Geoff Gannon February 24, 2020

The “Element of Compound Interest”: When Retaining Earnings is the Key to Compounding and When it Isn’t

In my first two articles about Warren Buffett’s letter to Berkshire Hathaway shareholders, I talked about Berkshire’s year-by-year results as a stock and about Warren Buffett’s approach to holding both stocks and businesses. Today, I want to talk about a very interesting section of Buffett’s letter that doesn’t (directly) seem to have all that much to do with either Berkshire or Buffett. This section starts with the name of a man Buffett has mentioned before “Edgar Lawrence Smith”. It also mentions a book review Buffett has mentioned before. In 1924, Smith wrote a book called “Common Stocks as Long Term Investments”. Keynes reviewed that book. He said two very interesting things in that review. The one Buffett quotes from this year goes:

“Well-managed industrial companies, do not, as a rule distribute to the shareholders the whole of their earned profits. In good years, if not in all years, they retain a part of their profits and put them back into the business. Thus there is an element of compound interest operating in favor of the sound industrial investment. Over a period of years, the real value of the property of a sound industrial is increasing at compound interest, quite apart from the dividends paid out to the shareholders.”

This is obviously the most important concept in stock investing. It is the entire reason why stocks outperform bonds over time. Investors – even after this book was published – tend to overvalue bonds and undervalue stocks. Academics call this a “risk premium” for stocks. But, on a diversified basis – it doesn’t make a lot of sense to say it represents long-term risk. It does represent volatility. It also represents uncertainty as to the exact size of performance and the timing of that performance. But, in most years, there really hasn’t been a lot of uncertainty that a 25-50 year old putting his money 100% into stocks will end up with more value when he’s 55-80 than the 25-50 year old putting his money 100% into bonds. The market doesn’t usually undervalue the dividend portion of stocks. Sometimes it does. There have been times – most a very, very long time ago – where you could buy a nice group of high quality stocks yielding more than government bonds (and even less commonly, corporate bonds). To this day, individual stocks sometimes do yield more than bonds. I can think of a few countries (a very few) where you can buy perfectly decent, growing businesses yielding more than the government bonds in those countries (though this is usually due to very low bond yields, not very high dividend yields). And I could think of a few stocks that yield more than some junk bonds right now. But, there’s an important caveat here. The stocks that seem safe and high yielding retain very, very little of their earnings and grow by very, very low amounts. In other words, the element of compound interest is often smallest in the stocks with the highest dividend yields. For example, I happen to know of a perfectly safe seeming super illiquid stock that has often been priced to yield between 8-10% a year. That’s wonderful. But, it doesn’t retain any earnings. And it doesn’t normally grow any faster than the overall rate of inflation. So, your total return in such a stock might be like 10-12% a year if you bought it on an average trading day and held it forever. Well, that’s good. Since Buffett took over Berkshire Hathaway, the S&P 500 has compounded – with dividends included – at 10% a year. So, something that seems very certain to return 10-12% is obviously an adequate substitute for the S&P 500. It belongs in the mix of a stock portfolio. But, while it is clearly a better alternative to bonds – it’s not so clear it’s a better alternative to stocks. A return in the 8-12% range is really about what you could expect historically in stocks that pay out far, far less in dividends than anything like an 8-10% yield.

Investors tend to have a preference for a regular, quantifiable amount of income paid out today instead of a far more irregular, and far less precisely quantifiable amount of much greater income paid out way down the road. This tendency isn’t universal. There are plenty of times – 1929, 1965, 1999, and today – when it pretty much disappears. For example, many of the biggest stocks in the S&P 500 don’t pay out dividends and do have very, very high P/E ratios for stocks so large. There is more pricing in of far distant expected cash flows into the biggest U.S. stocks than has normally been the case. Normally, investors – especially value investors – prefer a certain dividend yield, book value, earnings per share, etc. over something that hasn’t quite hit those numbers.

Are they right to prefer this?

We can do some math on the compounding and see. Although there are some complications, the easiest way to sketch out our expectations in any stock we buy is that our return will be somewhere BETWEEN the earnings yield and the return on equity. Now, I mean this in a “spirit of the law” not “letter of the law” sort of way. You can’t assume that $1 of reported earnings at an ad agency and a cruise line are equally valuable. The cruise line will tend to have less than $1 of cash free to be paid out in dividends, buybacks, etc. when it reports $1 of EPS. The ad agency will tend to have a bit more than $1 of cash actually free to be paid out than the $1 it reports. The earnings of most banks, insurers, and perhaps some real estate companies and such should be thought of in a “comprehensive” sense. Basically, how much is their book value per share (or better yet, if you can calculate it, how much is the fair market value of their assets less liabilities) growing each year – not necessarily how much are they reporting in earnings. But, you get the point. How much richer are shareholders at the end of 2019 than they were at the start of 2019? That is the earnings of the stock. I like to adjust it for how much dilution I expect the company to have due to issuing stock options and granting shares and all that. For S&P 500 companies, that’s running at like 1-2% a year right now.

Given where stock prices usually are – most stocks you buy will have a lower earnings yield than a return on equity. This means you should expect your return in the stock to rise over the years to the extent that: 1) The company retains a lot of its earnings and 2) You hold the stock for a long time. For example, if a company has a 20% return on equity, it reinvests 100% of its earnings (so that the business is actually growing at 20% a year), and you hold the stock for your entire lifetime – you’re going to end up with a return much, much closer to 20% a year than to whatever the earnings yield was when you bought the stock. Right now, it’s probably typical for an investor to buy a stock at about a 5% earnings yield (P/E of 20) and an ROE of 15% and then hope that – over time – the initial 5% return will drift upwards toward the 15% a year ROE limit.

If I’m being honest – I don’t think that will happen for a lot of the S&P 500 going forward. The ROE of the index is probably the least meaningful it’s ever been right now, because the index as a whole has been using all of its reported earnings to pay dividends and buyback shares. You can calculate the value of dividends paid to you (it depends on the tax rate you pay and the rate at which you can reinvest the dividends – that is, your stock picking prowess if you’re 100% in individual stocks). The return on buybacks depends on the price at which the purchase is made. If a company buys back its own stock at a 5% earnings yield and a 15% return on equity – the return on that investment will (like you own return in the stock) be bounded by 5% in the short-run and 15% in the long-run.

There are two really interesting aspects to what Buffett said about this element of compound interest in common stocks. Both have to do with how investors think about compounding in stocks. Here’s the best of what Buffett had to say about that book:

“It’s difficult to understand why retained earnings were underappreciated by investors before Smith’s book was published. After all, it was no secret that mind-boggling wealth had earlier been amassed by such titans as Carnegie, Rockefeller, and Ford, all of whom had retained a huge portion of their business earnings to fund growth and produce ever greater profits. Throughout America, also, there had long been small-time capitalists who became rich following the same playbook.”

What Buffett says is totally true. No one was getting rich in the 1800s and early 1900s in the U.S. simply through clipping coupons or collecting dividends. The richest Americans got that way by retaining extremely high proportions of their earnings in the same enterprise and growing it bigger and bigger. Sometimes, this was commented on. Buffett mentions Rockefeller. Standard Oil’s dividend policy was often questioned and criticized. And the company was believed to be too conservatively run financially. Eventually, there may have been truth to this. But, early on, the retaining of huge amounts of earnings was very useful in being able to take advantage of competitors who did not retain much in earnings and did use a lot of debt. What’s interesting to me, is that in most discussions I’ve ever seen of the dividend policy, cash position, borrowing, etc. of companies like Standard Oil – there’s little or no discussion of the concept of compounding. If I looked through newspapers from the time, I’d probably be able to find such references. But, just working from my memories of books I’ve read of the corporations run by the men Buffett mentions there – the concept of retaining earnings as opposed to paying dividends in order to compound the size of the overall enterprise just isn’t something I remember reading about. Complaints that the dividend policy of these corporations was unusual, stingy, etc. is something I remember reading. Discussion of the concept of compounding isn’t.

Today, the reverse is true. The biggest companies in the U.S. – with a couple exceptions – focus on retaining earnings instead of buying back stock and paying dividends. The investors in them keep talking about compounding despite their size.

This brings us to the last point regarding your long-term return in a stock.

It’s obviously determined by the return on INCREMENTAL retained earnings – not the past return. For example, Berkshire Hathaway’s rate of compounding – both in terms of book value and stock market value – peaked around the late 1990s. It was roughly around the time of the transformative General Re merger that shifted Berkshire’s overall asset allocation out of stocks and into bonds. Berkshire’s best years were from 1965-1998. For some of those years, the rate of incremental return on retained earnings was probably higher than 25%. Today, it’s probably lower than 10%. You could argue that maybe Berkshire’s stock portfolio tended to be overvalued in the 1990s and undervalued now. That Berkshire as a stock was overvalued then and undervalued now. And that book value was more meaningful then and less meaningful now. Even if you assume all those things – I still think you don’t get a tendency to compound incremental retained earnings at any less than 20% a year from 1965-1998. And I don’t think you get a tendency to compound incremental retained earnings for the last ten years, for what the next ten years will likely be, etc. at better than 10% a year.

For that reason, any dividends paid by Berkshire during the 1960s, 1970s, and 1980s would’ve had a disastrous influence on Berkshire’s long-term compound record. If you reduce the weighting you are putting into a higher returning asset (Berkshire in the 1960s, 1970s, and 1980s) and put it into a much lower returning asset (something like the S&P 500 after you’ve paid a tax on your dividends) – your compound result will get much worse. Berkshire’s advantage over the S&P 500 was once something like 15% a year during Buffett’s first 15 years or so. Dividends would be paid regularly over time, so it’s not quite like shifting out of something returning 20-25% a year into something returning 5-10% a year – but, it’s pretty close. It’s worse than making a portfolio allocation decision to shift from stocks to bonds. And, honestly – during Berkshire’s best years – paying a dividend would have the same size impact as like an individual investor shifting from a 100% allocation to the S&P 500 to a 100% allocation to T-Bills. In fact, the difference between those two asset classes has usually NOT been as high as 10-15% a year in lost compounding. Berkshire would’ve destroyed immense amounts of potential shareholder wealth by paying a dividend.

That was true for the first 30+ years of Berkshire’s history under Buffett.

It’s not true today.

There just can’t – mathematically – be much of a difference between Berkshire retaining 100% of its earnings or paying 100% of its earnings out in dividends. Even if you assume Berkshire will compound at as high as 10% a year over the next 10 years (which, honestly, I consider to be on the very aggressive side of what’s humanly possible), most investors don’t think the S&P 500 will do much worse than like a 6% or so return over the next 10 years. Obviously, a lot of them think 10% a year is possible (though I don’t think it is possible from today’s prices).

To use that analogy I made earlier – at worst, we’re talking about a capital allocation decision akin to going from 100% stocks to 100% long-term bonds (a difference of like 5% a year or less). More likely, we’re talking about the difference between a portfolio that’s like 100% stocks and 0% bonds to one that is 60% stocks and 40% bonds. In the long-run, it’s a mistake. In a long enough run, being less than 100% in stocks is almost always a mistake. Over your investment lifetime, any allocation to anything other than stocks is probably dragging down your eventual ending net worth unless you are expertly skilled in timing (you’re a good trader).

Given how expensive stocks are, I still think it’s likely that Berkshire is better off retaining 100% of its earnings than paying 100% out in dividends. However, it’s getting awfully close to a coin flip. And, a few years down the road – because of the anchor on performance created by compounded asset levels – I think it will be a coin flip and stay that way.

There’s one big exception though.

The more earnings Berkshire retains – the more of its own stock it could buy back if it wanted to. Berkshire works a little like a closed-end fund. Less so than it did before 1998. But, it still does to some extent. Berkshire owns listed stocks and reports their changing market value. But, Berkshire is also itself a listed stock. What tends to happen with any listed company that owns other listed companies is similar to what can happen when investors misjudge a cycle. In a misjudged cycle, investors put the lowest price-to-earnings ratio on a stock at the moment where earnings are most depressed and the highest price-to-earnings ratio on a stock at the moment where earnings are at their peak. Something like this did actually happen with Berkshire before the General Re deal. Berkshire’s stock portfolio – things like Coca-Cola – were trading at very high multiples of their earnings, sales, etc. and then Berkshire (which was carrying those investments at those high multiples) also traded at a high price-to-book ratio. This tends to lead to wild distortions in the “look through” earnings multiple. Basically, Berkshire is likely to have an especially low price-to-book ratio at the same moment that its book is likely to be UNDERSTATED relative to intrinsic value (due to carrying stocks at their market values – not at their intrinsic values).

As a result, Buffett might get the opportunity to spend a lot of Berkshire’s retained earnings buying back the company’s own stock when its price-to-book ratio is low and its stock portfolio is undervalued in the market.

The math on that works well. It’d be fairly easy for Berkshire to generate high returns on the portion of cash it uses to buy back its own stock in a market downturn. The catch here wouldn’t be the return on incremental capital deployed in this way. The catch would be the weighting. It’s usually pretty hard for companies to continually devote large amounts of capital to buying back their own stock. One, the buy back can help push up the stock price – and so, be self-defeating as a use of capital. Two, the stock can simply fail to stay cheap enough for long enough. This is more of a problem for illiquid stocks than for stocks on an exchange. Big, listed stocks like Berkshire tend to have pretty high share turnover – so, if they devote as much capital as they can to buying back the maximum allowable or prudent level of volume each month, they can buy back a meaningful part of the market cap. Illiquid stocks don’t turn over enough to make this possible. So, this is a rare case where size is oddly helpful. Having to spend billions on buybacks to move the compounding needle is unhelpful. But, the advantage of size as a stock is that you have a lot of passive owners, short-term owners, computers, etc. basically trading your stock. It’s a lot easier to find shares to buyback in the open market. There just isn’t much of a long-term, active shareholder base hoarding your stocks when it’s cheap. Instead there are more shareholders who are shorter-term oriented and less sensitive to the price you’re offering. The fact they think they can get out and back in again becomes an advantage to you in buying back stock – because you only need to perform half the operation they do. You’re just buying the stock. You never need to sell it again.

Finally, there is one obvious problem Berkshire might face in trying to commit enough capital to a buyback – Buffett. Berkshire’s best chance to buy back a ton of stock would obviously be if the decision was made by someone other than Buffett. That’s not going to happen while Buffett is running the company. So, there is the potential problem that a Berkshire buyback ordered by Warren Buffett would act as a signal that he believes the stock is cheap. Buffett is considered to have very good investment judgment. He knows more about the value of Berkshire than anyone. And, historically, he’s been very reluctant to buy back stock. A really big buyback could quickly defeat any chance of continuing to allocate a lot of capital to Berkshire shares, because Buffett isn’t Singleton. At Teledyne, Singleton wasn’t really all that famous. And the idea of buybacks was very poorly understood. Buffett is incredibly famous and buybacks are now very, very common. Buffett has a huge disadvantage versus Singleton. So, he’s unlikely to be able to commit as much capital as he’d like to any buyback, because the simple act of Buffett aggressively buying back shares could actually move the market in the wrong direction for him to continue getting a good return on his investment.

Become a Focused Compounding Premium Member

Get a Free Weekly Email from Geoff (enter your email in the sign-up box)