Sunday, August 14, 2016

Gender Pay Equality

I’ve been reading, for decades now, about the gender pay gap. That women earn less for doing the same job. I remember hearing this the first time at a Laurie Anderson concert in 1991. It turns out that it’s true - if you calculate the average salary of a man it is greater than the average salary of a woman. However, what many of these sources don’t mention is that once you account for a laundry list of factors. Things like 1) career choice, 2) hours worked, 3) level of education, 4) amount of experience, etc. then that gap narrows to (depending on the study) between 0% and 9% - most have it around 3%. This essay isn’t intended to be a refutation of the wage gap myth. If you’re interested, and I hope you are, I recommend reading Dr. Warren Farrell’s Why Men Earn More.

Instead, what’s always interested me is why this myth has survived for over five decades, and shows no sign of dying. In addition to the numerous studies there’s also common sense questions one could ask, like:

  1. Why would companies risk violating numerous laws which make it illegal to underpay women?
  2. If companies could underpay equally productive women why even hire men? - that’s a HUGE cost savings.
  3. If women in their 20’s and 30’s out-earn men, why does that change once women start having children?

So here’s my theory - I think it comes down to two things…

The first is that many people buy into these myths because it comes from a seemingly reputable source. We believe it without examining the facts. I believe politicians and businesses exploit this myth, and our laziness for their own gain. I don’t intend to dwell on this hypothesis in this essay.

Recently I’ve come to realize that there’s a second reason. When I saw President Obama giving a speech about pay discrimination towards women I was quite surprised. Is he unaware of any of these studies? The first one was done by the US Dept. of Labor in 1973? Is he just pandering to the female vote? What gives?

Then I read the US Congress’ report titled Gender Pay Inequality. It lists factors like a women's greater propensity to:

  1. Take time off to care for children or other family members.
  2. Work part-time.
  3. Study in lower paying fields.
  4. Work occupations that pay less.
  5. Not be leaders.

But these are all choices right? Then it hit me - it seems that women want the freedom to make these choices and have pay equality.

I’m going to dive in and examine only the first item on the above list: women’s (so called) unpaid labor. What would it look like if society paid women for this work? I’m going to run a very crude simulation as a thought experiment. In this simulation the worker:

  1. Works for six years.
  2. Takes a six year break.
  3. After the break returns at their original salary.
  4. Gets a 2% raise each year (when working).
  5. Has a 42 year long working career.

Here’s a plot of men (no-break) vs. woman (break) annual daily wages:
Salary-unpaid-break.png
So lifetime earnings would look like this:
Lifetime-unpaid-break.png

Taxes

Higher incomes come with progressively higher taxes. Here’s a fictitious progressive tax rate topping out at 35% for $70K pre-tax net income:
Annual Tax.png
Using this to estimate taxes paid by gender, we have this post-tax salary:
Salary-Net-unpaid-break.png
The lifetime taxes paid has a shape similar to the earnings plot:
Lifetime-Taxes-unpaid-break.png
We can see that because men don’t take time off from work, they reach the higher tax bracket six years earlier. Once taxes are taken into consideration the lifetime earnings gap drops from 78.8% to 80.3%. However, because of the progressive tax scale women paid 74.9% for every $1 in taxes paid by a man, even though her post-tax income 80.3% as her male counterpart. I’m not arguing that this is unfair, only that the benefits of higher income aren’t quite as they seem until all factors are taken into consideration.

Achieving Pay Equality (Paying for Unpaid Labor)

So how do we achieve pay equality when the mother isn’t working at a company, but is instead at home caring for children? One solution is to pay women for this gap. What would that look like? In this scenario the government, or possibly the company, would continue to pay them their salary. Something like:
Salary-Net-paid-break.png
However, workers that remain continue to improve and increase their value to the company, and consequently get raises. This accrues over lifetime income:
Lifetime-Net-paid-break.png
In this scenario women made (after taxes) 80.3% of men when unaid for that six year break. When paid for that break the earnings gap dropped by just over half to 91.2% - but it didn’t go away. The only way to eliminate the earnings gap caused by an absence is to continue to give raises as though the break was never taken. This presents a variety of questions:

  1. Why should the company pay somebody to not work? For that matter how could a company afford to do so?
  2. Why should the company give raises for somebody that not only isn’t gaining skills, but is likely losing skills?
  3. Is this fair to the other workers that do not take a break?
  4. How can a company compete internationally with companies in other countries that do not impose these laws?
  5. Do women earn an average caretaker salary, or is it based on their pre-break salary?

But where does the money come from?

There is only one answer: it comes from workers (men and women) that do not take time off to care for children and family. Because these breaks can easily become several years long (or more), it’s unrealistic to place this burden on companies, leaving only one real solution: the government. Yes, we could force husbands to pay their wives, but since many children are raised by single parents the only real solution is the government - i.e. taxes.

In this scenario the government pays the worker's salary during break, which includes raises. The company is then obligated to rehire that employee at the new higher salary, as though they were never absent. This is a big chunk of cash, so taxes will need to be raised. Let’s raise taxes by 50%:

Annual Tax_150.png
This isn’t very progressive, and it only narrows the post-tax lifetime earnings gap to 92.9%. In order to fix this we need a more progressive tax scale:

Annual Tax_Progressive.png
With this progressive tax rate, which maxes out at 80%, we can finally achieve the desired goal of lifetime gender pay equality:
Lifetime-Net-paid-break-progressive.png
With the salary curve looking like this:
Salary-Net-paid-break-progressive.png
You can see the progressively larger drops in salary as seniority goes up. There are just too many perverse incentives to this tax plan that I don’t see it as being anything close to workable.

A Man Tax

Another proposed solution is a tax levied only on men - AKA a “man tax”. This idea has been floated around for a while now. If men pay 23.8% more taxes than women, we’ll have this tax rate:
Annual Tax-Men.png
And the career lifetime salaries would be:
Salary-Net-paid-break-man-tax.png
With lifetime total earnings at:
Lifetime-Net-paid-break-man-tax.png
However, lifetime tax payments would be:
Lifetime-Taxes-paid-break-man-tax.png
Which means that men pay 50% more taxes than women equating to 59.6% of taxes overall. This would likely disincentivize men from either working, or climbing the corporate ladder, having a negative overall effect on the economy.

Summary

In this thought experiment I pretended as though time off for child/family care was the only cause for the earnings gap. Given that women work less, take less risks, avoid dangerous jobs, pursue lower paying careers, and many other factors, I’m really at a loss as to how to solve this supposed problem. In all honesty I don’t really think there is a problem. People are free to make their own individual choices. Many of those choices result in higher or lower income, but that's fine, let the individual decide what s/he wants. The thought of putting government in this mix, and forcing companies to pay more for some jobs, pay for less hours, etc. seems frightening Orwellian, and will only serve to destroy our economy and the freedoms both men and women enjoy.

It sounds as though women are asking to be able to work less (and more flexible) hours in less valued careers while still be paid the same as others that work harder and longer. What I’d really like to hear is exactly what women (and men) are asking for here, and how do you expect to achieve it?

Tuesday, March 15, 2016

Women in STEM

I’m a software developer and my wife is a teacher in Silicon Valley. We have two kids about to enter high school so, as of late, we are hearing a lot, and I mean a lot, about STEM - specifically about how to get girls into STEM. We already have overt diversity quotas on tech company boards today, and there’s also evidence of soft quotas (like tying manager bonuses to diversity metrics) in tech. As bad as that is (and it is), you really can’t solve the diversity problem, if there even is one (more on that later). Giving jobs to non, or under-qualified, candidates is a recipe for disaster.

Common sense would dictate that you need to examine the entire pipeline, all the way back to infancy, if you really want to address the social forces supposedly steering females away from STEM. In this post I’m going to take a quick look one step back: college.

For this analysis I was lazy and grabbed some data that was handy - a dataset of three UK colleges provided by a Guardian article: The gender gap at universities: where are all the men? I’m sure the subjects of study differ from national averages, but given the dataset covers three universities, and that the female:male ratio is the same as the US overall average, I’m going to consider this sufficient for this analysis.

Total-now.png

This shows 38% more women seeking undergraduate degrees than men. Women are more likely to graduate, and men less likely. Forbes reports that women now make up 60% of graduating students1 - a full 50% more graduates than men.

Back on topic… Companies can’t just hire non-qualified candidates, universities need to produce them. If you want gender equality (i.e. 50/50) amongst the engineering staff, then you need gender equality amongst college graduates. You need to somehow balance the graduates in those fields.

By the way - this entire analysis is predicated on the assumption that women currently enrolled in a non-STEM subject could be convinced, and have the aptitude, to switch to STEM and obtain a degree. This is clearly impossible. This is really a thought experiment to examine what college would have to look like in order to enable the gender diversity currently desired by many companies and individuals.

Looking at the breakdown of the subjects studied we see this participation rate. I’ve highlighted the STEM: (Science, Technology, Engineering, and Math) in red.


STEM today.png
One observation is that, again for these three universities, all other fields are already dominated by women.

So how do we go about solving this? One approach is to bring in new female students - students not currently destined for university. These students would either enter the STEM fields, or displace students in other fields so that they may enter STEM.

STEM New.png
This is quite unrealistic for many reasons. First off, who will pay for these new students? How many students not currently attending can qualify for admission? How will you take women, and not men? You still need to get women to enter STEM - and if those new female students have the same proclivities as the currently enrolled students then that won’t solve the problem. I haven’t heard of anybody suggesting this approach, but did want to run the numbers.

So, assuming that there are no qualified, but not currently attending students, then we need to migrate them from one field of study to another. The second approach is to embark on a campaign to convince women to switch majors to one in STEM. So there would now be less women in the non-STEM subjects.
Screen Shot 2016-03-15 at 12.23.38 PM.png
A problem here is that we are now (likely) over producing STEM graduates and under producing non-STEM grads. Computer science has nearly doubled the number of graduates. This will depress the market for engineers and likely raise the salaries for non-STEM graduates now in low supply.

I would say that the above, encouraging women into STEM, is our current approach. Sure, the economics won’t play out as simply as I have represented them here, but this is the general idea.

A third approach, one that maintains the same number of graduates for each subject, but also ensures gender equality in STEM graduation, is where men are displaced from STEM subjects into non-STEM subjects.

STEM push out.png
Interesting to note that both solution #2 and #3 result in more men than women in business.

However, this crude analysis, even if done correctly by an economist, still ignores the absolute most fundamental issue. Even more than salaries, assuming one is earning a living wage, people want rewarding careers. Not everybody wants to be a software engineer - even those fully capable of becoming one. Our real measure must be equality of opportunity and not outcome. If twice as many women as men want to be engineers, and women are equally competent, then that should be reflected by an engineering force that is two thirds women.

As crude as my analysis is, and it is, I feel that the targets (50/50 in STEM) are even more crude. They do not take into consideration what students actually want to do for a living. They are already resulting in quotas, both in industry, and at the university level. They are pushing people into careers that they may not find rewarding or for which they are under qualified. They are also barring others from entering their chosen career. This is unfair to everyone: women, men, and the companies that employ them.

I’m lucky enough to have two amazing kids. My daughter is a self confident straight-A student. We’ve had both kids in summer camps to learn computer programming: both Java and Python. She and I have done basic electronics - soldering. She also loves to rebuild the carburetor on her motorcycle. That being said she does seem to be leaning towards subjects more associated with female interests: communications, art, law, biology, music. It frustrates me that the current STEM initiative, so popular today, is trying to steer her into tech when I know that her brother wants much more to be in that field. I’m seeing what appears to be the formation of a system within society that will unfairly give the job that he craves to her - and she doesn’t even want it. This is wrong.

Wednesday, February 3, 2016

The Big Questions Analysis

I recently watched an episode of the BBC’s The Big Questions: Does social media reveal men's hatred for women? (YouTube link).


The speakers were:
  • Connie St Louis
  • Kate Smurthwaite
  • Milo Yianoppolous
  • Ella Whelan


There are other analysis that have focused more on this discussion. Specifically Rantzerker 33: TL; DR All the men hate the wimmins. Like this analysis, I too had the following general impressions:


  1. It wasn’t really a debate over yes or no - it was assumed that men hate women, and that social media reveals that fact.
  2. Milo is generally put up against two feminists, and a moderator that is mostly in the feminist camp.
  3. The feminists get more time to speak.


So, my goal lately is to challenge my assumptions - with data. Is my general impression of something actually correct, or is my confirmation bias leading me to the wrong conclusion? So let’s dive right in.


First off the episode was only 19:25 long (including intro, 10 seconds of silence, etc.), but the total amount of talk time, for everyone, was 20:23. There was a lot of people talking over others. Transcribing this episode, and teasing apart who was interrupting whom took hours and hours.


There were 234 statements from all speakers.

Interruptions

Before we get started I want to discuss interruptions. There was a ton of interrupting - by everyone. It was quite a vigorous debate, but interruptions are disrupting so I felt it was important to consider these in my analysis.


Not all interruptions are equal. Some are hardly noticeable and do not interfere with the other speaker, while others drown out the other speakers. This analysis only attempts to classify interruptions as either warranted or unwarranted. A warranted interruption is only these two specific conditions:


  1. You have been accused of a crime.
  2. You have been accused of lying.


Most people would not let those types of accusations go unanswered, and I think that most would not hold that against the interruptor. Interrupting because you disagree with someone else’s points, or facts, is considered unwarranted - regardless of how disruptive that interruption was.

Talk Time

This is the talk time for each speaker (ignoring the moderator and audience).


Interruptions
These times ignore interruptions. So, hypothetically, if in the middle of speaker A’s one minute statement speaker B talked on top of speaker A for ten seconds, speaker A is still counted as having spoke for one minute, and B for ten seconds. If a speaker had to break their statement and resume, either with or without repeating words, then these were broken into separate statements. If they were interrupted, but continued on, then they were recorded as a single statement. Either way the “interruptor” was flagged as having interrupted another speaker - more on that below.

Longest Statement Time

If I got a full minute to speak, and so did you, but in the middle of your minute I interrupted with a ten second statement then I think that should be counted. In that example the total speaking times are 1:10 vs 1:00, but having a solid minute to form a more cogent argument is an advantage in a debate. Here is the maximum speaking time for each speaker.

Interruptions

Because there were so many interruptions I felt it was important to take a look at who was interrupting, and who was being interrupted.


Number of times Interrupted another speaker
Number of times interrupted another speaker (unwarranted)
Number of times interrupted by another speaker
Number of times interrupted by another speaker (unwarranted)

Word Counts

Total words spoken
Words spoken during an interruption


Words spoken during an unwarranted interruption

Calls to Silence Another Speaker

I did notice several instances where speakers specifically asked the moderator to silence another speaker.
Ella’s request was to deny, not entirely seriously, was for Connie, Kate, and Milo to all be ignored. The other eight of Connie and Kate’s requests/complaints were all have Milo silenced.

First Call to Silence

Att 4:39 by Connie requested that the moderator stop Milo from speaking. Up to that point here were the stats:


Speaking Time


Interruption Count


Word Count
So at that time her (Connie’s) side already had nearly twice the number of minutes and words as the opposing side and had made a dozen interruptions - with none by either Milo or Ella.

Summary

I think my impression of imbalance was probably formed at the 25% mark when it clearly was imbalanced, and when Connie and Kate started calling for Milo to be silenced (see prior section). After that Milo corrected that imbalance, but could only do so by adopting their tactics. It was very clear that Kate was responsible for a disproportionate number of interruptions. I wonder what the debate would have been like had Milo been more gentlemanlike in his manner (sorry I just watched Pride and Prejudice). I doubt he would have got a word in edgewise.

I think this is why those in the men's right's community like Milo. He's articulate, informed, and a fighter. I look forward to his future appearances.