Hi Joachim, I wonder if weighting contributes to the situation? A stock which is over valued will have a higher market capitalisation, and will have a higher number of samples contributing to the index. Meanwhile, companies trading below fair value will be under- counted (less of their stock will be included). The collapse of a bubble sees the index suddenly re-weighted from the frothy sector toward the broader economy.
That is an interesting thought and I haven't really taken that into account so far. But if higher-valued stocks have a return distribution that becomes increasingly skewed to the downside due to the overvaluation then that would add towards a drift back to fair value.
So, I don't know whether the valuation effect is in practice big enough to materially increase the likelihood of ending at fair value. I have my doubts, but can't prove it.
I would think that if each stock is 'independently' being reviewed and 'objectively' be valued that the value should 'independent' and the errors generally cancel out so the the overall valuation of the index used, say S&P 500, should follow a normal distribution and be more likely than not close to the 'fair' or, maybe better, 'intrinsic' value at a given point in time. This does not preclude material over- or under-valuations but does follow that, all things considered, the market is 'fairly valued' on average.
The problem of your argument lies with the word “independently” that is absolutely not the case as all the research on behavioral finance and information cascades shows. Just look at my recent post on the impact of tweets by CEOs vs companies. In fact I guess that on average about once every other week I write some phenomenon that violates the concept of investors making up their minds independently. And once that is violated that means that individual errors do not cancel each other out.
Ingeniuos article. Thanks for the input. Can‘t wait until I read part II.
Hi Joachim, I wonder if weighting contributes to the situation? A stock which is over valued will have a higher market capitalisation, and will have a higher number of samples contributing to the index. Meanwhile, companies trading below fair value will be under- counted (less of their stock will be included). The collapse of a bubble sees the index suddenly re-weighted from the frothy sector toward the broader economy.
That is an interesting thought and I haven't really taken that into account so far. But if higher-valued stocks have a return distribution that becomes increasingly skewed to the downside due to the overvaluation then that would add towards a drift back to fair value.
The problem just is that mean reversion effects have become so much weaker in the last two decades as I have described for example here: https://klementoninvesting.substack.com/p/short-termism-and-momentum-investing
So, I don't know whether the valuation effect is in practice big enough to materially increase the likelihood of ending at fair value. I have my doubts, but can't prove it.
I would think that if each stock is 'independently' being reviewed and 'objectively' be valued that the value should 'independent' and the errors generally cancel out so the the overall valuation of the index used, say S&P 500, should follow a normal distribution and be more likely than not close to the 'fair' or, maybe better, 'intrinsic' value at a given point in time. This does not preclude material over- or under-valuations but does follow that, all things considered, the market is 'fairly valued' on average.
The problem of your argument lies with the word “independently” that is absolutely not the case as all the research on behavioral finance and information cascades shows. Just look at my recent post on the impact of tweets by CEOs vs companies. In fact I guess that on average about once every other week I write some phenomenon that violates the concept of investors making up their minds independently. And once that is violated that means that individual errors do not cancel each other out.