The first in the section "Do MBAs Make Better CEOs?". If the labour market is efficient we would NOT expect CEOs with MBAs to outperform CEOs without MBAs due to selection effects. We may expect them to outperform if we sample randomly from the general population that has MBAs versus that which does not, but this is not the sampling scheme in effect. If boards are able to accurately select non-MBA CEOs whose other characteristics compensate for the lack of an MBA, then we would expect a zero within-group correlation in the selected population. It is like if I was to hire a quantitative researcher for a machine learning research team. I could pick a PhD from a top school, or I could allow a non-PhD into my team as long as their other characteristics are sufficiently great (e.g. competition math in high school), but after appropriate selection effects there is no within-group correlation between holding a PhD and not holding one. If I looked at the average PhD versus the average non-PhD in the general population though, then an effect becomes apparent, but that's not the sampling scheme I've used.
The second problem is in the section "Is CEO Performance Persistent?". They used stock market returns which is flawed as expected CEO performance should be baked into the stock price from the beginning, meaning we expect zero excess return in the second sample even if the CEO was truly exceptional. This section would unlikely pass peer review into a top finance journal.
The only valid thing I can see is research about "Share Price Performance for CEOs Who Ran Multiple Companies", since nobody knows the second company ahead of time so the information shouldn't be baked into the price. But a lot of details are missing. Skewness and kurtosis of returns will impact the way they've quantized the data, among other things.
The first in the section "Do MBAs Make Better CEOs?". If the labour market is efficient we would NOT expect CEOs with MBAs to outperform CEOs without MBAs due to selection effects. We may expect them to outperform if we sample randomly from the general population that has MBAs versus that which does not, but this is not the sampling scheme in effect. If boards are able to accurately select non-MBA CEOs whose other characteristics compensate for the lack of an MBA, then we would expect a zero within-group correlation in the selected population. It is like if I was to hire a quantitative researcher for a machine learning research team. I could pick a PhD from a top school, or I could allow a non-PhD into my team as long as their other characteristics are sufficiently great (e.g. competition math in high school), but after appropriate selection effects there is no within-group correlation between holding a PhD and not holding one. If I looked at the average PhD versus the average non-PhD in the general population though, then an effect becomes apparent, but that's not the sampling scheme I've used.
The second problem is in the section "Is CEO Performance Persistent?". They used stock market returns which is flawed as expected CEO performance should be baked into the stock price from the beginning, meaning we expect zero excess return in the second sample even if the CEO was truly exceptional. This section would unlikely pass peer review into a top finance journal.
The only valid thing I can see is research about "Share Price Performance for CEOs Who Ran Multiple Companies", since nobody knows the second company ahead of time so the information shouldn't be baked into the price. But a lot of details are missing. Skewness and kurtosis of returns will impact the way they've quantized the data, among other things.
All in all, unpersuasive.