Margin of Safety

Suppose you estimate a share's value at $100 and can buy it for $75. The $25 gap looks like room to make a mistake.

But who established the $100? If that number rests on earnings the business cannot sustain, the cushion can disappear before you buy a single share.

The subtraction is easy. The useful question is whether the gap survives a less comfortable version of the business.

A buffer depends on the estimate

A margin of safety is the discount to estimated intrinsic value, expressed as a percentage of that estimate. Intrinsic value is what a share is worth based on the future benefits you expect as an owner.

The buffer gives your estimate room to be wrong. It does not cap your loss: a worsening business can be worth less than even your discounted purchase price.

In his 1992 shareholder letter, Warren Buffett credits Benjamin Graham with the principle. He pairs the price buffer with businesses he understands and whose future cash flows he can reasonably estimate.

Calculate the discount, not the upside

Divide the gap by estimated value:

For our $100 estimate and $75 price, that is ($100 − $75) ÷ $100 × 100 = 25%.

Reaching $100 from $75 would instead require a $25 ÷ $75 × 100 ≈ 33.3% price increase. Same $25 gap, different starting point. The discount compares the gap with value; the potential gain compares it with what you pay.

Neither percentage tells you whether the price will reach $100.

Put Harbor's estimate under pressure

The optional DCF model valued Harbor from future cash. Here, test whether its past P/Es can support a price buffer.

Our fictional coffee business reports these annual diluted earnings per share and year-end prices. Divide price by EPS to get its trailing P/E:

EPS and P/E are rounded here; calculations use full EPS. For FY2, that is $142.5 million of profit ÷ 51 million diluted shares.

A multiple-based estimate needs sustainable EPS and a justified P/E. Borrow the lowest and highest past P/Es as a test: FY2 and FY3's multiples give $62.27 and $66.00 when applied to the latest $3.00 EPS. This range describes past pricing, not an independently established intrinsic value.

Against the $66 quote, the gaps are about −6% and 0%. A negative gap means the price exceeds the estimate. Neither endpoint offers a discount.

The zero is built in: $66 ÷ $3 × $3 = $66. Using the latest price to set the multiple just returns that price. Those three past prices record what buyers paid. All three years could have been expensive.

When the quote falls

Return to the opening lesson's hypothetical 20% quote drop: $66 × 0.80 = $52.80. At unchanged $3.00 EPS, the comparison values stay put. The discount to the low estimate is ($62.27 − $52.80) ÷ $62.27 × 100 ≈ 15.2%; to $66.00, it is 20%.

Then keep that $52.80 quote, but suppose sustainable EPS is 20% lower: $3.00 × 0.80 = $2.40. Keep the two P/Es fixed. Both value estimates fall 20% too, to $49.82 and $52.80.

All amounts are dollars per share:

The gaps return to about −6% and 0%. The cheaper quote has bought no extra room for error.

Equal percentage cuts cancel the apparent improvement exactly. This tests how fragile the estimate is; it does not predict that price and earnings will fall together.

What supports the estimate

Harbor's rising reported free cash flow gives the earnings case some support: it grows from $106 million in FY1 to $120 million in FY3. That is evidence about earning power, not proof that past multiples were fair.

FY3 operating profit of $220 million covers its $20 million interest expense 11 times. Those numbers help you judge whether earnings can last; the multiple still needs a reason.

More uncertainty calls for more room. Morningstar's framework, for example, requires a larger discount for its highest stock rating when value is more uncertain. There is no percentage that makes every business a safe purchase.

For Harbor, $52.80 leaves a 15.2% gap only if $3.00 EPS lasts and FY2's multiple fits. Persistently shrinking profit margins or weaker cash generation would undermine that case.

At $2.40 EPS, that gap is gone. Without a convincing earnings case and a fair multiple, a lower price alone does not support a purchase. That is the problem behind value traps: a cheap-looking share whose business keeps shrinking.

Durable earning power, explored in economic moats, strengthens the estimate. Next, CAPE and market valuation ask whether history can help judge the price of the whole market.

In short

• A margin of safety is a discount to an estimate, not a limit on losses.
• Divide by estimated value for the discount and by price for possible upside.
• A lower quote adds no buffer when the value estimate falls by the same percentage.
• A useful estimate has evidence behind it and a clear reason you would revise it.