Back to Community
Investing in Female CEOs - sector neutral, a different benchmark and new data

Months ago, I shared my side project with the community. I've continued looking at investing in Female CEOs and analyzing the data in different ways. I wanted to share some of the different ways in which I have looked at the data.

This notebook includes the first version of my algorithm, but it also includes a sector weighted approach. There was a lot of feedback that I might have a large sector bias in the data. It turns out that 30% of the female CEOs in my data set were in the consumer space. I created the sector weighted approach to understand how much of an impact that had on the overall performance.

The female CEO data used to run this notebook is available here. To use it, clone the notebook and then load the file titled FemaleCEOs_v6.csv into the data folder of your research environment.

Loading notebook preview...
Notebook previews are currently unavailable.
Disclaimer

The material on this website is provided for informational purposes only and does not constitute an offer to sell, a solicitation to buy, or a recommendation or endorsement for any security or strategy, nor does it constitute an offer to provide investment advisory services by Quantopian. In addition, the material offers no opinion with respect to the suitability of any security or specific investment. No information contained herein should be regarded as a suggestion to engage in or refrain from any investment-related course of action as none of Quantopian nor any of its affiliates is undertaking to provide investment advice, act as an adviser to any plan or entity subject to the Employee Retirement Income Security Act of 1974, as amended, individual retirement account or individual retirement annuity, or give advice in a fiduciary capacity with respect to the materials presented herein. If you are an individual retirement or other investor, contact your financial advisor or other fiduciary unrelated to Quantopian about whether any given investment idea, strategy, product or service described herein may be appropriate for your circumstances. All investments involve risk, including loss of principal. Quantopian makes no guarantees as to the accuracy or completeness of the views expressed in the website. The views are subject to change, and may have become unreliable for various reasons, including changes in market conditions or economic circumstances.

22 responses

Another piece of feedback that I got, was that the S&P500 was the wrong benchmark to use. A better one would be all the companies of the Fortune 1000, but unfortunately getting the historical constituents was too pricey for me.

Using a new universe ranking and filtering API (which isn't available to everyone yet, but which will be soon!) I was able to create a version of the Fortune 500, which I have nicknamed the Quanto500. Since the Fortune 500 is the top companies in the US by revenue, I used the new API sort the entire universe by revenue and then select the top 500 to invest in. I re-rank and select the stocks each month and then rebalance the portfolio.

Loading notebook preview...
Notebook previews are currently unavailable.

With the addition of the new Quantopian Store, I'm also able to access CEO gender in the EventVestor CEO Changes dataset. With the help of James Christopher, we analyzed this data in a variety of ways over the summer. Just a few of the different algorithms are shared here.

This dataset is infinitely more interesting than the initial one I built. It not only includes all female CEOs changes, but also all male CEO changes, as well as the announcement date, the trade date and the effective date of the change. There are many more things to explore using this data set.

Loading notebook preview...
Notebook previews are currently unavailable.

Thank you Karen for sharing your notebook. Not only does it highlight an important economic fact - women are better managers of large corporations than men, but the extensive notebook is a great template for anyone looking to do research using the Quantopian platform.

I do have a couple of observations that I would share.

First, you compare an equal weighted portfolio, women-led companies where equal dollars are invested in each and rebalanced at random times (newly ascended women CEOs), to a capitalization-weighted portfolio, SPY. Perhaps you might consider using an equal weighted SP500 ETF like RSP which has outperformed the SPY by 60% since inception since 2004. That is a fairer comparison. Generally, equal weighted portfolios have a small-cap bias.

Some of the charts show discontinuities, e.g. Female CEOs in green around 2008 above.

Finally, could you comment on the difference between your simulation which outperforms the SPY by a factor of 3X and the actual performance of funds that implemented similar strategies (PXWIX and WIL) where they have each significantly underpeformed the S&P 500.

Thank you for your work and go girl power!

Looking at my algo vs RSP is a great suggestion. People are often saying "use another benchmark" but I haven't heard of this one. I'll be sure to look at it in the future.

As for the differences in PXWIX and WIL, PXWIX doesn't share a ton of detail about how they select which companies to invest in, but one major difference is they are a global fund, not just limited to US equities.

WIL tracks a weighted index of 85 U.S.-based companies that are listed on the NYSE or NASDAQ, have market capitalizations of at least $250 million, and have a woman CEO or a board of directors that’s at least 25% female. What I think is most interesting about this is the board of directors piece. I have not looked at board participation myself at all (I don't have the data). I did talk to a team of MBA students at Sloan who spent the semester researching this. They were unsuccessful at finding a link between performance and board participation.

There was an interesting article in Bloomberg on this topic recently that showed a few other funds and indicated the returns are a bit all over the place.

Karen,

The 'Equal weight benchmark' is a good suggestion. You can also try to look directly at several comparison groups:
a) Companies that replace women CEO's with male CEO's. Do they then show worse performance?
b) Companies that replace male CEO's with female CEO's. Do they show a bump?
c) 'Matching peer companies.' Try to find a universe of similar companies (there are only 80 female CEO's here?). Matched by industry and EBIT margin and trailing period stock price momentum. Do the female led companies outperform a custom universe of equal weighted similar 'companies' after the females are named to leadership positions?
d) Fundamental analysis. Don't look just at stock performance. Look at 1,3 and 5 year sales growth and EBIT and Gross Margin levels. Are female led companies outgrowing and/or improving bottom line margins more then 'matching companies' without females?
e) Introduce a 'random range' for an 80 stock buy-and-hold universe. An 80 stock universe is small and will have a huge range of 'random drift' that is possible over time. So, include probability estimates for 'chance outcomes' to see how unlikely the observed result is after controlling for a) market cap and b) sector and c) industry and d) company type (sales growth rates and margin rates).

It would be very interesting (and socially important) if you could show that women are better CEO's. But, I would keep digging on the 'comparison' group and the controls to eliminate all other variables as much as possible.

Might also be interesting to look at 'sentiment' and SMA30 and SMA100 sentiment over time after a female takes over. Does avg. sentiments go up? Fall? Stay the same? Does it stay up over time? Or to see if you can untangle what types of companies women take over on average (hi growth or struggling)... and how the companies change. i.e. does employee turnover fall or rise? Does % of sales to marketing fall or rise? Along with margins and revenues, might uncover some interesting stats in here. But, I would still be careful about linking these too much to 'gender.' There is a way too small pool of female CEO's to really look at. They may share underlying personality traits apart from gender, such as drive and motivation that are much higher then the 'averages' for men because of the challenges of becoming a female CEO in the first place. But, will be very interested to read what you find.

I finally got a chance to pull together an algo buying RSP and rebalancing monthly to use as a benchmark.

Here are the returns of it vs. my algo vs. the SPY.

Loading notebook preview...
Notebook previews are currently unavailable.

Karen - it really looks like your quanto-500/quanto-1000 is getting hit by a data error in 2005/2006 - equal-weighted indices are unlikely to drop 50% in a week.

EDIT: I suspect that would change those results quite a lot

Good point! I had wondered about that drop, but hadn't dug into it. Will make sure to do so.

Here is a revised version of the Quanto 500 graph above, you can see v1 and v2 of the Quanto 500 and Simon is correct, there was some kind of issue in the v1 version of it.

Interestingly, when I did a code compare of these two algos, the code is actually quite different. This work was done while the API was in development, and there was a significantly modification to the API along the way. V1 was done during the first version of the API, and v2 was done during the second. Additionally, in the time span between these two backtests runs (which were run on 9/28 and 10/5 respectively) we also made some significant changes to how dividends were being handled by the API.

I don't actually know which specific change impacted this algo.

Loading notebook preview...
Notebook previews are currently unavailable.

This is the algorithm for the Quanto500_v2 in the chart above. Interestingly there is still a blip around July of 2005. It's much smaller, but you can see it in the leverage chart.

I'm hesitant to share the code for the Quanto500_v1 algo, because it will no longer run and I don't want anyone to get confused and use it as an example.

Clone Algorithm
49
Loading...
Backtest from to with initial capital
Total Returns
--
Alpha
--
Beta
--
Sharpe
--
Sortino
--
Max Drawdown
--
Benchmark Returns
--
Volatility
--
Returns 1 Month 3 Month 6 Month 12 Month
Alpha 1 Month 3 Month 6 Month 12 Month
Beta 1 Month 3 Month 6 Month 12 Month
Sharpe 1 Month 3 Month 6 Month 12 Month
Sortino 1 Month 3 Month 6 Month 12 Month
Volatility 1 Month 3 Month 6 Month 12 Month
Max Drawdown 1 Month 3 Month 6 Month 12 Month
# Backtest ID: 561328a76b7a1110a6d57c81
There was a runtime error.

Interesting! Eyeballing it, the period of (female CEO) outperformance is concentrated in 2004-2007, particularly that dramatic spurt in 2004.

Hi Karen,

Thank you for investigating an equal-weighted benchmark for your equal-weighted strategy. As your research shows, much of the performance difference between female led companies and the "market" disappears. I am sure that if you also considered the sector weights between your strategy and the benchmark, any remaining out-performance would likewise narrow or even flip.

The economic rationale, that female led companies have an advantage over male led companies seems flawed, to me. Imagine if the thesis was flipped, that male dominated companies had an economic advantage. I think it would take a lot of convincing to prove this assertion.

IMHO, the funds that market this as an investment theme are doing a disservice to their investors. Sure, investing in female led companies is a valid social investing meme, but to make claims of enhanced performance where size, sector and other biases explain the tracking error with benchmark returns strikes me as marketing over reach.

Here is the tear sheet of this strategy using pyfolio.

Loading notebook preview...
Notebook previews are currently unavailable.
Disclaimer

The material on this website is provided for informational purposes only and does not constitute an offer to sell, a solicitation to buy, or a recommendation or endorsement for any security or strategy, nor does it constitute an offer to provide investment advisory services by Quantopian. In addition, the material offers no opinion with respect to the suitability of any security or specific investment. No information contained herein should be regarded as a suggestion to engage in or refrain from any investment-related course of action as none of Quantopian nor any of its affiliates is undertaking to provide investment advice, act as an adviser to any plan or entity subject to the Employee Retirement Income Security Act of 1974, as amended, individual retirement account or individual retirement annuity, or give advice in a fiduciary capacity with respect to the materials presented herein. If you are an individual retirement or other investor, contact your financial advisor or other fiduciary unrelated to Quantopian about whether any given investment idea, strategy, product or service described herein may be appropriate for your circumstances. All investments involve risk, including loss of principal. Quantopian makes no guarantees as to the accuracy or completeness of the views expressed in the website. The views are subject to change, and may have become unreliable for various reasons, including changes in market conditions or economic circumstances.

I've continued working on this strategy recently. Attached is my latest version, which uses the Eventvestor CEO change data set and the pipeline API to create my own version of the Fortune 1000 and the CEO change data to identify and invest in female CEOs when they start in the position.

I presented on this version at Pycon 2016. This version runs until 5/1/2016. If you are using the free Eventvestor data, change your backtest end date to 5/29/2014.

Clone Algorithm
12
Loading...
Backtest from to with initial capital
Total Returns
--
Alpha
--
Beta
--
Sharpe
--
Sortino
--
Max Drawdown
--
Benchmark Returns
--
Volatility
--
Returns 1 Month 3 Month 6 Month 12 Month
Alpha 1 Month 3 Month 6 Month 12 Month
Beta 1 Month 3 Month 6 Month 12 Month
Sharpe 1 Month 3 Month 6 Month 12 Month
Sortino 1 Month 3 Month 6 Month 12 Month
Volatility 1 Month 3 Month 6 Month 12 Month
Max Drawdown 1 Month 3 Month 6 Month 12 Month
# Backtest ID: 57484e64b2778f0f7d561ed9
There was a runtime error.

This is an updated version of my initial algorithm. It included data from 2002 through May 2016 for all the female CEOs of the Fortune 1000, as given to me by Catalyst.

In the code of the algo, you can see the shares spreadsheet of the data I have used.

Clone Algorithm
18
Loading...
Backtest from to with initial capital
Total Returns
--
Alpha
--
Beta
--
Sharpe
--
Sortino
--
Max Drawdown
--
Benchmark Returns
--
Volatility
--
Returns 1 Month 3 Month 6 Month 12 Month
Alpha 1 Month 3 Month 6 Month 12 Month
Beta 1 Month 3 Month 6 Month 12 Month
Sharpe 1 Month 3 Month 6 Month 12 Month
Sortino 1 Month 3 Month 6 Month 12 Month
Volatility 1 Month 3 Month 6 Month 12 Month
Max Drawdown 1 Month 3 Month 6 Month 12 Month
# Backtest ID: 574cb8edfebbb40f8a21dba7
There was a runtime error.

Karen,

Thanks for the update. Sure seems like the differences between female and male risk aversion is something that should continue to be studied. Googled around for something to expand upon this thought, and found the following:

http://cdp.sagepub.com/content/21/1/36

Tried to implement in a few lines of code per attached. I figured this would be a good way to start tooling around with this concept. Thanks again for the data!

Clone Algorithm
5
Loading...
Backtest from to with initial capital
Total Returns
--
Alpha
--
Beta
--
Sharpe
--
Sortino
--
Max Drawdown
--
Benchmark Returns
--
Volatility
--
Returns 1 Month 3 Month 6 Month 12 Month
Alpha 1 Month 3 Month 6 Month 12 Month
Beta 1 Month 3 Month 6 Month 12 Month
Sharpe 1 Month 3 Month 6 Month 12 Month
Sortino 1 Month 3 Month 6 Month 12 Month
Volatility 1 Month 3 Month 6 Month 12 Month
Max Drawdown 1 Month 3 Month 6 Month 12 Month
# Backtest ID: 5752d8de14e9b50f8d39ca1a
There was a runtime error.

This is very cool. Interesting area of research to follow. A question about the data. How many female CEO's have there been? Were there enough to get any kind of statistical confidence in the results? Also, have you checked to see if the results were driven by one or two particularly successful or unsuccessful companies?

Tar,

Below is a mod to Karen's algo that displays the # of female led companies owned during the time series. The numbers start around 13 in 2004, and reaches about 50. Probably too small a sample to get statistical confidence. More interesting though is the steady and stable increase in the quantity over time. The question I contemplate is how this increase is being driven. Probably a combination of out performance of the CEOs, and a lowering of the historical barriers to entry. The steady incline makes a case for out performance. Interested to see what returns will look like when female CEOs constitute ~15% of the index.

Big losers include: Rite Aid, Ann Taylor, and The Phoenix Companies and are all during the malarkey of 08-09
Big Winners are not very concentrated in terms of out sized gains or timing. Kraft and AMD are on the list.

  • the algo with the visual for quantity of female CEOs is below
Clone Algorithm
9
Loading...
Backtest from to with initial capital
Total Returns
--
Alpha
--
Beta
--
Sharpe
--
Sortino
--
Max Drawdown
--
Benchmark Returns
--
Volatility
--
Returns 1 Month 3 Month 6 Month 12 Month
Alpha 1 Month 3 Month 6 Month 12 Month
Beta 1 Month 3 Month 6 Month 12 Month
Sharpe 1 Month 3 Month 6 Month 12 Month
Sortino 1 Month 3 Month 6 Month 12 Month
Volatility 1 Month 3 Month 6 Month 12 Month
Max Drawdown 1 Month 3 Month 6 Month 12 Month
# Backtest ID: 57530feca8e8b20f86412f15
There was a runtime error.

Hi Tar,
I cover the number of CEOs and looked into some individual companies in my presentation here - https://www.youtube.com/watch?v=ll6Tq-wTXXw

There are 80 female CEOs at 74 companies from 2002 - 2014. I think there was one more added between 2014 and May of 2016.

I took a look previously at removing the outliers (the top 3 and the bottom 3 performers). This effected the returns, but didn't remove the outperformance.

I think this could be turned into an interesting market neutral strategy

1) go long the 10 companies with the most recent male to female ceo announcement
2) go short the 10 companies with the most recent female to male ceo announcement
3) positions are kept to approx 5% of portfolio value