid equally, when in fact the average male wage was $112,500, and the average female wage was $82,500. We can argue whether it is reasonable to compare a man deep in his career to a woman just starting out in hers, or if in fact it is necessary to make those comparisons because a glass ceiling is the real reason so many women don’t make the big bucks. We can also acknowledge the tendency for male-dominated disciplines like engineering to pay more. What we can do now is show the distribution of women and men in entry-level and senior roles.
|Employee Group||% of Women in Entry-Level Individual Contributor Roles||% of Women in Senior-Level Individual Contributor Roles|
|Programmers, Testers, Product Managers, Seattle||34%||17%|
|Programmers, Testers, Product Managers, San Francisco||29%||50%|
The Percentage of Women in Entry-Level & Senior-Level Roles at Redfin by Group as of February 2018
We limit these comparisons to individual contributors because we do not have enough women and men in different levels of management within a group to make a comparison. We also excluded from this table whole groups that did not have enough people at senior levels for a comparison.
When we conduct this entire study in 2019, we expect our minimum to increase from two of each gender to four, so that we only compare roles with at least four men and four women; the company will have grown to a size where this still allows for a large number of comparisons, and we’ll avoid groups that are so small everyone can guess whose pay is an outlier.
We’re glad to see that many groups at Redfin have no major gap in average pay between men and women. But that doesn’t mean we’re satisfied. We constantly ask ourselves how we can be better at running this company fairly. If you have questions or comments about that, fire away!