5 ms·

I've done these same analyses before as well, and more. This is the correct analysis to do if you're looking to build an alpha-generative trading strategy. It i

by darawk 8y ago

I've done these same analyses before as well, and more. This is the correct analysis to do if you're looking to build an alpha-generative trading strategy. It is not necessarily the correct analysis to do if you want to measure actual CEO performance. If you want to measure CEO performance, you need to look at growth in fundamentals, not share price. The share price is going to reflect the market's beliefs about the CEOs. Which, if the market is efficient (as it seems to be for these variables), should show exactly no effect. If there were an effect here manifest in share price returns, we could all read this article and arbitrage it away. This would mean that the next person who looked for this effect would conclude "CEO performance has no correlation to share price returns!". I'm kind of surprised they would do all of this statistical work and not be aware of this gaping hole in their reasoning. Now they do make the point at the end that they're arguing against stock-based compensation here, which is a good point. If the market prices in your value as a CEO the moment you're hired, you don't want to have share-price based compensation. You want comp tied to growth in fundamentals.

lordnacho 8y ago
> You want comp tied to growth in fundamentals. This has the same problem as tying it to the market. You don't know what the baseline is. Would a good CEO always grow the fundamentals? What if it's a declining industry? And vice versa.
- darawk 8y ago
  It is a problem, but it's not the same problem. You can try to address the issue of secular trends in the industry by looking at say, performance relative to industry peers. How you define industry peers of course is a bit tricky. Another option is to look at share price returns from the day before the new CEO was announced. That would incorporate the market pricing in the new CEO. However, I don't think this analysis did that.
- starcraft 8y ago
  It's not the same problem because expectations of CEO outperformance can be internalised into the price at time 0.
zby 8y ago
That depends on how the effect is being measured - if the price growth is measured from the point where the CEO is appointed publicly - then you are right. But if they start from the moment before the announcement - then there would be no way to arbitrage the effect away. The interesting thing to do would be to measure the effect of the announcement itself.
- darawk 8y ago
  Ya, I agree. However, i'm pretty sure they're doing it from the moment of appointment, not the moment of announcement. They don't call it out specifically in the article, but i'm familiar with the datasets they're using, and I don't believe announcement time is in there. Though it's certainly possible that they have data I haven't seen.
pembrook 8y ago
Except, we've learned via a robust body of research over the last 5 decades that the market is largely efficient. Therefore, a change in share price over time (not measured in quarters, but over 3+ year periods...as was done in this study) is an excellent proxy for change in fundamentals.
- darawk 8y ago
  The market is going to price in expected change in fundamentals due to the new CEO. They're going to price it in immediately - before the fundamentals actually change. The market does not proxy change in fundamentals. It proxies change in expected future fundamentals. These are different things.
  - pembrook 8y ago
    We could be arguing semantics here, but my point is; the market will of course price in prevailing sentiment at the time, but this will be corrected over the next 3 year of earnings releases whether prospects for future fundamentals have actually improved (hence why the study used 3-6 year time periods). The price at year 1 will be based on the prospect of future earnings. The price at year 3 will still be based on the prospect of future earnings...but with 12 quarters of hard data allowing you to more accurately price in the new CEOs specific strategy and its effects on growth prospects.
    - darawk 8y ago
      Right. But any of that future price movement that was predictable based on the known attributes of the CEO at the time will be priced into the initial move. If it were not, then you would have a profitable trading strategy that say, went long Harvard CEOs and short Princeton CEOs (strictly a hypothetical example).
- throwawaymath 8y ago
  > Except, we've learned via a robust body of research over the last 5 decades that the market is largely efficient. Oh? Care to cite some of that research? And let me preempt a few common citations which don't actually work: 1. Warren Buffett's hedge fund bet does not prove this. Among other reasons, he made a bet concerning the performance of a fund of funds, not any particular outlier. 2. Eugene Fama's research on the Efficient Market Hypothesis (EMH) does not prove this (while we're at it, Fama would disagree with you that the market is efficient). His work applies the EMH as a means of studying efficiency, not asserting efficiency. One of the most common misconceptions of Fama's work is that it's descriptive instead of prescriptive. 3. When you actually do the math, the number of hedge funds that has ever existed is nowhere near what you'd need to explain the outliers like RenTech or Baupost. So the common coin flipping analogy actually doesn't prove it either. I can save you a trip to look for citations - the "robust" body of research you're talking about doesn't exist. There is no such consensus. The efficient market is a model which only approximates real world conditions - it do so imperfectly, and in the best case it only works locally. Here is an article [1] written by Cliff Asness on the topic of the EMH which might be educational. Asness earned his PhD under Fama and founded AQR, one of the more successful hedge funds. He also regularly comments on, and published, research. There remains a great feal of inefficiency at the macro level, even on horizons measured in years. ______________________ 1. https://www.institutionalinvestor.com/article/b14zbgrj5pflsc/the-great-divide-over-market-efficiency https://www.institutionalinvestor.com/article/b14zbgrj5pflsc...
yborg 8y ago
Did you read the article? The objective was to prove or disprove the Jensen thesis that CEO compensation should be tied to gains in stock price because CEOs affect stock price. They were NOT attempting to answer the question of whether or not CEO quality and compensation affects the success of their underlying businesses.
- darawk 8y ago
  Did you? They were attempting to do both. I specifically called out that it was a legitimate way to do analyze the compensation issue.
jiveturkey 8y ago
Thank you. I also wish they would have measured startup (ie, private) CEO performance. Of course that's quite hard because financials are private and all the failed companies have no records left, and if they did, good luck teasing the data out.

by darawk 8y ago

lordnacho 8y ago

> You want comp tied to growth in fundamentals. This has the same problem as tying it to the market. You don't know what the baseline is. Would a good CEO always grow the fundamentals? What if it's a declining industry? And vice versa.

darawk 8y ago
It is a problem, but it's not the same problem. You can try to address the issue of secular trends in the industry by looking at say, performance relative to industry peers. How you define industry peers of course is a bit tricky. Another option is to look at share price returns from the day before the new CEO was announced. That would incorporate the market pricing in the new CEO. However, I don't think this analysis did that.
starcraft 8y ago
It's not the same problem because expectations of CEO outperformance can be internalised into the price at time 0.

zby 8y ago

That depends on how the effect is being measured - if the price growth is measured from the point where the CEO is appointed publicly - then you are right. But if they start from the moment before the announcement - then there would be no way to arbitrage the effect away. The interesting thing to do would be to measure the effect of the announcement itself.

darawk 8y ago
Ya, I agree. However, i'm pretty sure they're doing it from the moment of appointment, not the moment of announcement. They don't call it out specifically in the article, but i'm familiar with the datasets they're using, and I don't believe announcement time is in there. Though it's certainly possible that they have data I haven't seen.

pembrook 8y ago

Except, we've learned via a robust body of research over the last 5 decades that the market is largely efficient. Therefore, a change in share price over time (not measured in quarters, but over 3+ year periods...as was done in this study) is an excellent proxy for change in fundamentals.

darawk 8y ago
The market is going to price in expected change in fundamentals due to the new CEO. They're going to price it in immediately - before the fundamentals actually change. The market does not proxy change in fundamentals. It proxies change in expected future fundamentals. These are different things.
- pembrook 8y ago
  We could be arguing semantics here, but my point is; the market will of course price in prevailing sentiment at the time, but this will be corrected over the next 3 year of earnings releases whether prospects for future fundamentals have actually improved (hence why the study used 3-6 year time periods). The price at year 1 will be based on the prospect of future earnings. The price at year 3 will still be based on the prospect of future earnings...but with 12 quarters of hard data allowing you to more accurately price in the new CEOs specific strategy and its effects on growth prospects.
  - darawk 8y ago
    Right. But any of that future price movement that was predictable based on the known attributes of the CEO at the time will be priced into the initial move. If it were not, then you would have a profitable trading strategy that say, went long Harvard CEOs and short Princeton CEOs (strictly a hypothetical example).
throwawaymath 8y ago
> Except, we've learned via a robust body of research over the last 5 decades that the market is largely efficient. Oh? Care to cite some of that research? And let me preempt a few common citations which don't actually work: 1. Warren Buffett's hedge fund bet does not prove this. Among other reasons, he made a bet concerning the performance of a fund of funds, not any particular outlier. 2. Eugene Fama's research on the Efficient Market Hypothesis (EMH) does not prove this (while we're at it, Fama would disagree with you that the market is efficient). His work applies the EMH as a means of studying efficiency, not asserting efficiency. One of the most common misconceptions of Fama's work is that it's descriptive instead of prescriptive. 3. When you actually do the math, the number of hedge funds that has ever existed is nowhere near what you'd need to explain the outliers like RenTech or Baupost. So the common coin flipping analogy actually doesn't prove it either. I can save you a trip to look for citations - the "robust" body of research you're talking about doesn't exist. There is no such consensus. The efficient market is a model which only approximates real world conditions - it do so imperfectly, and in the best case it only works locally. Here is an article [1] written by Cliff Asness on the topic of the EMH which might be educational. Asness earned his PhD under Fama and founded AQR, one of the more successful hedge funds. He also regularly comments on, and published, research. There remains a great feal of inefficiency at the macro level, even on horizons measured in years. ______________________ 1. https://www.institutionalinvestor.com/article/b14zbgrj5pflsc/the-great-divide-over-market-efficiency https://www.institutionalinvestor.com/article/b14zbgrj5pflsc...

yborg 8y ago

Did you read the article? The objective was to prove or disprove the Jensen thesis that CEO compensation should be tied to gains in stock price because CEOs affect stock price. They were NOT attempting to answer the question of whether or not CEO quality and compensation affects the success of their underlying businesses.

darawk 8y ago
Did you? They were attempting to do both. I specifically called out that it was a legitimate way to do analyze the compensation issue.

jiveturkey 8y ago

Thank you. I also wish they would have measured startup (ie, private) CEO performance. Of course that's quite hard because financials are private and all the failed companies have no records left, and if they did, good luck teasing the data out.