Structure and check for pay integrity

Introduction

Designing and auditing fair pay practices are steps organizations everywhere can take to create more equitable workplaces. The gender pay gap is a well-documented, well-researched, and persistent problem. The systemic underpayment of women chips away at fairness and equality and may have a real economic impact.

To set and check for fair pay practices, organizations should consider these actions:

Set a compensation philosophy. Your philosophy guides how you award compensation.
Structure compensation processes and pay for the role. Structure provides a framework for compensation decisions, such as paying for the role and thereby supporting equal pay for equal work.
Run a pay integrity analysis. Once the two pieces above are in place, check to make sure the system is working as intended by conducting regular and rigorous analyses.

Understand the research

Understand the research Francine Blau, a leading researcher in the field of gender pay integrity at Cornell University’s School of Industrial and Labor Relations, and her colleague Lawrence Kahn looked at a large sample of US workers in 2010 and calculated the pay gap between men and women.

The authors distinguished between an unadjusted and adjusted gender pay gap. They found that unadjusted gap was 20.7% (women making roughly 80 cents on men's dollar). However, much of this differential can be attributed to non-gender variables such as the fact that a greater proportion of women than men are in lower paying occupations. Other variables such as experience, unionization, and region also help explain some of the difference. After controlling for all of these factors, Blau and Kahn found an adjusted 8.4% pay gap. This research tells us that even when you account for occupation, industry, experience, and other factors, women are still earning only 92 cents for every dollar a man earns.

Research has also examined how structure and accountability can establish pay integrity in organizations. In one longitudinal study, Emilio Castilla at the MIT Sloan School of Management, worked with an organization that had historically seen statistically significant lower pay increases for women and minorities. In Castilla’s study, managers were trained to base decisions about pay increases on consistent criteria (i.e., performance ratings) and asked to justify bonus amounts. Then a newly created committee reviewed managers’ pay increase recommendations and had the authority to modify pay decisions to correct any problems. Leaders across the organization also received an annual report on all pay decisions made within their division.

Castilla monitored the next four years of pay data and found pay increases no longer differed by demographic group. The adoption of increased structure, supported by accountability and transparency, decreased the pay gap.

Structure your pay process

There are many ways to structure your pay processes. To get started, identify what roles exist in your organization (or you plan to hire for), and detail what is required to be successful in those roles. You can organize this information by identifying what roles exist in your organization, and what is required to be successful in those roles. This could include naming job titles, summarizing job tasks and responsibilities, identifying the knowledge and skills needed to do the job as part of a job analysis.

Once you’ve completed a job analysis, it’s time to understand what these jobs are worth in the market (i.e., what other companies pay employees for doing similar work) and how you want to position employees’ pay compared to the market (e.g., you could target pay at the 50th percentile of the market, the most common practice). You can define “market” based on where you get or lose your people (e.g., organizations in your industry, in the same geographic location). When thinking about how you want to position your employees’ pay compared to the market, be sure to keep your budget in mind.

Using the job descriptions from the job analysis, compare your roles to market pay data for similar roles. It is important that you base your comparison on more than just a job title as companies assign job titles differently. A vice president at a startup may have five years of experience and a small scope of responsibility, while a vice president at a well-established, multinational organization may have 30 years of experience and be responsible for large groups or budgets. You can get free market data from the US Bureau of Labor Statistics for standard jobs, and more detailed data based on surveys by industry organizations and consulting firms. If market data is missing or unavailable, you can consider filling in gaps using the median pay of your current employees in that role, or using the market data for jobs of similar scope, responsibility, or complexity.

Many large organizations set specific pay targets with wide pay ranges. Pay ranges represent the minimum and maximum pay for a given role based on the market data and are typically based on a percentage above and below a job’s pay target (e.g., +/- 20%). Implementing a pay structure that includes targets and ranges makes it easier to consistently pay for each role and helps you spot outliers.

Reference your pay structure every time you assign pay—from new hire pay to promotion increases. For example, new hires at Google have their compensation set in relation to a pay target associated with their role, level, location. Keep in mind that local legislation may actually prevent you from either asking about or using prior pay as an input.

As you evaluate your programs, remember that pay integrity doesn’t always mean “pay everyone the same amount.” Based on your compensation philosophy, there could be factors that you determine should differentiate pay among people in similar jobs. At Google, for example, it is expected that top performers earn more than others in the same role.

Prepare to conduct a pay integrity analysis

Setting up a structured compensation system should help you achieve pay integrity. To ensure your pay system is working as intended, you can conduct a pay integrity analysis to verify that there are no pay differences by gender or race/ethnicity. The analysis is a way of understanding whether pay at your organization is influenced by factors you want (e.g., type of job, location) or those you don't want (e.g., gender, race/ethnicity). This type of analysis can help reveal trends across groups as well as individual outliers. At Google, all compensation decisions (e.g., assigning incoming salary, salary adjustments, allocating bonus and equity) are driven by a pay-for-performance philosophy. The team closely monitors other processes, including performance management and promotion, and at the close of each cycle analyzes outcomes by gender and ethnicity.

Pay integrity analyses can be complex, so you’ll first want to have a structured compensation process, complete with a compensation philosophy and pay targets. Once you have that, and if you decide a pay integrity analysis is right for your organization, you’ll want to start by doing the following:

Understand the legal context. Before you start this analysis, connect with a lawyer who has expertise in employment issues. This analysis could have a wide range of legal implications and you should understand all the considerations.
Define the questions you want to answer. Clearly define what you are investigating in each analysis. For example, is there pay integrity among male and female new hires doing similar work?
Standardize your compensation variable. Pay targets may differ between jobs, so standardizing your compensation variable allows you to make comparisons within job and across counties and regions. One way to do so is to use the following steps: 1. Convert all compensation into one type of currency. If the majority of your company’s employees are in the United States, you might want to convert all compensation to USD. 2. Apply a logarithmic transformation to the compensation variable. These steps allow you to identify percentage (vs. absolute) differences in compensation.
Determine how to reliably detect if there any differences. This varies from organization to organization and depends on what variables you want to look at. Before selecting or discounting a sample group, see what sample sizes work best for your organization and analysis. For small organizations with under a few dozen employees, running a pay integrity analysis may not be practical, but examining compensation data is still valuable. For example, if there’s only one woman in an organization, you can check how her pay compares to that of male colleagues with the same job, level, amount of experience, etc., or calculate the average of men’s and women’s salaries with some simple control variables.

Once you’ve done the above, it’s time to begin the analysis.

Identify variables to test

Choose variables for your pay equity analysis

These are examples of variables that could be included in a pay equity analysis. Control variables will vary depending on your compensation philosophy, the indepenent variable will be related to what you are checking for (in this case, gender equity), and the dependent variable will depend on how you pay your people (e.g., salary, bonus, stock).

Accounting for these Is this correlated With this outcome?

Control variables	Independent variable	Dependent variable
Performance Scores:	gender	compensation
Job Level
Tenure
Job Role

Now that you’ve established who or what you want to compare, you can identify the variables you want to test.

Your independent variable is what you are testing to see if it affects the dependent variable or outcome you care about. The dependent variable is the outcome measure you care about that might be affected by the control and independent variable(s). For a gender pay integrity analysis, the independent variable is employee gender and the dependent variable is a pay outcome (e.g., salary).

Factors that should influence pay, as determined by your compensation philosophy, are your control variables. For example, your philosophy might mean that employees at a higher job level should receive more compensation or employees with lower performance ratings should receive lower compensation. Therefore, example control variables might include job level, performance, or work experience.

Looking only at the magnitude of pay differences can create false alarm where differences are not statistically meaningful, and using pay ratios helps fix the absolute value problems of exchange rates, geographies, levels, etc.

Analyze the data and look for variance

Once you have selected your variables, it’s time to begin the actual analysis. This requires someone with proficiency in statistics, including an understanding of standard deviation, variance, regression models, and the ability to interpret results.

Build and clean your data set. Good analyses rely on good data. Spot check your data to see if any data are missing or incorrect. If your data set is incomplete, understand why and determine if this might influence your results.
Identify substantially similar jobs. Pay integrity analyses are most useful when comparing individuals doing similar work. You may have already grouped them when creating pay targets (e.g., entry level analysts in Sales may also be grouped with entry level analysts in Finance).
Check for multicollinearity, which occurs when two of the control variables are highly correlated and can result in drastic changes to your results. For example, tenure and job level could be highly correlated; the longer you’re at an organization the more likely you are to have moved up in level. Decide which variable is most important to control for and remove the other one from your regression analysis.
Run a regression analysis. The most common and rigorous method for conducting a thorough pay integrity analysis is running a regression model. A regression will tell you if there is a significant relationship between a number of variables (e.g., analyzing if gender actually impacts pay when you control for other factors that should influence compensation). It helps you avoid conflating differences in compensation due to the independent variable (e.g., gender) and another variable that could legitimately explain differences in compensation (e.g., job level, which should influence compensation). Specifically, an ordinary least squares regression will allow you to analyze all the control variables at the same time (see how to run one in a spreadsheet or in R). Enter your control variables (e.g., job level) as step one in your regression. Next, add your independent variable (e.g., gender), and then the dependent variable (e.g., pay ratio).
Test for significance. Looking at your regression output, are there any statistically significant gaps between groups, as determined by a significance test? If so, conduct an effect size calculation to better understand the magnitude of any differences. Then, analyze which of the variables in your analysis account for any of those large gaps.
Check your work. Have a colleague review your methodology, your spreadsheet formulas or code, and your assumptions (e.g., how you define similar work). Review your descriptive statistics (e.g., averages, variance, correlations) to ensure that they make sense (e.g., does anyone have negative compensation? are there any outliers like someone earning $1 per year?). Talk through your results with someone to see if they make logical sense.

Simple regression output

A regression analysis will tell you if there is a significant relationship between a number of variables (e.g., does gender impact pay when you control for other factors that should influence compensation?).

This is an example output from an ordinary least squares regression in R where the independent variable is gender, the dependent variable is pay ratio, and there are four control variables determined to be important to account for in this organization's analysis (e.g., performance scores, job level, seniority, work experience). In this example, after accounting for all control variables, women on average have a statistically significant 0.01 higher pay ratio than men.

	Estimate	Standard error	1 value	Pr (>/t/)
Intercept	0.70	0.70	0.70	0.70
Gender	0.01	0.01	2.31	0.02
Control_1	0.04	0.01	0.83	0.41
Control_2	0.01	0.02	2.74	0.00
Control_3	0.03	0.01	1.63	0.11
Control_4	0.01	0.01	0.86	0.39

Act on your findings

If you find pay integrity, high five! Check again after your next compensation planning cycle.

If you don’t, it’s time to dig deeper:

Seek legal counsel. Before taking any action, check back in with your legal advisor to understand the context and requirements for making changes.

Check your people processes. And do it often. If you do find inequities, first look into what’s driving these differences and if your people processes are working as intended. For example, if you’re not certain that your performance management system is totally objective, consider running your analysis with and without these variables to check for bias in performance ratings or promotions. Also, take note of all the decision points within your processes. Who is responsible for which decision? How might people respond differently based on the training, norms, and incentives surrounding the decision? Your results may surprise you. For example, in 2010 Google found that female engineers were self-nominating for promotion at a lower rate. This resulted in significant promotion rate differences. The solution? Showing employees the data, reminding everyone to put themselves up for promotion if they’re ready, and making sure managers were accountable all led to promotion rates equalizing.

This is also the time to look at your compensation processes again and ask questions like:

Is your compensation philosophy fair?

Is the organization consistently using pay structures?

How might the assumptions you made in your analysis affect the finding?

Consider pay adjustments. It’s important to make changes in the context of your compensation philosophy. If you see a significant difference between groups where no difference should exist, consider a one-time pay adjustment. For example, in 2015 Salesforce identified a gender pay gap at their company, and has since committed to compensation integrity While immediate adjustments can help achieve integrity in the short-term, it’s critical to investigate and fix the root causes as well so you can ensure inequities don’t creep back in over time.

Hold decision-makers accountable to reducing bias. In most organizations, managers play a big role in addressing pay integrity as they are responsible for the majority of pay and performance decisions. Start by educating managers about the importance of gender pay integrity and how their own preconceived notions might creep into the pay process. During compensation planning cycles, remind managers to be aware of gender pay integrity, let them know their pay decisions will be reviewed, and share back any appropriate data with them afterward. Some organizations also form internal committees to regularly review gender pay data.

Ensuring gender pay integrity requires long-term commitment from everyone. Whether you’re starting with a simple spreadsheet or conducting a full compensation analysis, organizations of all sizes are responsible to create fair and equitable workplaces.