In a 12 year period. So what about a 15 year period? Or a 20 year period? Or in a 12 year period but 10 years ago?
Results like that are generally bunk, or if they really cared, they would have made a broader statement (in these and those periods they performed like that).
I have only looked at the Credit Suisse paper that I think started this meme, and they clearly use a cut off point that makes women led companies look good. If they start their comparison a couple of years earlier, suddenly it looks different.
That's just standard methodology for companies like Credit Suisse whose main business is convincing people to buy stock. That same method is used to make stock market funds look good all the time (so that gullible people pay the 5% premium for the seemingly brilliant manager of that fund).
Other explanations of course could be: categories of businesses were trending that were more attractive to women (like digital services vs mining or something like that). Or the market in the time favored a risk averse approach and women tend to be more risk averse (don't know that) - there was a bad economical recession in that period of time (as they claim women "manage risk better" - well there are times when more or less risk taking are called for).
Don't get me wrong, I don't mind if women are CEOs and lead companies. I'm just wary of "narratives" that sound too good or too ideologically motivated.
1. A colleague of the author has commented in a lower thread that 12 years was the size of the data set available.
2. If n is very small in a several-year period, you wouldn't be able to do meaningful analyses on that period. How many women CEO's were there in the Fortune 1000 in the 80's? 90's? I'm going to guess epsilonically few.
3. Possible explanation that isn't hand-waving: selection bias, somewhat akin to the immigration effect. Because it's more challenging for women to get on a career path that eventually leads them to become a CEO, only the most determined/resourceful/hardworking will get there and be measured in the sample, whereas fewer men get weeded out. (e.g., IQ scores of H1B recipients from Asia is not representative of Asia as a continent, where IQ is a standin for whatever measurable trait you want here; and overall Asia and North America may well be quite comparable along most dimensions.)
So suppose they had done the study, and found no improved performance for women. Would they have published it? I guess the chance to get a result was roughly 50% (since stocks tend to go up or down, and the sample set of women-ceo-companies is probably small).
As to your explanation 3: that's just the kind of thing people want to believe who cherish such studies. At the same time there were also women who failed as CEOs (as the article said). Sure, it's a valid hypothesis, but there would be many ways to test it. For starters, there must already be studies correlating IQ to CEO success. What do they say?
Well I just looked at the first paper, and now look at the credit suisse paper I linked in another comment ( https://www.credit-suisse.com/newsletter/doc/gender_diversit... ) which has charts for 2005 to 2011. There was a "crash" in mid 2007 that somehow reversed the trends for companies with or without women on the board. Before that, zero-woman companies were outperforming the woman-companies.
Now it seems rather curious to me that they choose 2004-2008 for their comparison (in the first paper you linked to). So strangely, they catch exactly that crash point in their interval.
Certainly not. But they're common defaults. The worry is that if someone picks a non-standard interval, it suggests that they gave conscious thought to what interval to pick and came up with that one. The natural question in that case is why: Is there some economic reason that 12 years would be special, or some matching limitation to their data set? Or did they try every interval between 5 and 20 years and pick the one where the result was most favorable?
Also, why are they comparing companies ran by women in the Fortune 1000 against the S&P 500? Is it because the Fortune 1000 has the best performance by female CEOs while in the same time period the S&P 500 had the worst performance they could find?
Also, what percentage of companies in the S&P 500 where ran by women? I didn't see that anywhere.
The S&P 500 is the go-to index for a lot of people when they think about the "market return;" an important idea in finance. It's a portfolio of a large enough number of companies that the portfolio return converges pretty thoroughly to the market return. Not sure why the S&P 1000 isn't used as much, but using the S&P 500 is a convention.
Why choose all companies with female CEOs from the S&P 1000? The S&P 500 as of last year had 26 female CEOs, and if I had to guess, I'd say that number is higher than it would have been in 2002. The portfolio of "S&P 500 companies with female CEOs as of 2002" wouldn't have been diversified enough to reliably converge to the "market return of companies with female CEOs" that they were trying to investigate.
In a 12 year period. So what about a 15 year period? Or a 20 year period? Or in a 12 year period but 10 years ago?
This is a great point. Anytime you see a study where the time period seems contrived, it's usually because it was used to tweak the outcome to get what the author wanted.
Never invest in the stock market!!! (Source: S&P500 from Jan 02 to Jan 08)
One of the authors colleagues has posted in this thread stating that the selection of 12 years was because that was the amount of data available in the system they used.
I have only looked at the Credit Suisse paper that I think started this meme, and they clearly use a cut off point that makes women led companies look good. If they start their comparison a couple of years earlier, suddenly it looks different.
I'm not sure which one I looked at (thought I would have saved it to hd but can't find it right now). Googling I find for example this (it looks different than the one I remember, but still): https://www.credit-suisse.com/newsletter/doc/gender_diversit...
Only skimmed it, but they have for example a chart for 2005 to 2011. Between 2005 and mid 2007the zero-women companies outperform the at-least-one-women companies, then there is some sort of crash and the roles reverse. Would be curious to see the chart for pre-2005...
You can check out the full analysis and data set in a shared IPython notebook.
The study was done over a 12 year period because that's the full length of the pricing data available in the backtester Karen was using to do the study. It would be interesting to extend the analysis back in time as well.
NB: I work with Karen at Quantopian and we both have two X chromosomes :)
If you have a Bloomberg terminal, you can get more pricing.
Also, if you have a source of split-and-dividend-adjusted prices, you don't have to do a daily simulation, you can just simulate on the days when there is a buy or a sell.
Also, why does the leverage vary? Shouldn't it be 1.0 every day?
> ...finding that women CEOs in the Fortune 1000 drive three times the returns as S&P 500 enterprises run predominantly by men.
Apples to oranges. Why not compare female-CEO companies in the Fortune 1000 to male-CEO companies from the Fortune 1000? Why not include how male-CEO companies from the Fortune 1000 compare to the S&P 500? Why not compare female-CEO companies in the S&P 500 to male-CEO companies in the S&P 500 to the average company in the S&P 500?
Most importantly, why do people's critical thinking skills go right out the window when female positive news stories hit?
Lack of data. If I could find the historical lists of the S&P500 including CEO gender, or the entire list of Fortune 1000 companies by year, I'd look at any of the alternatives you suggest. The reality is getting the data is pretty tough, so we work with what we have. I've actually just gotten a dump of data from Morningstar that should allow me to expand the scope, but as with everything, there is work involved to make it usable....so stay tuned.
First, sorry but no excuses for releasing questionable studies that will spread like wildfire due to confirmation bias. Just because something is 'girl positive' doesn't mean we should all abandon our critical thinking skills. It's a small fraction of these feminist studies that go viral that can withstand any amount of scrutiny which makes the movement look worse.
I figured out how to get Fortune 1000 companies arbitrarily far back, and the S&P500 back to 2005. But the information is under copyright. I spent a long time thinking about the implications of helping you subvert copyright and decided that it's not something I'm going to do even though I'd like to see the study done better.
If you want the 2014 Fortune 1000 with contacts (i.e. CEO names) it costs $1799 from Fortune, the historical data from 1996-2013 costs $399/yr though it's unclear if that includes CEO names. (http://www.fortunedatastore.com/) Fortune's contact for historical data is female, she may be sympathetic to your cause. The S&P500 data might cost another $5-10k.
Since what you're doing is commercial your company has a responsibility to go through the proper channels to get your data. Really, the project is being used for advertising and brand-recognition. I also think your company has a responsibility to make sure you have the data you need to do proper analysis when you're representing the company.
Honestly I can't believe a company focused on quantitative market analysis doesn't already have this data!
Once you have CEO names it should be fairly straightforward to assign gender.
How about, instead of attackin the methodology, in some suspiciously misognistic approach to proving women cannot be the cause, we propose some hypothesis for why this might be true.
- women in business today are like female scientists of merely one or two generations ago. Only the brilliant / driven can overcome the obstacles, and brilliant driven people will outperform, and as CEO they drive the company to perform.
- companies that have sufficient internal mobility that a female CEO is able to be appointed, also are likely to have many other qualities that will allow the company to outperform.
Personally I prefer the second as an explanation for why the companies shown are doing better, and the first as an explanation of why those particular women are CEOs.
(I'm not a big fan of the superstar CEO theory)
Yes. We should take methodological criticism seriously.
If the methodology is flawed, why should effort be put into searching for hypotheses? If the methodology is flawed, then the results are not valid, and any hypotheses are meaningless as well.
The methodology is being attacked because the results are so spectacular that one assumes they are self-promotion/marketing by the hedge fund behind it. Otherwise why reveal such a profitable piece of data?
Free Money! Just buy the stock of all Wiltshire 5000 companies with a female CEO and we should be good. Who wants to subscribe to my index fund?
More seriously there's research on company performance before and after either Norway or Sweden put a quota of woman members on company boards. I have heard differing things on it, positive and negative and would be delighted if someone who wasn't skiving off work could add to the discussion by citing some of that research. In general female and male executives are dissimilar in their career trajectories. Women are much more likely to be company lifers for one. That's not going to be the only difference but it is plausible that there are more intervening factors than whether or not the CEO has XX chromosones.
Maybe CEOs of ill performing companies are more likely to be fired, and their replacement less likely to be a woman (since there are few candidates overall).
Just guessing - that's probably not it. But I think results like that are generally interesting as something seems to be going on. I just can't stand it if people jump to conclusions, usually to the ones that support their biases. It would be great to learn what is really going on.
Selection bias. Only the best and brightest get to run companies. Much easier for men to run companies. Which obviously says something bad about our society - but it does not mean that women run companies better than men.
Hey moderators, did this set off the flamewar detector? Because I don't think that's the article's fault, and people may well have things to say that aren't flamebait.
I wonder if this is more an effect that because a fewer number of women become leaders of the companies that the leaders they get are better then average.
Similarly to how during World War 2 when African American pilots were finally allowed to fly (The Tuskegee Airmen) they outperformed many other squadrons due to all the barriers put up ensured that they had the most dedicated, smartest and toughest pilots.
My concern is that he didn't calculate internal rate of return (IRR) correctly, and adjust for market conditions when the female CEO was hired/fired. It's hard to tell without verifying the details.
For example, suppose a company hired a female CEO during the market bottom in 2002 and fired her before the crash in 2008. Then, that would cause the "female CEO" statistic to be inflated relative to the S&P 500. If you calculate the relative IRR, it's more correct.
I.e., instead of a buy-and-hold S&P 500 investment vs buying/selling when the female CEO is hired/fired, you should instead buy $100k of the S&P 500 and female CEO stock SIMULTANEOUSLY, and then cash out both holdings when the female CEO is fired.
If you buy-and-hold the S&P 500 for 12 years vs buy/sell during a female CEO tenure, that's an apples-to-oranges comparison. That isn't properly adjusting for the market conditions when the female CEO was hired/fired.
He didn't provide enough details for me to check. I'd need a list of female CEO Tickers, hire/fire dates, and split-and-dividend-adjusted share price on the two dates.
There's a slightly different methodology here, but one consistent with what you're looking for.
On one line, buy-and-hold the S&P 500. Re-invest all dividends. You are 100% in the market at all times.
On the other line, buy-and-hold all companies run by female CEOs, weighted by the number of companies. Rebalance your portfolio every time a company is added or removed. You are 100% in the market at all times.
I think that if you look at the IPython notebook that the Fortune article refers to you can find the details spelled out in code.
(FWIW, the calculations are done by a she, not a he! It's my colleague Karen.)
"Now anyone who disagrees, has never actually heard this, or is bothered by opening with a sweeping prejudiced claim, has stopped reading. Let us proceed to strengthen the confirmation bias of those still reading..."
It really should be no surprise that these progressive companies have performed better than average. Society places higher expectations on women, so of course they must overperform!
Results like that are generally bunk, or if they really cared, they would have made a broader statement (in these and those periods they performed like that).
I have only looked at the Credit Suisse paper that I think started this meme, and they clearly use a cut off point that makes women led companies look good. If they start their comparison a couple of years earlier, suddenly it looks different.
That's just standard methodology for companies like Credit Suisse whose main business is convincing people to buy stock. That same method is used to make stock market funds look good all the time (so that gullible people pay the 5% premium for the seemingly brilliant manager of that fund).
Other explanations of course could be: categories of businesses were trending that were more attractive to women (like digital services vs mining or something like that). Or the market in the time favored a risk averse approach and women tend to be more risk averse (don't know that) - there was a bad economical recession in that period of time (as they claim women "manage risk better" - well there are times when more or less risk taking are called for).
Don't get me wrong, I don't mind if women are CEOs and lead companies. I'm just wary of "narratives" that sound too good or too ideologically motivated.