Three ways of calculating gender pay gap

Every day, we see women leaving for their offices, travelling by public transport and returning to their houses after a strenuous shift in the night. Things surely seem to have changed. Women are equal contributors to the goods and services production of India. Then what exactly is causing the gender pay gap, that has been actively discussed in the West as well as the East?

A stringent legislation, namely, the Equal Remuneration Act, 1976 was passed by the Government of India, to ensure equal pay to both the genders and prevent any kind of discrimination on the grounds of sex. Yet, India has slipped to the 112th position from the previous 108th in 2018, according to the Gender Gap Index. While in the US, the data from the American Association of University Women (AAUW) shows that women are still paid 82 cents to every dollar paid to a man. Why does this pattern continue?

Let’s understand the basics first.

What is gender pay gap?

It is the disparity between the salary of a man and a woman, working in similar positions or roles and shouldering similar responsibilities. This takes into account the hours that they give to their work, the experience they possess and the tasks that they are required to perform.

Organisations have been focusing a lot on what is the right and the wrong method of calculating gender pay gap. But, the fact is that there is no ‘right’ and ‘wrong’. Each and every calculation gives us a new understanding of the pay dynamics in the following ways:
a) Within an organisation, who gets paid more, in totality — be it due to position, experience, pay discrimination, assignments allotted, etc.? After all, the pay gap is also a gap of opportunities, jobs lost, unemployment issues, and so on.

b) Is there a disparity among the new hires and the senior employees (fun fact — out of 75 per cent men and 74 per cent women who start their careers as single contributors (they are not managers), 36 per cent men qualify for the position of a supervisor, while only 30 per cent women make it to that category. Also, in the age group of 20-24, women earn 90 per cent of what their male counterparts earn!)

c) Is everyone eligible for a pay raise, across the board? Should controls be filtered out to see if the women actually deserve the percentage of raise? Here, statistics comes to rescue.

d) Is there a rampant occupational segregation? For instance, why are maximum nurses women? Why are women mostly lurking in the bottom lines of the organisation? Are they automatically endowed with a prejudiced compensation or are their tasks assigned with a lower comparative value? But then, what is the value assigned to a particular role? Yes, you can find that out too. Data is magic!

Types of pay gap — average, equal and median

The three traditional ways of calculation give us the average pay gap, the equal pay gap and the median pay gap. What are these?

1. Average pay gap: This is also called the ‘unadjusted’ pay gap, because no controls are used to determine if the gap percentage will vary according to varying characteristics, such as age, seniority, work experience, years of education and so on.

Yes, this does work. We get a rough idea of the opportunities that women are losing out on. Whatsoever the reason of disparity is, one should keep in mind the biases that come to play in the hiring process and while assigning merit to a job.

This is derived by calculating the average/mean salary of the male employees in the organisation and then, doing the same with that of the women staff.

{(average male salary-average female salary)/average male salary} x 100

For instance, if the average man earns Rs 40,000 and the average woman earns Rs 35,000, the difference in their salaries will be Rs 5,000. The formula given above will then be followed to find out the solution. The percentage pay gap in this case, applying the aforementioned formula, will be 12.5 per cent.

2. Median pay gap: Unlike mean pay gap, this helps us find out the disparity in ratio of men and women in higher positions, within an organisation. In statistical terms, to calculate median, the employees of an organisation are to be listed in increasing order of their salaries, separately according to their gender. In the US, according to the median pay gap, women earn 79 cents for every dollar earned by a man!

If the listed employees are made to line up, two halves can be created. On doing this, from two extremes, one employee will get left in the centre with approximately the in-between income of the entire batch.

Suppose an organisation has five women earning five-figure salaries of Rs 40,000, Rs 30,000, Rs 20,000, Rs 50,000 and Rs 20,000 respectively.

If they are lined up in increasing order of salaries, it will give us-

Rs 20,000, Rs 20,000, Rs 30,000, Rs 40,000, Rs 50,000

On creating two equal halves, the figure in between the two halves is 30,000. Therefore, the median pay for this batch of employees will be Rs 30,000. Similarly, if an organisation has 105 male employees, we halve the closest even and add 1 to it, to get that employee whose salary would be the median salary. In this example it would be- (104/ 2) + 1 = 53

So, the 53rd employee’s salary would be the median pay. On comparing the difference in the median pays of both the genders, the median gender pay gap emerges.

[(Median salary of man – Median salary of woman) x 100]/ median salary of man = percentage of median p.g

Whereas the mean salary for the batch would be-

(20000+20000+30000+40000+50000)/5 = 160000/5 = 32,000

If this isn’t as accurate as the mean pay gap percentage, how does it help exactly?

a) If the male employees’ salaries rise drastically and retain the growth in the rise/transition, whereas the female employees’ salaries rise gradually or spike drastically in the end for a select few senior positions, we can infer from the median pay of each gender, as to how very few women occupy positions offering outstanding pay, with more power and responsibilities.

b) It may also highlight how many women are stereotypically lagging behind in roles that are less important or less paying. In women-dominated sectors, the pay is significantly low.

3. Equal pay gap: This has been talked about the most. It is the most appropriate way of calculating the gender pay gap, as the employees whose salaries are compared perform the same work and probably work for the same number of hours, even while possessing equal experience and holding the same responsibilities.

Besides these general mathematical calculations, there is an advanced way of calculating the gap using calculus. Studies show that this is the most accurate of them all. The annual salary of an employee gives us the male advantage in rupees. Whereas, taking a log of wages gives a better idea of the percentage-raise required to bridge the gap. (how can we ignore the cost benefits and feasibility of these solutions?)

Let’s try it ourselves.

The formula is-

Log of [{(average male salary- average female salary)/average male salary} x 100]

Which can be simplified as log of percentage of mean pay gap. When we receive the answer to this, we antilog the solution to go back to the real figure.

Using the above example, where the man earns Rs 40,000 and the woman earns Rs 35,000, let’s do it again-

= log [ (5,000 / 40,000) x 100]

= log (50/ 4)

=log (25/ 2)

= log (25) – log (2) {using the rule log (a/b)= log a -log b}

= 1.39 – 0.30

= 1.09

Antilog (1.10) = 12.58 (this will reverse the log of the percentage to give us back the figure we need)

If you compare the first average pay gap and the last average pay gap, i.e, 12.5 per cent and 12.58 per cent, a difference of 0.08 per cent is seen, which is a significant difference when it comes to salaries worth crores!

The best part of this is, we can extend the calculation too. Now, if there are n number of employees in an organisation, we can create a table of the data and create our own equation, to determine the coefficient. The equation that is used for a large number of employees, that is, to get the ‘adjusted’ pay gap, using controls varying from employee to employee will be-

Log Salary i = ?1 Malei + ?2 Controlsi + ?i

In this equation, if we take the log of the salary in the L.H.S, Malei stands for the male dummy whose salary is taken, Controlsi can be regulated to find out the required solution (for instance, 10, if the work experience of the employee is 10 years) and ?i is the standard deviation, that can be calculated with respect to the mean of the data given.

?1 is the coefficient, which remains if we calculate the linear equation using average male and female salary, regulating the controls. This gives us the approximate gender gap percentage, which is way more accurate than the unadjusted equation as we can change the controls to see if there is a layered form of pay disparity.

We can also calculate the percentage of probability of an employee getting raise after a year if she/he has a certain work experience, or the rise in hourly wages after a certain period of education, or to find out the years of education required by a man and a woman to maximise his or her earnings.

Hence, once we acquire the coefficients through a systematic data analysis and the subsequent standard deviation, we can get our answers by switching the controls from age to work experience to years of education etc.

Since this cannot be done all at once, taking the first step first matters. There is no better technology than the one that aims at bridging gaps and eliminating century-old traditions!