How the “what’s your current salary?” question hurts the gender pay gap

In 2015 we added 8,214 employees to Google. And the women we hired, on average, received a 30 percent bigger salary increase upon joining the company, compared to men. Does that sound fair to you?

This op-ed was originally published in The Washington Post.

Study after study demonstrates that women are paid less than men in the United States. Equal Pay Day, three weeks ago, marked the number of additional days women needed to work into 2016 to "catch up" with what men were paid in the prior year. Indeed, in a recent paper, Cornell professors Francine Blau and Lawrence Kahn found that women were paid 79 cents for each dollar a man was paid. Even after adjusting for type of job, industry, experience, location and education, the gap remained 92 cents for each dollar.

A number of factors contribute to that difference. Some are behavioral. For example, as many have noted, women are less likely than men to negotiate salaries or ask for a raise.

But the fundamental problem is hidden in the way we make pay decisions. Fixing it requires us to focus on paying people what a job is worth — and not basing a pay decision on someone’s current salary.

We are all susceptible to something called anchoring bias, where our brains get stuck on an initial number. One of the first studies conducted on anchoring, by professors Amos Tversky at Stanford University and Daniel Kahneman at Princeton, asked people to spin a wheel with numbers on it and then estimate how many African countries were United Nations members. But the scientists secretly controlled where the wheel landed, setting it to hit either 10 or 65. When the wheel came up 10, people guessed that 25 percent of African nations were in the U.N. When it landed on 65, they guessed 45 percent. They had unconsciously anchored on the initial number.

When it comes to making pay decisions, we anchor too much on someone’s current salary instead of what the job is worth. Imagine hiring two accountants. One (call her Eliza) currently makes $50,000 and the other (Alexander) makes $58,000. And let’s say the average accountant in your company makes $60,000. It feels natural to offer Alexander a salary of $60,000, just like everyone else gets. But for most managers, it feels wrong to give Eliza the same salary. After all, that’s a $10,000 raise! Wouldn’t that be unfair to Alexander, who only got a $2,000 raise? And why not save a few bucks by paying people based on their past salaries?

But that approach ignores the reality that women are consistently, and unjustly, paid less than men for the same work.

There are many challenges when it comes to equality in the workforce. In the technology industry, we’re working to grow female representation within both our engineering teams and leadership ranks by getting more women into the field and better supporting their advancement. But ensuring women and men are paid equally for equal work is something companies should and can fix today, with two straightforward steps.

First, avoid anchoring to someone’s current salary. Instead, pay for what the job is worth by setting a pay target for each job when hiring and promoting. At Google, recruiters typically ask about salary as a data point, but neither they nor hiring managers make decisions about pay. Instead, offers are determined by our "people operations" team, which provides a bulwark of objectivity and fairness. If a candidate's current pay is below our target for that job, we simply ignore the prior salary and offer our target.

We calculate pay targets based on industry surveys, but you can simply use the average of what your current people make, or find decent estimates online. The further someone’s current salary is from the target, the bigger raise she or he gets when they are hired. That’s why we’d give Eliza a $10,000 raise.

By paying for the role, not the person, you start with a clean slate and mitigate any bias embedded within Eliza’s prior compensation. In other words, you correct the pay bias that exists in society.

Second, make sure your system continues to work as intended. Even if pay is fair upon hire or promotion, what if women are promoted more slowly than men? What if bias creeps into pay for people who aren’t promoted? We watch for these issues, and correct them if they arise. We’re making information available now about how organizations can run their own pay equity analyses.

Applying an approach similar to Blau and Kahn’s — where we control for the type of job, education, experience, location, etc. — there is no difference between what women and men are paid at Google.

The systemic underpayment of women chips away at our society's values of fairness and equality — and has a real economic impact on our daughters, sisters and mothers. It also diminishes men if we stand by and allow this bias against women to persist.

But we can change the system. If, instead of falling prey to anchoring bias, we corrected women’s wages with every hire and promotion, I believe we’d eliminate the gender wage gap within a decade. Let’s begin.