I am referring to an article in Money Sensethat appeared to suggest that investors aversion to going back into the market place was based on behavioural biases built up over the last 3 years, and that this was irrational.
What concerned me was the last paragraph:
“Investors should keep in mind the concept of reversion to the mean. This principle is based on the premise that stock market cycles and asset classes, over the long run, tend to return to their historical average (or mean). For example, the historical average for stock returns is in the range of 8% to 12%. A non-emotional approach to long-term decision making will help investors avoid mental mistakes, control impulsive, emotional judgments and earn a share of what the markets have to offer.”
Global developed markets have gone sideways since the late 1990s, enough time for a great many people to have retired and departed to the great whatever in the sky. Add in management fees and transaction costs, and you are not dealing with irrational aversion but a justified response to the academic and industry cheerleading for absurd return expectations at the end of the 1990s; these were expectations that were often spurred on by naïve beliefs in general equilibrium/efficient market theories. The same people who are castigating investors for not investing, through some misplaced behavioural bias, were pontificating on long term market returns of 8% to 12% at the end of the 1990s, blithely ignoring all the valuation excesses.
To take only a 3 year view is not only irrational and ignorant of investment risk, but absurd. So I guess the writer is saying everybody should get back on the roller coaster, because after 3 years markets have cleared and we are back to the type of equilibrium that supports mean returns of 8% to 12%.
But, it is not clear from the article as to whether the writer understands what he means by mean returns within a modern portfolio theory context:
Mean reversion within a modern portfolio theory context rests on a belief in a stable market equilibrium and a normal probability distribution of returns. If you average the distributions derived from say a Monte Carlo simulation, and you inputted a mean return of x%, you would get, more or less the mean return. What the writer does not say is that on any one probability distribution run, ex post returns could be below the mean, and that the mean return, even in equilibrium cannot be guaranteed. If you live 1,000 or more lifetimes, then on average, returns will revert to the mean, but the mean itself is unknown and will probably never ever be known.
Also, the equilibrium’s parameters, the return, the standard deviation and the correlation, are all dependent on the nature of the equilibrium – mean expected returns can shift and are not guaranteed to be in the 8% to 12% range even in a general equilibrium. This is mindless academic insanity!
The writer assumes, erroneously that a) the return relationship of the past will be repeated, b) that the one distribution of returns that we have experienced is representative of the mean of all distributions and c) that we are in a general equilibrium. All 3 assumptions are wrong!
We are not in a general equilibrium and for many years leading up to 2007 we have spent moving further and further away from equilibrium conditions. In other words the distribution of future returns were no longer a bell shaped normal, but warped and effectively heavily negatively skewed distribution.
Underlying all returns are earnings, and it is this fundamental, as opposed to valuation based returns, that is more likely to revert to a mean type figure. But, because we are more often than not out of equilibrium, in order to get back to the average return over time, we also have to correct on the downside, which means that as earnings readjust for periods of unsustainable excess that they are just as likely to move well below a certain period. You cannot do this in a general equilibrium context: there is no theoretical basis for undershooting the mean.
Earnings growth is dependent on the long term growth profiles of global economies, which may well be impaired for many reasons at this point in time. An 8% to 12% range is also a nominal figure which may not translate to current economic dynamics. Additionally, since earnings relative to GDP are also way above their historical averages (and I am not talking statistical means here), there are many reasons to believe, that at a more fundamental non equilibrium level, that we should be taking the opposite stance to that recommended. I am not saying that this is what people should do, just that the unwinding from a long period of excess has yet to work its way out of the system and could easily place downward pressure on real earnings growth for some time to come. Such an outcome does not compare well to an 8% to 12% expected mean return.
It would appear that behavioural economics is being used to support modern portfolio theory outcomes and dogma than what it really should be doing, which is to explain why markets and economies move outside of equilibrium relationships rendering ineffective the ordered systematic risk framework of modern portfolio theory.