*[Originally published 2/19/2014]*

A few months ago I picked up a copy of Seth Klarman’s “Margin Of Safety.” While there were many valuable nuggets within the book, one thing that’s become a central part of my investment process is the use of conservative assumptions.

While many value investors are familiar with Ben Graham’s focus on margin of safety (i.e. buy something for 50 or 60 cents that has an intrinsic worth of $1), Klarman takes it one stage further.

Rather than just work on current, or even average assumptions, he talks about using conservative assumptions when calculating intrinsic value, and *then* applying a margin of safety. It’s a bit like using a double margin of safety.

This has become more relevant in current times since the application of margin of safety has expanded beyond tangible book values to estimating intrinsic value based on future earnings (e.g. using discounted cash flow, or DCF, models).

How do I use this when thinking about various business margins (e.g. operating margin, net margin)? Well, it helps to put margins into context. By looking at 10 year histories of financials for an interesting company, I want to know what is the range of operating and net margins for that business.

Also, with reversion to the mean being a central tenet in my investing approach, I want to get a sense of whether margins are high or low compared to their mean values.

When looking for companies with reliable earnings growth, available at attractive prices, the first thing I want to see is a steadily growing top line. Then, I look to get a sense of what is a reasonable or conservative assumption for the margins.

There’s usually some variation in the margin figures over 10 years, so I use numbers or a range which are well within repeatable, rather than outstanding. Equally, I’m not ultra-conservative, so I’m looking for something that is near or below the mean reverted range. This is the margin that I then apply to revenue estimates to derive earnings estimates. (Hence the focus on steady growers, as revenues are more predictable).

How does this work in a practical sense? Let’s use an example. A while back I wrote about Medtronic (NYSE: MDT). Over the last 10 years, the net margin for MDT had a low value of 14.2% in 2009 to a high of 22.8% in 2007. The mean for 10 years was 19.8%.

If using a mean reversion figure for net margin, a little under 20% would work (say 19.5%), or if using a conservative assumption, I would go down to about 15%. This gives me a range (as Klarman discusses, rather than single hard and fast numbers).

Let’s assume MDT is able to continue growing revenues at the same absolute amount over the next 5 years as the last 5 (this is easier for a large cap to achieve than grow revenue at the same CAGR – another element of conservatism), this would suggest revenues of just over $19.5 billion in 2018.

Applying the net margins indicated above, this gives us a range of earnings from $2.95bn to around $3.83bn. Assuming the number of shares outstanding remains at the current level of 1,009 million, this gives us an EPS range of 2.92 to 3.80 in 2018.

Now, MDT also has a steady history of buying back shares, reducing the outstanding number from 1225 million to 1027 million over the last 10 full reported years. So, allowing for a small reduction over the next 5 years at a similar ongoing rate could lead to a float of 950 million shares by 2018.

Plugging this into the calculation gives us an EPS range of 3.11 to 4.03. So, our most to least conservative estimates for EPS projection ranges from 2.92 to 4.03 for 2018.

Admittedly, this is quite a wide range, but that’s OK. It comes down to how you feel about the assumptions you’ve plugged into your model and if there are other factors which, though not captured in a model, cause you to lean more towards one end than the other (e.g. qualitative factors).

But what is interesting is that the EPS for 2013 is already 3.74 (fully diluted, as are my EPS range projections). This suggests, that using conservative assumptions, MDT does not have much if any upside in EPS over the next 5 years, and may well decline.

If we use a crude a fair value measure -- say, a PE ratio of 15x -- this suggests a fair value of $44 to around $60. Given that I look for a 15% CAGR, this means I need a double from current price to 5 year fair value – resulting in a buy price somewhere $22 – $30. As it happens, MDT seems to be a steady consistent performer as a business so I might lean closer to the $30 buy price.

But it’s still a long way from the current price of $56. Of course, if it gets either a growth or momentum wind up its sails, it will never come close to my buy price. That’s fine, I have a watchlist of companies I like where I am willing to wait for them to become attractively priced.

When using the approach above, I will certainly miss some trading opportunities by using conservative assumptions and considering reversion to the mean. Stuff has to become almost deep value before I can buy.

But what it also means is that when I do buy, there will be a margin of safety in both the price paid AND the assumptions that went into the model. It also teaches patience and directs energy into the search for a few good situations. Buffett relays this in his description of the 20 hole punch card:

*“I always tell the students in business school they’d be better off when they got out of business school to have a punch card with 20 punches on it. And every time they made an investment decision they used up one of those punches, because they aren’t going to get 20 great ideas in their lifetime. They’re going to get five, or three, or seven and you can get rich off five, or three, or seven. But what you can’t get rich doing is trying to get one every day.” *Warren Buffett, 1991

**Raman Minhas*** writes about using value investing, saving and psychology to help you reach financial independence. If you enjoyed this article, join his free newsletter.*

**Disclosure**: None