Throughout a classic education in finance and economics, one core concept that became an assumption of my daily life was that volatility was risk. To a certain degree this intuitively makes sense: if I'm buying an asset, and I have little sense what its value is going to be in an hour or two days from now, I should buckle up because it's going to be one hell of a ride.
In a relatively preset environment, volatility would certainly approximate risk. For example, if we were to live on Mount Kenya as a subsistence farmer, where mean daily fluctuations of temperature equate to 11.5 degrees Celsius (20.7 degrees Fahrenheit), this could surely be seen as one approximation of the risk of being able to generate a successful crop, although another might be seasonal weather fluctuation, such as that seen over a year. Still another would be fluctuation in weather over several years. Thus these three time spans (daily, seasonal and yearly) of weather fluctuation could probably closely approximate your risk of survival (of course there would be others: see bears).
So why might a security's volatility not closely approximate its risk? One answer would be the failure to translate observed volatility to actual volatility, although we would certainly run in to this problem when observing weather. The other problem would come with what this volatility actually represents. In the case of weather, it is the state of our atmosphere, something far beyond our control (although the aggregation of humans is doing a pretty good job: see climate change)
Is An Individual Investor Equivalent to a Subsistence Farmer?
In an asset market, for volatility to approximate "risk of survival" like it does in the natural world, it would have to retain some key characteristics:
- Unchangeable by one or several individuals
- Relatively constant over time
With the first item, we find little support that a small subsection of the population cannot move prices and ergo influence volatility. By its nature, asset markets are a wealth adjusted voting machine, so were one to be so inclined, with the proper amount of money one could dramatically influence the price environment of one particular asset. This further topples the second characteristic as big money moving in and out of trades can dramatically affect the volatility of individual assets.
Because volatility is easily manipulated and not an inherent characteristic of an asset, its value as an accurate predictor of risk is severely diminished if not totally obliterated. Thus, while volatility might be high for an asset currently, there is no inherent dynamic that suggests that it might or should stay at this level for any amount of time going forward.
The next article in this series will look at sticky volatility induced by selection bias followed by an article exploring problems with the risk/return correlation.
As another interesting exercise, what if you look at the volatility of an asset's price over time? There's evidence that stock prices, and even GDP, might have unit roots in them, meaning that their variances goes to infinity over time!
ReplyDeleteThus, if you're using the current variance in an asset as an approximation for risk, and the variance is actually increasing over time, you would not only be wrong in your current assessment of the risk, but you would probably also be wrong in using the variance as a proxy for risk at all.
This brings to light another interesting question. What is the risk of an asset following a unit root process, and consequently with infinite variance? And, if stock prices, like GDP, do have unit roots, are we setting ourselves up to fail with our current approach to finance?
Hey Vulcidian,
ReplyDeleteThat's a really interesting idea. Do you have any links to scholarly papers or any other further reading on that topic?
To learn more about unit roots in time series, the best place to start is the classic wikipedia (http://en.wikipedia.org/wiki/Unit_root).
ReplyDeleteHere's an example of a short article related to random walks and stock prices (http://www.chicagobooth.edu/faculty/selectedpapers/sp16.pdf).
Enjoy!