Seth Klarman on Beta - Inviting Contrarian Views
In his legendary book 'Margin of Safety', Seth Klarman deplores the use of beta as a measure of risk. He argues that beta is simply a measure of historical volatility of a stock and doesn't say anything about 'risk' in its literal sense.
For instance, a low priced security may have a historically high beta, but going by the tenets of value investing, the value of the security may be much higher and locked in by an impending catalyst, say a liquidation. In this case, the actual risk would be very low, contrary to the high risk demonstrated by beta. Therefore, beta as a metric for risk is not useful for value investors, but only for short term traders.
I agree with Klarman's proposition, but I have two questions:
- Why is beta still widely accepted as a measure of risk even by long-only value investors? What makes it useful even for value-based long term investments?
- If beta is not a good measure of risk for value investors, is there any other metric that can 'quantify' risk? I'm talking about the literal "risk" of making a loss on an investment, not volatility.
What do you think?
If there was an easy standard of calculating the "risk" of a position, everyone would be using it, thus the acceptance of Beta. The literal risk is much more subjective, totally depends on what kind of risk you think lies beneath the surface. It's hard to quantify something like that, but that's why the big boys get paid
I made this post BECAUSE no measure of the sort seems to exists today, which is astounding to me. We're living in a world which created ABS, MBS, CDO and I can't even scratch the surface of what else. Sure, some portion of risk can never be quantified, but in a world full of quants, you really believe nobody can quantify a basic idea like the riskiness of an investment?
What comes to mind is credit rating, but no sophisticated investor uses that because they're not at all reliable. Similar to credit rating, is it not possible to construct a standardized set of parameters that can provide, at the very least, a relative measure of risk for everything from equities to LBO's?
@"Simple As...", @"BlackHat"... care to chime in?
I don't know if I've ever heard beta given any shred of credence at my shop. To quote one of our PMs: "I went to Wharton, I know the beta isn't real"
Is this guy your PM?
//www.youtube.com/embed/fB0weaPFiiE
Your risk is how much you can lose on the position if just about everything went wrong. An example of the difference between a value investor and a "value" investor would be AAPL. "Value" investors look at the P/E and say it's cheap ex-cash and hence has limited risk since the multiple can't possibly go that much lower but value investors would see 2/3 of the company's profit at risk from a single fad-type product. Not that AAPL is bad here but that's the difference in thought process.
And fwiw your risk should not be dictated by some arbitrary symbol, even if it's greek!
I don't know that risk is something that can be quantified or that it's something that a person should even want to distill down to a single metric... A huge part of understanding risk is understanding that you don't know what you don't know. And if you don't know what you don't know, how can you attempt to quantify something that you're not even aware of?
This, plus everyone has their own understanding of what they do/don't know and thus have their own range of valuations.
OP, I think you're trying to look at risk from a PE perspective. Sure, you'll do a lot of the same analysis that a HF might perform, but the subjectivity of your assumptions and scenarios will be a lot more broad with a simple public equity investment. In a HF, you don't have that ability to shape the investment toward a value that works best for you, you're a slave to the market. Thus more potential for downside and why value investors want a bigger cushion. That's why activist investors are trying to grab more control of companies. They can help lead the investment toward an estimated valuation and lessen the estimated risk, kind of similar to what someone at a PE firm would do (obviously they have their differences). When you really can't control what's going on with a company, your downside could be little to huge, even after calculating what a liquidation scenario might bring.
I think the reason no metric exists is because there's not even a good/rigorous definition of what risk is, which I think is basically what Simple As... is getting at. I've been thinking about how to think about risk in general for a while, with no real success at all (which is basically what I expected, but I figured it can't hurt to think)
I personally can't get past thinking of my certainty of investment outcomes as a probability distribution that evolves over time as I get more evidence about the nature of the investment. Sadly the entire Bayesian thing doesn't lend itself to "hard" assessments of risk, even if you came up with a good definition.You just have all sorts of bias starting in your base rate all the way to which evidence you even look at, and none of it's particularly tangible.
Quantitative risk assessment - in my experience - often quickly devolves into a game where everything's made up and the metrics don't matter. But I also don't work at a quant shop, so the guys at Wilmott might disagree.
There was a study done that showed that traditional risk measures, like volatility, were only about 1/10th as important to accurately forecast as expected returns. PM's don't get paid to make a bet on the riskiness of a security they get paid to make a bet on where it's most likely to go.
Of course that's all backward looking and if you truly think that you understand the forward looking, downside risk better than the market then it could definitely be helpful. But I would still make the case that stock prices are mostly indicative of the market's expectations on returns, not risk (for individual securities)
As an add-on to that, most PM's use very similar risk models often from third parties. The differentiation comes from varying opinions on expected returns.
Thanks for the responses, I get that risk is too qualitative and there are too many variables to distill it down to a single metric. I gave all this considerable thought during my commute and at first agreed that any quantification of risk will not be reliable. But, bear with me for a moment. This is pretty far out there so take it purely from an academic point of view.
When ABS were created, every tranche carried a rating from a rating agency, which indicated the riskiness of that tranche. Yes, I know where that led, but assume for a moment that in a hypothetical world, the tranches were assigned realistic ratings.
Also, despite the innumerable risks associated with stocks, there also exists an entire industry called equity research, which assigns buy, hold or sell ratings. Again, yes, I know, most of those reports are not worth a dime, but occasionally you do get star analysts consistently churning out quality recommendations.
Now look at a typical PE mezzanine investment, which comprises of a convertible security possibly carrying contractual cash flows, and common stock. I think it's safe to say that the debt portion of the mezzanine security can be rated comfortably, similar to the manner ABS tranches are rated. The equity portion of the investment can be rated by analyzing the company in the same way as done in equity research, after adjusting for potential effects of a PE investment. But instead of assigning buy or hold ratings, the rating can be on a scale, similar to credit rating. The debt and equity ratings can then be combined to create a scale-based rating for the entire investment.
What's the point of all this? Well, some pension funds and endowment funds invest through private equity funds. They sometimes also have the power to dictate that their funds be deployed only in investments that prescribe to certain norms. If the above were to become a reality, such pension and endowment funds would have the option to say that their money should be invested only in deals carrying a rating above a particular level.
Am I simplifying things too much? Yes. Will it ever happen? Probably not. But just consider the above from a purely academic point of view and tell me the flaw in my reasoning.
I like this idea and it makes sense to me intuitively for what "risk" is trying to measure. Volatility is kind of a weird metric in that it measures the magnitudes of both upswings and downswings, but investors are only really concerned with an excessive downside case when it comes to risk in some sense. Within fixed income, credit risk, which is what credit ratings try to measure, is solely focused on the downside (unlike interest rate risk, etc). Tough to implement, but it might have merit for equities as well.
I'm running the danger of ranting, but I'd be genuinely curious to learn of a situation where a PM is not paid to make a bet on the riskiness of a security
I understand your idea of focusing on ER over risk, but there's a rather complicated set of interactions between the two that make discussion returns without risk rather useless in my opinion (eg levering your FI to have ER of equity markets). If you don't understand the probability of loss, how can you believe your assessment of the securities return pdf and the resulting ER is correct?
I don't disagree with you - risk is a very important thing to take into account. Ignore financial advisors/wealth managers (as they're not trying to generate alpha) and quantitative funds (that might have a lot of leverage or have very unique strategies) for the moment. The point I was trying to make was that attempting to forecast risk (variances, covariances, betas, whatever) more accurately than the market is not going to drive risk-adjusted performance nearly as much as forecasting returns more accurately than the market will. I definitely don't mean to say risk isn't important to take into account though.
https://faculty.fuqua.duke.edu/~charvey/Teaching/BA453_2006/Chopra_The_…
Edit: -I agree with your first example. If you think (accurately) that the portfolio is much less risky and are able to lever it up without taking any additional haircuts, that will definitely help performance. Bonds in general probably do depend more on risk (than equities do) because of the lack of upside. -To your Third Point I would say that expected returns and risk are only tied together in theory (unless you're an EMH proponent I guess). In the real world, it's important to measure each independently, but its more important to get the expected returns right.
I still think saying that ER is generally more important is a pretty sweeping generalization. You might have data I have no access to, but in general I think that statement needs to be looked at on a case-by-case basis; I'm not comfortable making a generalizing statement I have no reason to believe is true.
I'm by no means a strong/medium form EMH proponent, but like anyone interested in public markets I believe/hope that there's some long term equilibrium. If I believe in said equilibrium then the assessment of risk matters, since it will determine the discount a company "deserves".
The core of my thought process really boils down to looking at risk-adjusted returns as ER/Risk. As I think we agree on, if your denominator is small, even seemingly small deviations in risk matter. It's the percent change that matters in both ways.
I really don't know how risk is viewed by private players so I'm lost in that debate. All I know is that if I could add non-recourse leverage to my holdings I'd be all over it.
Perferendis dolores debitis fuga. Corrupti repellendus consectetur dolorem est sed ea.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...