Margin of Safety

The Concept

A margin of safety leaves room for a plan to go wrong without causing a disastrous loss. You might keep spare capacity in a system, cash for unexpected expenses, or time between appointments. The buffer gives you more options when conditions change.

The gap between your car and the one ahead gives you time to react if the driver brakes suddenly. A similar allowance can help a system cope with a load greater than expected. Its usefulness depends on what goes wrong and how much stress the buffer can absorb.

Sometimes things fail for reasons beyond our control. A lot of tricky and highly consequential decisions, though, fail for preventable reasons. When we don’t consider how things might go wrong and plan for them in advance, we’re left flat-footed when they do go wrong. Then we end up reacting instead of reasoning. It’s much easier to plan for things that could go wrong in advance when you’re calm and open-minded than it is to respond when things are in the midst of going wrong. That planning relies on imagining extremes rather than depending on averages—just as an engineer doesn’t design a bridge for the average load but for the worst plausible conditions, we too need to ask, “What if I’m wrong? What if things don’t go as planned? How much extra capacity do I need to handle the unexpected?”

When failure is expensive, it’s worth investing in large margins of safety.

The LTCM Example

If you’re an investor, you’ve likely heard the story of Long-Term Capital Management (LTCM), a hedge fund founded in 1994 by a prominent investor who managed to get two Nobel laureates on his board. LTCM had a high risk portfolio that was highly acclaimed for its incredible returns—over 21 percent in its first year, then 43 percent in its second year, and 41 percent in its third year.

Imagine being an investor in this environment. You see this hedge fund taking off, and your friends are boasting of their success and urging you to join in on the bonanza. They tell you about the amazing people working there—people with incredibly high IQs, including two Nobel Prize winners, who are also experienced in their respective fields and who have invested substantial amounts of their own money.

You watch your friends double and then quadruple their investments. You start wondering whether you should invest everything as well. Your own portfolio is returning 8 to 12 percent year over year—good returns, but not 40 percent! Is the rest of the world going to get rich while you play it safe?

Consider now two scenarios. In the first, you decide to follow your friends and invest everything you have in the fund. A few months later, Asia and Russia experience a financial crisis. That crisis together with LTCM’s highly leveraged investments loses them $4.6 billion in less than four months. In this scenario, you (and your friends) end up in financial ruin.

Imagine a different scenario. It’s November 1997, and LTCM’s recent returns look impressive. You could still allow for a loss by limiting how much you invest. This limits your exposure to the fund. In value investing, the related idea of a margin of safety usually means a gap between the purchase price and your estimate of underlying value. Both leave room for error, but they address different risks.

A margin of safety is like having insurance. If you knew in advance you wouldn’t need to make a claim this year, it’s a waste of money to buy insurance. The problem is that you don’t know in what year you’ll need to make a claim, so you buy it every year. It might seem like a waste of money in years when nothing happens, but it shows its real value in years when something does.

The size of a buffer depends on the loss you need to survive and the resources you can spare. For the second scenario, consider what a loss in 1998 could mean. Investing only one tenth of your portfolio in LTCM would limit the direct loss on that holding during the 1998 crisis to at most 10 percent, assuming no borrowing or further obligations. The rest of the portfolio remains outside the fund, although it could also suffer losses in the crisis.

Preparation vs Prediction

Social pressure can make the first scenario tempting: your friends are doing well, and you don’t want to miss out. Confidence in recent returns can also make a loss seem unlikely. You end up planning as though LTCM’s next year will resemble its first three, leaving little room for a different outcome.

In the second scenario, you allow for your forecast to fail. Asking “What if I’m wrong?” can help you decide how much you could afford to lose before making the investment.

Warren Buffett has a saying that I often come back to: “Diversification is protection against ignorance. It makes little sense if you know what you are doing.” The thing is, most of us rarely know what we are doing with the confidence required to go all in. When you don’t know what you are doing, a margin of safety saves you from the worst outcomes. Even when you do know what you’re doing and you make the best possible decision at the time, things can change.

An unused buffer has a cost, but that does not mean it was a mistake. Assess it against the risks you faced when you made the decision, rather than only the outcome that happened.

When Margins of Safety Are Needed

We can’t prepare for everything. Some events overwhelm even substantial reserves. Still, we can prepare for recurring sources of disruption without knowing when they will occur.

The key is to always assess how high the stakes are. The greater the cost of failure, the bigger the buffer should be. On a personal level, the events we can expect include:

• Grief from losing a loved one
• Health issues
• Relationship changes
• Financial pressures
• Challenges in meeting our career goals

On a more macro level, they include the following:

• War and political dissent
• Natural disasters
• Environmental and ecological changes
• Economic fluctuations: both collapse and growth
• Technological advances and resistance to them

In all these areas, what seems historically extreme may not be the true limit. Conditions can surpass former worst cases, so simply referencing the past is insufficient. A margin of safety is meant to handle not only what has occurred, but what might occur under plausible but more severe circumstances.

How to Build a Margin of Safety

Suppose a bridge is expected to support 5,000 tons at once on an average day. A capacity of 5,001 tons would leave almost no allowance for heavier traffic, estimation error, or deterioration. Capacities of 10,000 or 20,000 tons illustrate larger buffers, but choosing an appropriate capacity requires an engineering analysis of plausible loads and failure modes. The average load alone is not enough.

Backups can also help a system keep functioning after a failure. Alternative routes to work may help when a road closes; redundant aircraft systems can preserve essential functions when a component fails. Check whether the backup would survive the same event that disables the primary option.

Businesses can leave room for uncertainty in their forecasts. For example, a team might wait for a response to a follow-up email before treating someone who downloads a white paper as a qualified lead. That stricter definition can reduce overestimation of demand, though it may also miss some genuine prospects.

Doubling an estimated need can be a useful starting point for a rough plan, but it is not a general test of an adequate margin of safety. The appropriate buffer depends on the uncertainty, the consequences of failure, and the cost of carrying the reserve.

For example, if you want to feel financially secure even if you lose your job, you can estimate how long it will take you to gain employment again, and then save enough to live off savings for double that amount of time.

But we need to adapt our margin of safety to the individual and situational circumstances. If the cost of failure is high, and outcomes are more consequential, you want a large margin of safety. For instance, if you’re worried about losing your job, and you’re in a sector or economy that’s volatile, you’ll want to increase the length of time you can take care of yourself while unemployed.

When failure would be cheap and easy to recover from, a smaller buffer may be reasonable. A long record of success can inform your estimates, but it does not remove the possibility of a different outcome.

Yet even established patterns aren’t foolproof. As Nassim Taleb writes in The Black Swan, “Consider a turkey that is fed every day. Every single feeding will firm up the bird’s belief that it is the general rule of life to be fed every day by friendly members of the human race ‘looking out for its best interests,’ as a politician would say. On the afternoon of the Wednesday before Thanksgiving, something unexpected will happen to the turkey. It will incur a revision of belief.” Our outcomes can sometimes upend even our most well-established expectations.

Better information may support a narrower valuation range. The account describes Warren Buffett as seeking discounts of 30–50 percent to estimated value, or a 30–50 percent margin of safety, while sometimes paying closer to estimated value for a business he understands well—perhaps leaving only a 20 percent margin. Treat these as illustrations, not a fixed buying rule. Estimated value can be wrong, and confidence alone does not establish how much protection a price provides.

One of Warren Buffett’s core tenets for buying a business is that if he doesn’t understand it, he doesn’t buy it. In other words, if he doesn’t have enough information to calculate a margin of safety, he doesn’t invest at all. He also knows that not all margins of safety will protect him—the goal isn’t to get it perfect for every stock he buys; it’s to use the best possible strategy for all his stocks in the big picture.

A buffer can change behavior. Someone who feels protected may take more risk, using up part of the intended benefit. Research on seat belts has raised this possibility, but that does not establish that seat belts make driving less safe overall. A simpler example is setting a watch fifteen minutes ahead: if you always subtract the extra time, the reminder no longer changes when you leave.

Larger reserves also have costs: unused capacity, tied-up cash, and slower work. A useful buffer covers a plausible disruption at a cost you can afford. Review it when the system, its environment, or the consequences of failure change.

Before committing to a plan, identify the assumptions that could fail and decide what room you need to recover. Check that the reserve will still be available when you need it.