[This is #3 in a series of Meditations on Risk.]
In the investment world standard deviation = risk.
This idea that how much something wiggles up and down equals risk is troubling me.
I am not sure why, but it might have to do with the fact that the word “risk” implies the unexpected. A surprise. Something that I can’t control.
When portfolios are designed, the idea is that we can find out how much risk you need and then design a portfolio with those precise risk characteristics.
If you want more risk, here, all we have to do it turn this dial. Tell me when it is enough and I will stop. Ok perfect, just the right amount of risk for you.
As if we can control “risk” like tuning in a radio dial.
I guess I have always thought of risk as something outside my control. Risk is the chance of something going terribly wrong. Something unexpected.
When I back-country ski, there is a risk of avalanche. There are things we can do to control for that risk, but after we do all we can to control things, the danger that’s left is called risk.
In fact to a large degree the greatest risks are the things that we have not even considered. The unknowable risks.
I think using a number like standard deviation gives us the false sense that all risk is measurable.
If we can measure it, then we can control it.
But, if we can control it, is it really risk?
Standard deviation measures what happens when the unknown becomes known. BUT, it does not tell us what will happen when the next unknown risk becomes known because we can’t know.
Sometimes this line of thinking is dismissed with the idea that history repeats itself and therefore there are no surprises.
Does that make sense how dangerous that it?
But if that is true, then why call it risk?
{ 11 comments }
Not only is it somewhat misleading to think that it can be measured, but I'd argue that it's not even measuring the right thing. (Or rather, it's not measuring what most people think it's measuring.) Ask any typical investor, and they'll say something to the effect of “risk = chance of losing money.”
People think we're talking about the probability that return < 0,
when in reality we're talking about the probability that return < “expected return.”
Absolutely loving this series, btw.
Thanks Mike. Yeah you are right.
Measuring the wrong thing, using the wrong measurement, and I am not even
sure that risk is measurable.
After having worked with “risk” as an student/academic for ten years, and as a practitioner for seventeen, I know that the “group think” in the financial economics world, is driven by an inability to publish generic quantitative risk papers if risk is not uniform to everyone. They need simplifying assumptions or the pretty math doesn't work.
The coefficients aren't stable; they aren't predictive. My view is that one must understand the “risks,” not the abstract “risk” of his positions, which occurs “randomly.”
* What will inflation do to my portfolio?
* What will happen if China's recent growth proves to be a fantasy?
* How does current Fed policy or non-policy affect me? How might it shift? What then?
* What of energy prices?
* Etc.
I wrote something on this recently: http://alephblog.com/2009/08/29/avoid-risk-make...
Keep it up. You write well, Carl.
I agree 100 percent with Mike that risk is properly defined as the chance of losing money. It is a terrible mistake to define risk as volatility of price. Warren Buffett has pointed out that, when you define risk as volatility, you end up thinking of a stock that doubles in price as being more “risky”than one that declines in price by 20 percent. This obviously makes no sense.
I believe that risk is measurable. The valuation level that applies for stocks on the day you buy a broad index fund tells you the level of risk you are taking on. Stock crashes are inevitable when you buy stocks at the sorts of price levels that applied from 1996 through 2008. We have never gone to those price levels and not seen a stock crash and an economic crisis in the years that followed. On the other hand, there has never yet been a time when investors did not see a strong long-term return on stocks purchased at times of reasonable or low prices. So the risk involved in stock investing is obviously much less when stocks are priced reasonably.
The more overpriced stocks are at the time you buy them, the more risk you take on by doing so.
Rob
Carl, This is good stuff. You are very correct in that we have tended to always think that measuring standard deviation was our way of trying to control risk. Well, last year, and in certain years past, that hasn't worked so well, especially for those of us in the MPT camp. Mike's post below is more to the point in that client's definition of risk is to not lose money. I know that you have even started to rethink MPT and what it does/does not mean when planning someones financial future. I read an article not long ago in which the author was stating that negative standard deviation is what we all need to be focused on. I may agree with that but Mike may have said it best in that we may not even be measuring the right things. So arriving at some conclusion may become the easy part, if we can simply figure out what all the right questions are.
Best Regards,
Chad Castle
>This idea that how much something wiggles up and down equals risk is troubling me.
>
>I am not sure why, but it might have to do with the fact that the word “risk” implies the unexpected. A >surprise. Something that I can’t control.
It seems to me predicting stock market returns over the next N years fills both those criteria- I certainly can't control it and I can't predict what it will be either. Just because I know the likely range of possibilities doesn't really predict anything.
If your plan assumes some minimum return (likely based on historical averages) then isn't getting less a real risk to your finances?
>If we can measure it, then we can control it.
I disagree- we measure a LOT of things that we can't control- the wind speed and direction of a hurricane, the returns of the S&P500….
>But, if we can control it, is it really risk?
Controlling exposure to a risk is very different than controlling the risk itself. Just because you know the likely extent of a risk (i.e. the standard deviation) and can alter your exposure to that risk that doesn't mean you have any control over the danger. In your skiing example you could eliminate the risk by never skiing or lessen it by skiing less. You didn't gain any more control over avalanches, but you really do reduce your risk by changing your exposure to that uncontrollable danger.
-Rick Francis
Thanks Chad. Yeah it is a lot to think about.
I think you are correct, time to think critically about the assumptions we
are making. These are not laws of physics!
Agree with MIke. This is an excellent series.
'Risk' is one of those words that is not well defined. In my world, it can mean either
a) The probability of losing money on a trade
b) The maximum possible loss from a given trade, or specific portfolio (as in how much is at risk?)
My business in trading, using risk-reducing option strategies. The discussion here is primarily directed to long-term passive investors, and the idea of risk must be different. None of these definitions of 'risk' is wrong. It's one of those words that depends on context.
Thanks Chad. Yeah it is a lot to think about.
I think you are correct, time to think critically about the assumptions we
are making. These are not laws of physics!
Agree with MIke. This is an excellent series.
'Risk' is one of those words that is not well defined. In my world, it can mean either
a) The probability of losing money on a trade
b) The maximum possible loss from a given trade, or specific portfolio (as in how much is at risk?)
My business in trading, using risk-reducing option strategies. The discussion here is primarily directed to long-term passive investors, and the idea of risk must be different. None of these definitions of 'risk' is wrong. It's one of those words that depends on context.
You're correct, “risk” is defined by standard deviation (ie. volatility, sigma). It isn't intended to predict the future, but academic research suggests it's the best predictor of future movement. Even market movements like we saw in 2007-2009 (down 50%) fall within the range correctly predicted by standard deviation. For example, the S&P 500 has historically had a sigma of about 19%. The normal distribution tells us that 68% of the occurrences happen within +/- one sigma, 95% occur within two sigma, and 99.7% occur within three sigma. If the S&P has a historical return of 12%, then 68% of the time, returns will be between -7% and 31% (+/- 19%). if we jump out to three sigma, then 99.7% of the returns should be between -45% and +58% (in a given year). yes, a wide range, but when I plot the returns of any major index in a histogram, it is a beautiful normal distribution (bell curve), with a mean of 12% and tails on each side almost exactly matching what historical standard deviation would predict.
Todd Larson, CFA
Adjunct Professor, Seattle University
Comments on this entry are closed.