The art of investment has one characteristic that is not generally appreciated. A credible, if unspectacular, result can be achieved by the lay investor with a minimum of effort and capability; but to improve this easily attainable standard requires much application and more than a trace of wisdom. — Ben Graham, *The Intelligent Investor.*

It’s not supposed to be easy. Anyone who finds it easy is stupid. — Charlie Munger.

Please use the link below to read this lecture on Moody’s. You will learn how a great investor used Buffett’s purchase of Coke in 1988 to make a case for buying Moody’s in 2000 at a high multiple (21) of earnings.

http://www.scribd.com/doc/69137839/Lecture-10-Moody-s-Corporation-Example

Note how this Great Investor is focused on quality companies. You will not learn in business school his creative comparison of two companies at different times and in different industries.

I would like to ask about some of these “return on reinvested income” calculations by a worked example.

Specifically, let us take LON:SN (Smith and Nephew), which is traded in the UK, and also the US, I believe. I have figures in pence, so I would like to use those if that’s OK by you. In very simplistic terms, SN operate in the health care equipment and services sector, where they provide artifical joints, keyhole surgery, and wound dressings (like I say, keeping it in simple terms). Here is an abstract of their figures:

Apologies in advance if the formatting comes out wrong.

My first question is: should I use basic EPS or adjusted EPS? I’m going to assume EPSa (adjusted EPS) for now.

The next question is: what years should I include in order to obtain “total adjusted EPS”?

In other words, should I use:

* 224.48 – the sum of 13.96 16.02 18.38 22.30 25.41 22.84 26.43 37.95 41.19

* 269.01 – the sum of 13.96 16.02 18.38 22.30 25.41 22.84 26.43 37.95 41.19 45.43

* 255.95 – the sum of 16.02 18.38 22.30 25.41 22.84 26.43 37.95 41.19 45.43

To my mind, it’s the first number that I should use. My reasoning is, if you look at a one year period at the beginning, say, then I would argue that it is the 13.96p in 2001 that is used to generate the income for 2002. In other words, don’t include the last set of numbers.

Assuming I’m right, then the total divend payouts are:

* 54.67p – the sum of 4.65 4.80 4.95 5.10 5.60 5.49 6.07 8.96 9.05

So, the total retained earnings ploughed into the company are:

* 169.81p : being 224.28-54.67

So, the EPS went from 13.96p in 2001 to 45.43p in 2010, which is an increase of 31.47p (=45.43-13.96).

This implies a return on retained earnings of 18.5% (=31.47/169.81)

That is, of course, an excellent rate of return. Do you agree with my calculations?

I have a couple of other ways of estimating “return on retained earnings”.

Method 1 is to do a least-squares exponential fit. If I do that – and I can’t show the calcs because it involves solving an equation – I obtain a value of 13.8% – somewhat under 18.5%.

Method 2 is what I call a “Graham shortcut”. For the defensive investor, he insists on a ten-year growth of at least one-third in per-share earnings. On a compound basis, this works out at 2.9% (take the 10th root of 4/3). What I do, to get an annual compound rate, is take the median of the first 3 years, and the median of the last 3 years, and take the seventh root of the ratio (assuming a 10-year period).

To illustrate: the median EPSa of the first 3 years is 16.02p. The median of the last 3 years is 41.19p. So the compound rate is 14.4% (=(41.19/16.02)^(1/7). That’s pretty close to the least-squares fit of 13.8% – and it’s one that anyone can do with a calculator. Anecdotally, I have found that “method 2” gives a pretty good approximation to “method 1” in most instances.

You may ask why I take the seventh root. The reason is that if you assume that the median EPS lies in the middle of the 3-year period, then there are seven intervals. In case that explanation is non-sensical, the median for the earlier period is in 2002. The median for the later period is 2009. That represents a gap of 7 years.

In fact, I use 7 years regardless of whether the median falls at 2002 and 2009. I wont adjust the period length.

What do you think of those methods of calculating compounded returns in earnings? Isn’t it “just as valid”?

What is also interesting is to cross-check these results with figures for ROE and retained earnings. Dividend covers and ROEs are of course never going to be constant throughout the whole period, but I think it’s interesting. Although I haven’t given you data, I have calculated that the median ROE over the 10 year period is 27.1%. The median dividend cover is 4.3. Earnings should therefore have grown by 20.8% pa (= 27.1 * (1 – 1/4.3)). That figure was too optimistic.

John, just FYI the link on page 4 apparently is not working.

Dear Logan James

I downloaded the document and double clicked on the link and it worked. Then there is a link inside that post which is this one:

http://csinvesting.org/2011/09/18/buffett-lecture-at-uf-1998-on-life-and-business-analysis/

The link takes you to the Buffett lecture.

Sorry you had trouble.