Defensive Value Investing: A 20-week course

“The price paid for each [company] should be reasonable in relation to its average earnings for the last five years or longer.”

Benjamin Graham

Using the tools we’ve looked at so far, relatively defensive companies should be easy to find. Just pick a company, check for a ten-year unbroken dividend record and calculate its growth rate, growth quality, profitability and so on. If it has consistent inflation-beating growth, powered by high levels of profitability and little debt, it could be worth investing in.

However, I have said nothing about price or value, and the title of this book is The Defensive Value Investor, not The Defensive Investor.

It’s no good finding a fantastically defensive company if it currently trades on a PE of 50 and has a dividend yield of 1%. In that case even if the company quadrupled in size over the next decade the share price might not go anywhere.

How does this work? How can corporate growth not result in a rising share price?

Imagine a company that investors thought was about to grow quickly. In that case they might accept a smaller earnings yield and dividend yield today, so they’re willing to accept the PE of 50 and dividend yield of 1% mentioned before. Let’s assume the company’s share price is 100p while earnings are 2p and the dividend is 1p.

After ten years the company has done exactly as investors expected. Its earnings have grown to 8p and the dividend is up to 4p, so its financial results have quadrupled in a decade, which requires a growth rate of about 15% a year. However, the company’s growth spurt is over and investors now expect growth in line with inflation, perhaps 2% a year or thereabouts.

With such paltry future growth investors may now only be willing to give the company a PE ratio of 12.5 and a dividend yield of 4%, both of which are fairly normal values for slower growth companies.

Surprisingly enough, with those valuation metrics the share’s price would still be 100p; unchanged from its level a decade before, despite the underlying company growing at 15% a year.

The lesson from this example is hopefully clear: the price you pay for a company is every bit as important as the company’s growth rate.

In this chapter I’ll work through how I think about price and value and in particular a couple of twists I apply to the standard PE ratio and dividend yield.

Problems with the PE ratio

When it comes to valuing companies, investors generally stick to using simple valuation ratios. The most common is the price to earnings (PE) ratio mentioned above, which compares the current share price to last year’s earnings per share.

The price to earnings ratio is calculated as follows:

PE = share price / earnings per share

Value investors will often look for low PE ratios and at first glance that makes a lot of sense. A low PE ratio means a high earnings yield and that means more earnings for each pound invested. Since earnings are the starting point for dividends, and are the source of much internal investment within companies, getting a higher earnings return on your investment is generally a good thing.

However, there are several problems with the standard PE ratio, the biggest of which is its focus on a single year’s earnings.

The problem here is that earnings can go up or down dramatically and unpredictably from one year to the next, even for quite defensive companies. For example, the earnings of Aviva, the insurance giant, went from 45.4p in 2004 to 86.6p in 2006 for a 91% gain in two years, but by 2007 they had dropped back again by 44%.

Does that mean Aviva was worth 44% less in 2007 than it was in 2006? I don’t think it does. I don’t think a property investor would say a house had halved in value just because rental income had halved in a single year (perhaps because of a problem tenant). A sensible investor would say that one year was a blip and that the true ability of the house to produce an income was more or less unchanged.

As an equity investor I want to think about companies in much the same way that property investors think about property. I want to ignore short-term ups and downs and instead think about the long-term value of a company rather than its earnings over a single 12-month period.

A PE ratio for long-term investors

The solution is simple. Instead of comparing price to last year’s earnings, it is better to compare a company’s share price to its average earnings over a number of years. That’s because a company’s average earnings over the previous ten years (known as a ten-year moving average) will change much less from one year to the next than a single year’s earnings.

For example, a company’s 2015 earnings could easily be double or half its 2014 earnings. However, if the company has been profitable in every year for the past ten years then its average earnings from 2004-2014 are unlikely to be very different to its average earnings over the 2005-2015 period. The stability of long-term average earnings gives us something significantly more robust to compare share prices against than a company’s most recent one-year earnings.

It was Professor Robert Shiller of Yale University who popularised the idea of comparing current prices to long-term average earnings, although the concept dates back at least to Benjamin Graham, who originally wrote about it in the 1930s.

Shiller’s approach is to use the ten-year inflation adjusted earnings average in a ratio which is now known as the cyclically adjusted PE (CAPE). By applying CAPE to the S& P 500 index (the US large-cap index), Shiller was able to show that the US market in the year 2000 was clearly expensive and likely to fall back towards historically normal valuation levels. At the time many investors were caught up in the dot-com mania and thought the market would just keep going up and up.

The decade of subsequent poor returns from that index (and the FTSE 100 as well, which was similarly overvalued) are strong proof that Shiller and his valuation tool were right. In other words, CAPE gave investors a broadly correct indication of long-term returns, which is something the standard PE ratio rarely manages to do.

Figure 6.1 shows the difference in volatility between real (inflation adjusted) annual earnings and cyclically-adjusted earnings (i.e. the ten-year average) for the S& P 500 index. I’m using that index rather than the FTSE 100 because its earnings data goes back over 100 years, which will help to show these volatility differences much more clearly.

INSERT IMAGE FROM BOOK (FIGURE 5.1)

Figure 6.1: S& P 500 earnings volatility over more than a century

Measured on an annual basis, earnings move around a lot. On the other hand, cyclically-adjusted earnings are far more stable, reflecting the relatively stable earnings power of the 500 largest companies in the US economy. This gives us a much steadier number to compare price against, and that applies whether we’re talking about the S& P 500, the FTSE 100 or even a single company (although this approach is most suited to relatively defensive companies).

Valuation mean reversion

The next valuable insight from CAPE is that it is a mean reverting ratio, which means it tends to head back towards its long-term average figure given enough time. Occasionally it will wander off to extreme highs (like 1999) or extreme lows (like 2009), but for the most part CAPE stays fairly close to its long-term average.

Figure 6.2 shows this mean reverting tendency for the S& P 500.

INSERT IMAGE FROM BOOK (FIGURE 5.2)

Figure 6.2: Mean reversion of the S& P 500 CAPE ratio

The index’s CAPE valuation has repeatedly swung back and forth, from above to below its long-run average, for more than a century. This tendency to mean revert is caused, at least in part, by what is known as the replacement cost of businesses.

As share prices increase it eventually becomes much cheaper to own a company (perhaps a widget manufacturer) by creating a new one from scratch than it is to buy an existing one through the stock market. This increases the supply of new companies and reduces the number of stock market buyers, both of which tend to lower share prices. The result is an effective limit to how high share prices can go.

The same principle applies when shares are very cheap. At some point it will make more sense to buy an existing company cheaply on the stock market than to start one up from scratch. This decreases the supply of new companies and increases the number of stock market buyers, both of which tend to raise share prices. The result is an effective limit to how low share prices can go.

In between those two extremes the market will bounce around randomly, driven by news and the opinions and emotions of investors.

Figure 6.2 shows that for the most part, investors are happy to buy and sell when CAPE is somewhere between 10 and 20. When the market moves well away from that range it eventually falls back towards its average value of about 16. It may take a few years to do it, but if history is anything to go by the odds are that this mean reversion will continue long into the future.

Mean reversion is an important idea because it suggests that if you buy an index when its price is well below its long-term average CAPE ratio, the ratio is more likely to move upwards and back towards its long-term average over the next few years.

Since the ten-year average earnings figure is relatively stable this increase in the CAPE ratio is more likely to come from a rising price than a decline in average earnings. If this occurs the upward movement in price will give a boost to shareholder returns (not to mention the higher yields which are also typically available at lower CAPE ratios). In other words, valuation mean reversion is the engine that drives the buy low and sell high tactic, which is at the core of any value investment strategy.

Although CAPE and the notion of comparing price to long-term average earnings is usually only applied to market indices such as the S& P 500 and FTSE 100, I think these ideas can just as effectively be applied to relatively defensive individual companies.

Calculating a long-term PE ratio

While CAPE is based on inflation-adjusted earnings, I think that virtually the same benefits can be achieved without adjusting earnings for inflation, which makes the ratio that bit easier to calculate.

The non-inflation-adjusted version of CAPE is usually known as PE10 and it is calculated as follows:

My rule of thumb for PE10 has as its starting point a quote from Benjamin Graham:

“We would not recommend a price exceeding twenty times [the company’s five-year average] earnings. This would protect the investor against the common problem of buying good stocks at high levels of the general market. It would also bar the purchase, even in normal markets, of a number of fine issues which sell at unduly high prices in anticipation of greatly increased future earnings.”

For our purposes Graham’s formula of 20 times the five-year earnings average needs a slight tweak. I use the ten-year earnings average, which for growing companies will typically be lower than the five-year average, and more so if the company is growing quickly. There is no magic correct number, but I have found that companies trading at more than 30 times their ten-year average earnings are unlikely to fit the description of a value investment.

On that basis I use the following rule of thumb:

I’ll calculate this PE10 ratio for a couple of real-world examples in a moment, but first I want to show you how these ideas about cyclically-adjusted earnings can be applied to dividends as well.

A dividend ratio for long-term investors

Much of our analysis into the quality and defensiveness of a company has been based on its dividend track record, where we have primarily focused on how consistently and quickly the company had been able to increase its dividend payments to shareholders.

While the consistency and speed of dividend growth are clearly important to an investment’s total return, just as important is the dividend yield. The dividend yield can be measured using historic dividends or a forecast of future dividends. I prefer to use historic dividends so here’s my definition of the (historic) dividend yield:

dividend yield = last year’s total dividend / current share price

This is the dividend yield that you’ll see quoted on almost every investment website. It’s also available from just about every data provider, including the ones listed in the appendix.

However, the dividend yield suffers from many of the same problems as the PE ratio. Perhaps not to the same extent as dividends tend to be more stable than earnings from one year to the next, but the basic problems that come from focusing on a single year’s results are still there.

As with earnings, if a company’s dividend is cut by 50%, does that mean the whole company is worth 50% less than before? Perhaps in some cases it does, but for high quality and relatively defensive companies I think in most cases it does not. A more likely story is that the company has hit some sort of short or medium-term problem that requires a dividend cut in order to rectify it. In my experience the dividend will often recover, although it may take several years.

That’s not to say I don’t look at the dividend yield, because I do, but it is not my primary dividend ratio.

I think a better way to find shares that are cheap relative to their past dividends is to use the dividend equivalent of the PE10 ratio, i.e. the PD10 ratio. This is exactly what it sounds like; it’s the ratio between a company’s current share price and its long-term (ten-year) average dividend. It provides a much more stable ratio than the standard dividend yield.

Calculating a long-term dividend ratio

To calculate the PD10 ratio you’ll need the company’s per share dividend payments (DPS) for the last ten years. The steps are detailed below:

My rule of thumb for this ratio is:

This maximum value of 60 comes from combining the PE10 maximum of 30 with the idea that a reasonable long-term level of dividend cover for most companies is two (dividend cover is the ratio between earnings and dividends per share and is calculated as EPS divided by DPS). So a company where earnings consistently covered the dividend twice over (i.e the dividend cover was two), and whose share price resulted in a PE10 ratio of 30, would also have a PD10 ratio of 60.

As with the PE10 maximum of 30, there is nothing magical about the PD10 maximum of 60 which says that shares below that level are cheap and those above it are expensive. However, shares trading beyond that sort of level are far less likely to be cheap and are more likely to be either expensive or relying on rapid future growth to justify the current price, and I am not interested in investing in either of those situations.

Stocks with PD10 ratios above 60 are also more likely to have below average dividend yields as well, which will be of interest to more income-focused investors.

Let’s take a look at some examples.

BAE Systems’ valuation ratios

POSSIBLY CHANGE THIS SECTION TO USE A CURRENT HOLDING RATHER THAN BAE.

These valuation ratios are simple to calculate, so applying them to a real company will only take a few moments. In this case I’ll apply them to BAE Systems, a FTSE 100 company that operates in the Aerospace & Defence sector. It is the UK’s largest manufacturing company. As at the 2014 results, BAE had a growth rate of 6.4% and growth quality of 81%, both of which are above average.

You can see BAE’s earnings and dividend per share over the last ten years in Table 6.1.

INSERT TABLE 5.1 FROM BOOK (OR UPDATE TO USE A CURRENT HOLDING)

Table 6.1: BAE Systems’ results to 2014

In addition to earnings and dividends per share for the last ten years, we also need the share price. For this example I’ll use the price as it was in May 2015, which was 509p. We can calculate BAE Systems’ PE10 ratio using the previously described steps:

Calculating PD10 is more or less the same:

Both BAE’s PE10 ratio and PD10 ratio are comfortably below my rule of thumb maximums of 30 and 60 respectively, so according to those rules BAE wasn’t obviously overvalued at 509p.

ARM Holdings’ valuation ratios

UPDATE THIS SECTION TO USE A CURRENT HOLDING OR SIMILAR WITH UP TO DATE FIGURES

ARM Holdings is a FTSE 100-listed technology company that develops and licences technologies which are included in many modern gadgets such as smartphones and tablets. As at the 2014 results the company had a growth rate of 23.1% and growth quality of 89%, both of which are among the highest of any company in the FTSE All-Share.

You can see the company’s earnings and dividend per share over the last ten years in Table 6.2.

INSERT TABLE 5.2 FROM THE BOOK (UPDATED FOR WHATEVER COMPANY YOU’RE USING)

Table 6.2: ARM Holdings’ results to 2014

Once again we need the share price and for ARM Holdings I’ll use the price as it was in August 2015, which was 865p.

We can calculate ARM Holdings’ PE10 and PD10 ratios using the same steps as before:

ARM Holdings was clearly trading way outside my rule of thumb maximums, indicating that the shares were not obviously cheap.

But saying the shares were not obviously cheap is not the same as saying they are expensive. If the company continued to grow at 20% plus per year for many years (as it had done in the past) then it may eventually justify those lofty valuation ratios. However, relying on rapid future growth to justify a company’s current share price is the province of growth investors, not defensive value investors.

This means that as things stand today, I would probably not consider investing in ARM unless its share price dropped to around 200p, which seems unlikely at the moment.

In the next chapter we will look at several ways to combine the various defensive factors such as growth rate and growth quality, with the value factors of PE10 and PD10, in order to find shares that have a good balance of both defensiveness and value.

But first, here are those valuation rules of thumb once again.