Hi… just went for an interview on a finance job scope. Interviewed by the company CFO and he ask me about Beta. From what i read in Kaplan notes, it states that Beta is a measure of systematic risk under the CAPM section. However he disagree with me and said that beta is not risk but volatility. He said that i’m wrong. Whoo… felt a bit sad when he said that to me… I google and found some articles that mention beta is measure of volatility. So is there any explanation bout this? I seems to be confused after this interview. thanks
Sounds like he was either being a d!ck or trying to make a point given the context of what he was saying - imo it’s kind of a BS semantic point. Besides, what’s “volatility”, there’s a lot of metrics for that too.
He’s a dick. It’s a bit of both. Beta is a measure of how much your security moves in relation to a market. Higher beta means bigger swings than the market = higher risk and higher volatility. I hate dogmatic people, they are blind.
Beta is a measure of risk and volatility is a measure of risk. Both measure risk. And risk shows up as volatility. Beta measures a specific kind of risk -> the risk that is correlated to the rest of the market (or, if you are using beta with another factor, the risk associated with that factor). Volatility includes this, but also includes company and/or industry specific risks. Basically two things go into beta: relative volatilities (is the asset more or less volatile than the market index), and correlation (how consistently does the asset move with the market). What makes beta different to work with than plain-old volatility is that it is a number that combines both ideas. The other thing that makes beta seem a bit strange is that it is a number that basically measures risk RELATIVE TO THE MARKET AS A WHOLE. So a beta of 1 means that it’s market-linked risk is just as volatile as the market, and a beta of 0.5 means it’s half as volatile. If a beta is low, it might be that the asset is highly correlated, but just not all that volatile (market goes up 2%, it goes up 20bps). Or it could be that the asset is just as volatile, but not very correlated.
Muddahudda is correct on both counts. If you believe in CAPM, then risk is volatility and beta is a measure of systemic risk. To be correct you would have to say beta can represent systemic risk under CAPM, but that’s really splitting hairs. Someone who argues about something like this is akin to someone arguing about whether long-term expected equity returns are 10% or 11% a year when the standard deviation is 20%.
One thing that beta is useful for is for figuring out how much of an asset’s volatility comes from non-systemic sources vs. how much just comes from the market as a whole. Basically, the systemic risk of an asset = beta*Mkt_SD And the non-systemic risk of an asset = SQRT( Asset_SD^2 - (Beta*Mkt_SD)^2 ) Once you have an estimate for the alpha of the asset, you can figure out how much non-systemic risk you need to take on to capture that alpha, and figure out if that’s worth having in your portfolio… (Also, you don’t truly have to believe in CAPM to find this useful… the index model of the market runs on a very simple assumption: that there are things that happen in the economy on a fairly frequent basis that affect most or all stocks very broadly. This is a very plausible assumption. Empirically, the index model works surprisingly well for something so simple. The fact that CAPM leads you to an equation that looks very similar doesn’t mean that you have to accept all the implausible CAPM assumptions and conclude that the market portfolio is the most efficient portfolio. i.e. beta can still be useful even if you don’t believe CAPM)
I suspect he was trying to gauge the depth of your understanding of beta and was hoping you would fight back with an answer along the lines of those given here. Or he just might be a d!ck.
@bchadwick Agree, you only need to assume that the idiosyncratic parts of your factor model have a zero correlation… However, I think it avoids the criticism that the interviewer was trying to make. He seemed to be saying that risk is not the same thing as volatility. What you describe is breaking up the standard deviation into the factor and idiosyncratic parts. What might be relevant to this CFO is perhaps the expected shortfall/conditional var of the factor and idiosyncratic parts. I generally feel that systematic risk has a meaning specific to CAPM (the risk that cannot be diversified away) and I do not see it used often in academic papers outside of references to CAPM. Personally, I find the factor/idiosyncratic risk more specific.
“and said that beta is not risk but volatility” i dont understand what that means. volatility is a measure of volatility. it’s a precisely defined concept, as is beta. as bchad said, beta can be decomposed into the volatility of the asset, divided by the market volatility, multipled by the correlation with the market. the correlation is a number whose absolute value is between 0 and 1. therefore necessarily a high-beta stock (i.e. beta >>1) also has high volatility (relative to the market vol). on the contrary, a low beta stock (beta <<1) may have low volatility, or it may have high volatility and small correlation with the market. to put it another way, an asset with high volatility may have high beta, or low beta (because of low correlation).
Looks like the CFO is a bit confused himself IMO. Beta is a measure of relative risk, and volatility is risk. Look at the formula for beta: [Corr(I,Mkt)*St.Dev.(I)]/St.Dev. (Mkt)
He must have wanted to point out that beta is not volatility in absolute terms but in relative terms. In a theoretical world, you could say that the asset/security is twice as much volatile as the market but that doesn’t tell you how risky the asset really is ( as the market might not move at all). In the real world though, I feel that if you you tell me something is twice as volatile as the market, I can safely deduct that asset class if twice as risky since volatility is considered by most as being risk. On the other hand you can go into a detailed discussion, if volatility is risk. As a portfolio manager I would consider more risky an asset that has the potential to go down in price with no price volatility than one that has volatility but stays within a range or goes up in price. So is the statistical measure of volatility a good measure of risk? I can argue not but I’m not smart enough.
I can argue not = I can argue that it isn’t 
So there are several possibilities here: 1) The CFO is confused about beta vs volatility. 2) The CFO is a jerk and threw you this question for the fun of watching you squirm with the answer. 3) The CFO knows it’s a tricky question, and the real issue is less about whether you got the right answer and more about how many different aspects of risk you bring into the answer: relative vs absolute volatility, correlations, systemic vs total risk, risk vs uncertainty, upside vs downside risk. I don’t know where you are in the CFA program, but this question is a deceptively simple one, and I never felt that the CFA curriculum did a good job of explaining this stuff (even if we were asked to mechanically apply it), so don’t beat yourself up too much over it… it’s a harder question than it seems. We should start a thread on deceptively tricky questions that come up in interviews this one on beta vs volatility is a good one. Another one that comes up is the difference between paying dividends and repurchasing shares. I’m sure that there are others. One of my favorite “strange questions” is to figure out what is the difference between “to rob” and “to steal”?
After I posted the above, it struck me that this is the correct answer: Q: Is Beta a measure of volatility or measure of risk? A: Yes.
There’s an important point that has been touched on but not clearly identified. In finance, “risk” has taken on a meaning that is quite different from standard usage. Any dictionary will emphasize “risk of loss”, “danger”, “hazard”, etc. In brief, risk usually means downside volatility. Upside volatility entails no risk. Academic finance has muddled this considerably (as e.g. Buffett will tell you every year in Omaha), beginning at least with Markowitz and MPT: “risk [is] variance of return (1952)” Here’s a question: two investments, Uponly and Downonly. Uponly will either maintain price or go up 50%; Downonly will either maintain price or go down 50%. Being a measure of volatility, beta (and other symmetric measures) rank these investments the same. To most people, their risk is entirely different. Refs: http://www.riskexpertise.com/papers/risk.pdf http://www.ams.sunysb.edu/~rachev/publication/DesProp-Rev2.pdf
This guy sounds like he would have been a great boss.
True, DH, and I agree with the main point you are making. However, in your Uponly and Downonly example, you would still have different expected returns (one positive, and one negative), and can make a judgement about that. Risk in the MPT sense is defined conceptually as deviance from your expected return. In risk, the real challenge seems to be twofold: 1) assets with the same expected return and SD but one has fatter tails than the other, so should fatter tails have a larger risk premium, and what should that be, 2) uncertainty, i.e. do we really have any idea what the true distribution of risks are, including risks of asteroid collisions, wars, regime changes, etc… I know you know about this, but I thought I’d add it to the thread.
Honestly, I’ve had similar discussions on here before, and could easily write for days on this innocent looking subject. First, a definition of risk. Risk is holistic, it is the simple fact that “more things can happen than will happen.” This is the best definition I’ve heard. Basically, Beta is neither volatility nor risk in it’s entirety, it is a section of both. It is a very neatly defined portion of covariance with the volatility that excludes idiosyncratic volatility and only shows the degree to which it moves with the market. So it is only a clearly defined portion of volatility. As a risk measure, it’s an even less clearly defined COMPONENT. As a historical measure of covariance, it is great if life continues without disruption. However, we’ve all seen correlations shift over the past few years as regimes change. Since risk is a forward looking concept, obviously, a rear looking indicator based on a correlation that will surely change under stress is not really a realistic measure of risk. Also, risk is a personalized concept. If you are super liquid and can hold to maturity or market recovers, then risk is a binary credit event, not market fluctuations. Same holds true if the instrument is illiquid and not priced often. If you are un-diversified, like some hedge funds, then covariance means nothing. You’re worried about the total variance. You also must account for risk profile in the form of skew and kurtosis. It’s worthwhile to look at downside deviation or semi-variance (perhaps the best singular risk metric, period.) in this regard. Here you can see if a security is exposed to large infrequent losses and common but small gains, this may be a risk you cannot bear that may not be reflected in the market or “beta”. With any historical measure, you’re going to run into the fact that markets don’t always price appropriately, and as such, risk of market mis-pricing (probably your biggest true risk) frequently goes unaccounted for. For instance, you could be holding shares with a historical beta of 2, but if your blue chip pharma company with a beta of 0.5 goes under due to manufacturing short-cuts, then that would be the “unknown” and frequently overlooked mis-pricing risk I’m referring to. This would obviously not have been accounted for by beta in this scenario. Also, pharma companies are exposed to many major risks, such as public policy (political) risk that are overlooked in beta as they have nothing to do with markets. Long story short, this is probably some of the thought process the guy was looking for. Probably wanted to see if you’re the type to take careful notes in class, ace all the exams and be able to create widgets perfectly over a full career with no real understanding or critical thinking skills, or whether you thought independently and understood these crucial financial topics with depth. The moral to this whole story is that it’s inappropriate to use a single measure, particularly a historical one for any given time period. Regimes change, and with them risk profiles, which means risk management process must change as well from phase to phase. While beta works well in periods of stability, it is useless in upheaval. As are most historical measures. The problem is spotting this changes. A risk team’s duty is to try, but this task is daunting. However, not everyone can afford to take a Taleb approach and buy OTM options either as we do not all share the same mandate. It’s an endless dilemma and one that has probably not seen it’s last market crash.
I agree that semideviation is a preferred measure of risk to standard deviation, but it’s a pain to calculate. If returns are perfectly symmetrical, semideviation = standard deviation (at least as parameters; the estimates will differ because of sample size), so to the degree that you are willing to assume that return distributions are symmetrical, standard deviation is a reasonable proxy, and you might be able to calculate skew instead and make some correction for that, rather than going through the extra code and statistical mess of calculating semideviation. An interesting observation someone made is that the odd moments of a distribution basically represent return opportunities (larger numbers better), and even moments represent risk (smaller numbers better). That’s always stuck with me.
DarienHacker that is what I was talking about too… Swan, nice post!