What actually drives the gender pay gap?

Using data from TransparentCareer, we analyzed the factors contributing to the wage gap between men and women.

As you have probably heard before, there is a gap in pay between men and women, and it has been widely documented. While there is no question that this gap does in fact exist, the more important question is why does it exist. Here, we embark on a journey to peel back the layers that contribute to this gap. We think the time and detail we put into analyzing this issue is well worth it and we hope you do too. Stay with us.


One of the hardest things to do when analyzing the wage gap is to create an apples-to-apples comparison. However, because of the depth of the data we collect at TransparentCareer, we are in a unique position to answer this question.

For this analysis, we looked specifically at the full-time jobs MBAs got right out of school. This helps to eliminate a lot of the issues normally found in studies like this, because the individuals in our dataset have essentially identical credentials and have access to the exact same set of jobs.

Now for the high level results:

The breakdown of the MBA gender wage gap. (n=2,323)

No surprise here: we found a wage gap. On average, men make $180K in total compensation versus women who make $166K — a gap of $14,000. Thats pretty huge coming right out of graduate school. As the chart shows, the largest gaps arise from differences in bonuses and salaries.

Now onto the real question we want to answer: why does this gap exist?

Choice of job function as a driver of the wage gap

Our first hypothesis was the job mix hypothesis: that women might be attracted to or situationally led into job functions which happen to have lower market salaries on average. So we broke down the pay gap by job function. Here’s what we found:

Okay…I know it’s a lot, so let me unpack what you’re looking at. For each job function (the general category the job fits into), we analyzed the relative proportion of men and women, their average compensation, and the average compensation for that function as a whole. From these, we computed 3 metrics: the wage gap (the average male compensation minus the average female compensation), a compensation index, and a female representation index.

For the female representation index, a score of 100 would mean that women are equally represented relative to their overall prevalence in our data set. An index of 60, for example, means that women are 40% less likely than average to choose that job function. An index of 120 would mean 20% more likely, and so on.

As you can see, women are far more likely to take roles in Human Resources or Marketing compared to men, and far less likely to work in financial fields like Private Equity, Venture Capital, Investment Management, and Investment Banking. Why this occurs is probably a topic for another study…

Now, for this difference in role choice to drive the wage gap, the jobs in which women are under-represented must pay more on average, and vice-versa.

We found exactly that. The scatter plot above shows an inverse correlation between the average total compensation and the proportion of women for a particular job function.

Beyond the correlation, we wanted to know how much of the $14K overall wage gap was driven by this difference in role choice alone. After running some regressions, we found that roughly $4.5K of the gap was driven by this difference in job function choice, or about 33% of the total.

If you’re interested, a different study at Stanford analyzed this exact effect and found job choice to be responsible for 53% of the wage gap.


So where does the rest come from?

First, we thought maybe men were more likely to negotiate their offers up, but we couldn’t find any evidence of that. Both men and women were equally likely to negotiate, with ~19% of each group saying they attempted to negotiate their offer. The results of successful negotiation were also no different, resulting in about $10K in increased pay for both sexes. (Side note: if appropriate, you absolutely should negotiate your offers…we even built a tool to help you benchmark your offers automatically).

Differing positions as a driver of the wage gap

As we showed above, just 1/3rd of the MBA gender pay gap came from differences in gender representation across job functions. That means the other 66% comes from differences in pay within each function.

If you look at the table (re-pasted below), you can see large differences in average pay even within functions. Investment Management, for example, shows a whopping $97K gap (!), but even functions with notoriously standardized compensation packages, like Consulting, show sizable gaps.

Where does this difference come from?

We initially thought that the actual positions/titles that men and women go into within these high-level functions might differ. To look at this, we drilled down into Consulting and Investment Management to see if the position titles and their corresponding compensation were different for men and women.

N=73

For Investment Management we found two things going on.

  1. The highest paying roles were for “analysts”, who made significantly more than “associates”. Those with the analyst title consisted of almost all men in our dataset.
  2. For the exact same job title in the exact same job function, there was still a significant difference in total compensation.

Within Consulting, both genders were hired into the same exact roles, but again there were striking differences in compensation, with men making more.

After further analysis, it appears that hiring MBA students into different roles was fairly localized to Investment Management, both because of the scale at which it occurred and because of the absolute dollar difference in pay between these roles. Overall, we estimate this factor to represent $1.3K, or 9%, of the overall wage gap observed.

Going deeper

So what about the difference in pay for the exact same title within the exact same job function? Where does that come from?

We see 3 potential drivers at play:

  1. Reporting Bias: Women are actually making the same amount, but are reporting lower compensation in our forms
  2. Company Attraction/Selection Bias: The companies that attract women for these same roles pay less on average
  3. Fundamental Pay Gap: There is a persistent and fundamental wage gap, meaning women are actually receiving less pay for the exact same job

Let’s explore these three possibilities.

1) Reporting Bias: women report lower compensation than men in surveys

While it may seem difficult to determine whether women consistently under-report their earnings, we came up with a methodology for doing so. We used data from a set of top consulting firms, which we know to offer completely standardized compensation to MBAs regardless of gender. We looked to see if there was a difference in how women and men reported these offers.

For companies where offers are completely standardized, women report lower total compensation than men (N=166)

Shockingly, we found that men reported their compensation $8,000 higher than women did. Within salary there was no difference, but on bonus and other compensation (which includes 401K matching and stock) men reported significantly higher wages.

This means that of the $12K wage gap in consulting, reporting differences may explain nearly two-thirds. According our estimates, across industries reporting bias accounts for a whopping 43% of the overall MBA wage gap.

What’s noteworthy here is that offered salary is an objective, immutable number. Performance bonuses and other compensation, however, can be interpreted more subjectively — based on a range candidates are given in an offer letter. Its likely that men are estimating they will achieve higher performance bonuses than women, which leads to a larger reported wage gap.

2) Women go to companies that pay lower on average

Next, we wanted to see if women were more likely within the same job function to work for companies that pay lower wages on average. Using the consulting industry for our testing ground for this hypothesis, we looked to see if there was an over-representation of women at companies with lower compensation and vice versa.

Based on our analysis, we did not find that women were more likely to work for companies that paid less. In fact, of the $12,000 wage gap in the consulting industry, we were only able to attribute $63 to this effect — basically nothing.

3) Putting it all together: is there a fundamental gender pay gap?

Okay we’ve been through a lot and if you’re still reading this we appreciate you joining us on this journey. Now that we’ve analyzed a lot of the potential drivers of the wage gap, let’s bring it all together.

We’ve assessed 5 potential drivers of the wage gap:

  1. Job Function Differences: Women may choose job functions that pay less than average
  2. Specific Position Differences: Within each job function, women may be hired into specific roles that pay less
  3. Reporting Bias: Women may under-report their compensation compared to men
  4. Company Selection Differences: Within specific industries, women may choose to work for companies that pay less
  5. Negotiation Differences: Women may negotiate their compensation packages less often or less successfully
Estimating the factors contributing the observed gender wage gap for MBAs

Based on the analysis, we were able to identify that the first 3 had a significant contribution to the observed wage gap. We took each of these drivers and estimated the proportion they contributed to the overall wage gap and believe these factors explain roughly 85% of the gap.

That leaves a gap of at least $2,200 that we are not able to explain. The remaining variation could due to a variety of factors, including a fundamental wage gap (i.e. women may actually be receiving lower pay for the same job than men).

While this is just one dataset of a specific group of people, we hope it illustrates the complexity of this issue: even for a very homogeneous population, the drivers of the wage gap are deep and numerous.

Beyond simply identifying these drivers, there are additional questions that need to be addressed. For instance, why do women report differently than men? Why do women and men choose the job functions they do, and is this a problem? If it is a problem, what should we do about it?

It’s clear that women should not be paid less than men for the same work. However, the complexity of this issue makes it difficult to separate when this is actually occurring, due to more nuanced variables that may be driving a gap. Nevertheless, each of these types of issues is a problem — from legal, ethical, and social perspectives — but each deserves understanding and better identification of when problems exist.

This analysis is by no means perfect or even close to it. We hope instead that it brings about further discussion on this topic, and how to develop datasets that can further address and identify problematic wage gaps.

Thank you for reading and for entertaining our attempts at analyzing something very complicated.

Discuss. Share. Give Feedback. Offer to help. We are open to it all.

The TransparentCareer Team

Note: If you have are interested in doing deeper analysis on the dataset, let me know and I can share a sanitized version with you. We’d love to get better insights and would love your help.