Risk and Return Within the Stock Market: What Works Best?

From Risk and Return Within the Stock Market: What Works Best? By Roger G. Ibbotson, Ph.D. and Daniel Y.-J. Kim, Ph.D:

Contrary to theory, low beta and low volatility portfolios outperform high beta and high volatility portfolios……. Overall the best returning characteristics are high earning/price, high book to market, and low turnover. On risk adjusted basis, the best performances were low beta, low volatility, and low turnover……Within this anomaly, a common theme emerges. Whether it be through factors that encode popularity among investors (turnover, growth), academic popularity (citations), or popularity caused by leverage aversion (beta, volatility), popularity underperforms.

Well what a surprise! Demand dynamics impact pricing of assets and hence returns. Who would have guessed.  This is something that value biased investors have known for a while.  From my Capitalism In Crisis 3 document:

Mean variance optimizers look at historic risk/return relationships. As noted in this analysis, risk and return for asset classes are impacted by demand and supply. Many of these linear allocation tools hold the risk and return relationships constant while determining optimum allocations to these assets/securities. This is clearly flawed, since only small changes in net demand need significantly impact price/risk and return; higher allocations to an asset not only increases its price and reduces its return per unit of risk but, also reduces the price of other assets and increase their prospective return per unit of risk. You cannot hold risk/return relationships constant when asset allocation/demand are being adjusted.

The price of a popular investment will be bid up by new money assets coming into the market portfolio and from a switch in demand from poorer performing assets in the market portfolio.   The poorer price performing assets will have lost liquidity, will tend to be held by longer term investors and will be bought by those few that see a lower price with stable yet boring fundamentals as a value proposition.    

The better performing assets will see larger upward price swings on the way up as demand crowds supply and larger downward price swings on the way down as supply crowds demand.  In other words it is the ratio of “short term risk sensitive/price focussed investors” to long term return focussed investors unconcerned over short term volatility that defines a significant component of price volatility.   The more popular stocks are going to be more easily sold and will be in greater demand and hence will have more liquidity.   Positive demand impacts price positively and return negatively and vice versa for weak demand dynamics.    Bid and ask prices will also be impacted by demand factors, with popular stocks costing more to buyers and less popular stocks costing less to buyers, depending on the demand dynamics.

High risk should equal higher return, but not all perceived risk can be incorporated in volatility, especially in an out of equilibrium world with irrational investors.  Higher risk can also be the prospect of lower returns: in other words a perception over performance risk given that investors prefer more return to less return and greater certainty of return to less certainty of return (risk seeking when assets are rising and risk avoiding when assets are falling).  A popular stock may well be seen as providing greater certainty of higher return and a less popular stock less so.  

MPT’s belief that higher volatility risk should equal higher returns is grounded in the belief that risk is defined by the sensitivity of assets to new random information, that market and economic relationships are in balance, and that assets are priced correctly, and that by virtue of this, expected volatility, which is assumed to incorporate information over the probability distribution of random price sensitive information, should be a predictor of return.   

The above is of course not the case and MPT allocators, MVO’s, not only ignore market dynamics of return as noted above, but also out of equilibrium dynamics.  

I do of course have concerns over the way in which the data set has been used (1971 to 2012), in that we are not looking at multi period returns to assess other factors that may have been involved.  I suspect that there are fairly long periods of time that low volatility assets have under performed, and the start and end dates may be impacted by the extent to which low and high volatility/low and high liquidity assets are relatively priced.  For example, If we started with a pricing environment that significantly under priced low volatility stocks and ended with a pricing environment where low volatility stocks were more richly priced on a relative basis the risk premiums on low liquidity, low beta will be higher than they might turn out to be going forward.  The ability to look at the entire data set and the assumptions, especially with respect to transaction costs, would have been useful.

See also:

Liquidity as an Investment Style By Roger G. Ibbotson, Ph.D., Zhiwu Chen, Ph.D., Daniel Y.-J. Kim, Ph.D., and Wendy Y. Hu, Ph.D.

Leave a Reply