Saturday, April 16, 2011

Margin of Safety: An alternative risk assessment tool?

I have lost count of the number of times I have been taken to task for not mentioning "margin of safety" in my valuation and investment books. In general, the critique is usually couched thus: "Instead of using beta or some other portfolio theory risk measure, why don't you look at the margin of safety?". While I see the intuitive value of paying heed to the "margin of safety",  I don't see the two as alternative measures of risk. In fact, I think that risk measures in valuation and margin of safety play very different roles in investing.

I know that "margin of safety" has a long history in value investing. While the term may have been in use prior to 1934, Graham and Dodd brought it into the value investing vernacular, when they used it in the first edition of "Security Analysis". Put simply, they argued that investors should buy stocks that trade at significant discounts on value and developed screens that would yield these stocks. In fact, many of Graham's screens in investment analysis (low PE, stocks that trade at a discount on net working capital) are attempts to put the margin of safety into practice.

In the years since, there have been value investors who have woven the margin on safety (MOS) into their valuation strategies. In fact, here is how I understand how a savvy value investor uses MOS. The first step in the process requires screening for companies that meets good company criteria: solid management, good product and sustainable competitive advantage; this is often done qualitatively but can be quantifiable. The second step in the process is the estimation of intrinsic value, but value investors are all over the map on how they do this: some use discounted cash flow, some use relative valuation and some look at book value. The third step in the process is to compare the price to the intrinsic value and that is where the MOS comes in: with a margin of safety of 40%, you would only buy an asset if its price was more than 40% below its intrinsic value.

The term returned to center stage a few years ago, when Seth Klarman, a value investing legend, wrote a book using the term as the title, published in 1991. In the book, though, Seth summarizes the margin of safety as "buying assets at a significant discount to underlying business value, and giving preference to tangible assets over intangibles".  Seth is a brilliant thinker (I love the letters he writes to investors..)  and the book has original and interesting ways of looking at risk. I learned a great deal about the ethos of value investing but it did not alter the fundamental ways in which I approached estimating intrinsic value, only the ways in which I used that value.

The basic idea behind MOS is an unexceptional one. In fact, would any investor (growth, value or a technical analyst) disagree with the notion that you would like buy an asset at a significant discount on estimated value? Even the most daring growth investor would buy into the notion, though she may disagree about what to incorporate into intrinsic value. To integrate MOS into the investment process, we need to recognize its place in the process and its limitations.

1. Stage of the investment process: Note that the MOS is used by investors at the very last stage of the investment process, once you have screened for good companies and estimated intrinsic value. Thinking about MOS while screening for companies or estimating intrinsic value is a distraction, not a help.
Proposition 1: MOS comes into play at the end of the investment process, not at the beginning.

2. MOS is only as good as your estimate of intrinsic value: This should go without saying but the MOS is heavily dependent on getting good and unbiased estimates of the intrinsic value. Put a different way, if you consistently over estimate intrinsic value by 100% ore greater, having a 40% margin for error will not protect you against bad investment choices.

That is perhaps the reason why I have never understood why MOS is offered as an alternative to the standard risk and return measures used in intrinsic valuation (beta or betas). Beta is not an investment choice tool but an input (and not even the key one) into a discounted cash flow model. In other words, there is no reason why I cannot use beta to estimate intrinsic value and then use MOS to determine whether I buy the investment. If you don't like beta as your measure of risk, I completely understand, but how does using MOS provide an alternative? You still need to come up with a different way of incorporating risk into your analysis and estimating intrinsic value. (Perhaps, you would like me to use the risk free rate as my discount rate in discounted cash flow valuation and use MOS as my risk adjustment measure... That's an interesting choice and worth talking about ... I know that Buffett claims to do something similar, but he discounts only the cash flows that he believes he can count on, making his cash flows risk adjusted cash flows.)

I know.. I know... There are those who argue that you don't need to do discounted cash flow valuation to estimate intrinsic value and that there are alternatives. True, but they come with their own baggage. One is to use relative valuation: assume that the multiple (PE or EV/EBITDA) at which the sector is trading at can be used to estimate the intrinsic value for your company. The upside of this approach is that it is simple and does not require an explicit risk adjustment. The downside is that you make implicit assumptions about risk and growth when you use a sector average multiple... The other is to use book value, in stated or modified form, as the intrinsic value. Not a bad way of doing things, if you trust accountants to get these numbers right...
Proposition 2: MOS does not substitute for risk assessment and intrinsic valuation, but augments them.

3. Need a measure of error in intrinsic value estimate: If you are going to use a MOS, it cannot be a constant. Intuitively, you would expect it to vary across investments and across time. Why? The reason we build in margins for error is because we are uncertain about our own estimates of intrinsic value, but that uncertainty is not the same for all stocks. Thus, I would feel perfectly comfortable buying stock in Con Ed, a regulated utility where I feel secure about my estimates of cash flows, growth and risk, with a 20% margin of safety, whereas I would need a 40% margin of safety, before buying Google or Apple, where I face more uncertainty. In a similar vein, I would have demanded a much larger margin of safety in November 2008, when macro economic uncertainty was substantial, than today, for the same stock.

While this may seem completely subjective, it does not have to be so. If we can bring probabilistic approaches (simulations, scenario analysis) to play in intrinsic valuation, we can not only estimate intrinsic value but also the standard error in the estimates.
Proposition 3: The MOS cannot and should not be a fixed number, but should be reflective of the uncertainty in the assessment of intrinsic value.

4. There is a cost to having a larger margin of safety: Adding MOS to the investment process adds a constraint and every constraint creates a cost. What, you may wonder, is the cost of investing only in stocks that have a margin on safety of 40% or higher? Borrowing from statistics, there are two types of errors in investing: type 1 errors, where you invest in over valued stocks thinking that they are cheap and type 2 errors, where you don't invest in under valued stocks because of concerns that they might be over valued. Adding MOS to the screening process and increasing the MOS reduces your chance of type 1 errors but increases the possibility of type 2 errors. For individual investors or small portfolio managers, the cost of type 2 errors may be small because there are so many listed stocks and they have relatively little money to invest. However, as fund size increases, the costs of type 2 errors will also go up. I know quite of few larger mutual fund managers, who claim to be value investors , who cannot find enough stocks that meet their MOS criteria and hold larger and larger amounts of the fund in cash.

It gets worse, when a MOS is overlaid on top of a conservative estimate of intrinsic value. While the investments that make it through both tests may be great, there may be very few or no investments that meet these criteria. I would love to find a company with growing earnings, no debt, trading for less than the cash balance on the balance sheet. I would also like to play shortstop for the Yankees and slam dunk a basketball and I have no chance of doing any of those and I would waste my time and resources trying to do so.
Proposition 4: Being too conservative can be damaging to your long term investment prospects.

So, let's call a truce. Rather than making intrinsic valuation techniques (such as DCF) the enemy and portraying portfolio theory as the black science, value investors who want to use MOS should consider incorporating useful information from both to refine MOS as an investment technique. After all, we have a shared objective. We want to generate better returns on our investments than the proverbial monkey with a dartboard... or the Vanguard 500 Index fund...


SiMpLy NiKhiL said...

Thanks a lot prof, your post as usual, certainly adds a lot of value to we readers.
Simple but most important for me here was MOS is as good/bad a tool as one's estimate of intrinsic value.

Unknown said...

Great post.

Many investors have latched onto the concept of MOS without understanding some of its implications.

MOS is a very simple but extremely useful concept.

As an individual investor, my potential investing universe is huge, so applying a MOS religiously ensure that I make fewer Type 1 mistakes.

I've gravitated towards microcap stocks for the simple reason that diversions from intrinsic value are much more extreme than in large-cap stocks like KO or JNJ.

See my recent post on IBAL.OB, a microcap manufacturing company that is trading around its cash balance with no debt, decent returns on capital, & solid although lumpy profits.

IMO, although it might not be shortstop for the Yankees, there are still opportunities to get a few at-bats in the big leagues..

Unknown said...

Thank you for the discussion.

I am very surprised to read that some investors still confuse this MOS concept and use it as a risk measure !

shlomi ardan said...

Thanks Answath, interesting post.
What bothers me, though, is not why you did not use MOS as a risk measure, but why use beta, which is a volatility measure, as a risk measure. does risk really equal volatility? can't the public really just be moody about something which is fundementally safe? or, vice versa, feel something is safe when it's really very dangerous?

Aswath Damodaran said...

My point is that if you don't like beta - and I understand your concerns - MOS does not really fill the void. You have to come up with a different risk assessment measure (and there are quite a few) to help you in your intrinsic valuation. So, this post is not a defense of the use of beta but an argument that MOS is not the alternative that you are looking for.

mercman said...

Thanks for posting your insights I find them very helpful. I agree that beta may not be as precise as one needs it to be and that risk and volatility are two different animals.
A couple of quick Google searches for terms such as "substitute for beta" did not yield much in the way of enlightenment.I would love to see this discussed in future musings.On a side note I want to say thank you for all of the information and insight you put out there it has been extremely helpful. it is not easy being an individual in a shark tank.

mahesh said...

I personally believe that Margin of Safety as a term is hyped a lot. It is nothing but age old wisdom of being a bit conservative in your approach to be better prepared in case of any unknown shocks. Be it being conservative in predicting cash flows, valuing tangibles, using risk premiums etc. Also, as mentioned by you, using MOS on any valuation is useful only if the valuation itself is right.

Money Chats said...

Most people are familiar with the concept of borrowing money to invest in property, and using the property as security. The same principle applies to a margin loan (also known as a share investment loan) where the amount you borrow for investment purposes is secured against the shares, managed funds, and cash in your portfolio.
Investing in Mutual Funds

John said...

I too thought this was a good post.

I have spent some time thinking about quantitative approaches to value investing. Until 5-10 years ago, I don't think quantitative investors had the tools necessary to adjust portfolio allocations based on the concept of margin of safety, but now I believe they do.

Meucci's Entropy Pooling approach to taking views combined with mean-CVAR optimization would allow you do to this. Entropy Pooling would allow you to take a view that the probability of equity returns being below the margin of safety is less than some percent. With this new distribution reflecting your views about intrinsic value, you could perform mean-CVAR optimization to create a new portfolio accounting for the lower expected tail risk of the value companies.

Mike said...

Hi Professor, very informative post! thank you.

I would urge you to go a little futher and respond to Adam's post about microcap ABAL.OB. It has all the characteristics he has cited.

However, in my opinion the stock is priced properly.
Main drivers: Low liquidity, uncertain management succession plan.
Hi Adam, even if you are correct and the market has priced it wrong, there is a high chance that the market will remain irrational for a long time, thereby keeping ABAL as a penny stock.

Unknown said...


Just want to make sure the stock you were looking up was IBAL.OB (not ABAL.OB)

While I think there can be some disagreement on how much value you place on the business, I think most people would agree that there is SOME value - IBAL is a business with tangible assets, valuable land, and $3m in cash in the bank, yet the market is pricing all of this below zero.

I don't think that is rational.

Could the market continue to be rational for awhile? Sure, but I'll take a boring business that throws off cash w/ possible short-term catalysts to help unlock that value.

VIKRAM said...

Sir ,Thank you for the Post.

I have a doubt. In the backdrop of MOS , value investors buy stocks whose intrinsic value is more than Market price to book profits when Market price levels or crosses intrinsic value. What if we had a chance to short those shares whose intrinsic value is much lesser than market price and exit when the market price bottoms near intrinsic value. Will this idea help in price realisation and price discovery? Do we have proper instruments to use this as a startergy. Correct me if I am wrong.

Diffusion said...

Thanks a lot prof! I especially like the discussion on the so called Type 1 Error vs. Type 2 Error.

My take away is that, instead of evaluating intrinsic value and then apply a constant MOS, an investor may better off to start with the top k undervalued stocks in a sector / industry and gauge its financial quality.

In this way all the 3 factors got addressed:
1. Market Risk, which can be gauged by the beta of the sector or the industry
2. Valuation, and
3. Quality

Gaurav Mehta said...

Just wanted to point out that beta is not that bad a measure of risk as it is normally made out to be.

Risk is on both sides positive and negative.... I would suggest watch companies that have beta lessser than 1 and betas greater than 1.4 when the market mood is extremely bullish or bearish. Lower beta stocks actually tend to under reach the market momentum and higher beta stocks overeact.... be it either a normal 10 -15% market correction or a market crash like 2008.

I haven'r really seen a better measure of risk than beta as yet.

Diffusion said...

It becomes clearer to me that "risk" means different things to different people.

Beta measures how fast a stock moves once the market moves. A portfolio manager may care this a lot because he has no control on the market's direction, so the best thing he can do is to understand what's expected loss (risk comes in here) he would have if the market suddenly moves (when such as earthquake hits Japan).

A related measurement is Volatility, which measures how fast the market would move. Short term investors (traders and hedge funds) may care about this a lot because there is ways to hedge (risk comes in here) or profit from sudden market move.

MOS measures the valuation error (thus risk) a value investor cares about. There are different approaches to value the intrinsic value of a stock, and none of them is perfect. So there is error built in and value investors want to avoid them. Because they focus only a picking stock, so the valuation error is the major risk they want to deal with.

Now do we have an agreement on what is risk? Maybe not. So I wouldn't hope we agree upon on what is the best way to measure risk.

Gaurav Mehta said...

Sir, please could you have a look at the email i sent you on a project.. Thx

Pranav Pratap Singh said...

Dear Prof,

At times I think that the concepts like MOS become so popular because they provide the investor with some conviction about his buying and selling price and hence provide her unwittingly with investing discipline. Even in a case the value that the investor has accorded to a share is not correct and lets say that the market value is closer to the fair value, by using techniques like MOS the invetor gets committed to buying the share at his purchase price (lower) and sell it at his selling price(higher) holding it throughout the period.
Markets being volatile, there would be high chances that the share will attain both his buying and selling prices and the investor might make money because of his committment even if the valuation or technique might not be appropriate.

Would you agree to this? Would there be a way to take out the effect of this conviction of the investor, which might be misguided at times, to evaluate intrinsic success of investing/valuation techniques?

Indira said...

Hello Aswath,

Thanks for sharing your thoughts.

As a user of MOS, and having conducted proprietary modeling with composite metrics on our 1,400+ company database, would like to point out that there are other uses of MOS than the one you have specified. MOS has proven to be the most valid approach for our investments early in the screening process in generating alpha for our long term investments.

Indira Amladi
Managing Member, Princeton I. V. Management LLC (General Partner)

Diffusion said...

Indira, would you please elaborate a little bit more on the "other use" of MOS? Thanks.

Gaurav Mehta said...


What you have pointed out...should we assume....if you hda to screen through 1400 companies to find out which of those are trading much below value using MOS after you a cettain value has been assined to each based on broad parameters including PE, P/B, ROE, ROC, PE, Margins, Debt/Equity, EBITDA ect. I normally use this process to screen my 500-600 name indian Companies watchlist to see if there is anything trading at a significant discount.

If thats the case then again MOS would have the same meaning as to what Prof has been stating i.e. MOS is nothing but a tool to buy or screen securities which are trading way below value however the valuation of those securities is normally done in a different way i.e. Relative Valuation for screening and DCF for Buying for a Value Investor. I have still to see a model which uses MOS to Value a security.

In case you mean something else or use MOS to value please specify.


Pranav Pratap Singh said...

MoS sounds more like an investing philosophy rather than a technique. As Seth Klarman wrote in his annual letter "Benjamin Graham’s margin-of-safety concept – to invest at a sufficient discount so that even bad luck or the vicissitudes of the business cycle won’t derail an investment – is applicable to the economy as a whole. Bridges intended for ten-ton trucks are overbuilt by engineers to hold vehicles of 30 tons."

The key for success of MoS could be being very sure of the value of the investment, so as to ensure safety. One has to be sure that the truck can hold vehicles of 30 tons to feel that safe. It would always be worthwhile to make investments where value is substantially more than price paid. But, may be, investment in businesses that are asset heavy or where valuation can be done on the basis of what exists rather than what may accrue in future, would go better with MoS concept. So, for example, it may be more suitable to use MoS for evaluating manufacturing companies with a substantial track record rather than companies in early stages of growth. Would it also imply that following only MoS philosophy would exclude investments in high growth companies?

Also I was not very impressed with Seth Klarman's book. However, took the effort of looking for his annual letters after going through Prof's accolades for him. It has been a very happy discover. Suggest you also try it, if you haven't already.

The Silent Assassin said...

Sorry Aswath, but I learned almost nothing from this post of yours.

Your first and second propositions (stage in process; substitution for risk assessment & valuation) can be easily gleaned from a single night's reflection on the MOS philosophy. You don't even need experience! Plain common sense would do the trick.

Your fourth proposition (Type 1 and 2 errors), while a bit harder to glean, can be obtained by ONE WRONG INVESTMENT DECISION resulting from a conclusion produced by an excessively conservative analytical framework. Every single investor out there has made at least one bad decision in the past in terms of opportunity costs. I admit I learned this the hard way D:

What piqued my interest, instead, was your third proposition. I personally find this too true. I tend to be quite lax in my MOS standards when I'm facing companies of low risk (to the point that I bought stock even when they were 20% overvalued... in a zero growth scenario, of course.)

And now you propose a way to systematize this using probability statistics? I'm highly intrigued on its practical application, because I for one would like to know how I would assign/determine the probabilities of the slow, fast, and stagnant growth scenarios I often employ in my DCF valuations.

Care to provide a hypothetical situation and your recommendation on how to determine these probabilities? I'd even welcome the chance to see something from your own works. It'll be a good foundation to build upon.

Unknown said...

Thanks Aswath for this valuable post. I am Risk assessment provider in UK and i know how important is risk assessment training for the people. Risk assessment list the hazards etc that can cause harm and then you find away to put control measures in place to reduce this or even eliminate the risk completely.

Arnold Brame
Health and Safety Training UK.