(To have your question answered on the blog, email Geoff.)
Someone who reads the blog emailed me this question:
“GuruFocus provides data on predictability of a business. Do you like this metric? Do you use this metric in your analysis? It seems to me that the more predictable a company’s historic earnings, the easier it should be to calculate the intrinsic value of the company. Do you agree with this assessment?
The problem with this for value investors is predictable businesses tend to not get nearly as mispriced (at least highly followed large cap stocks).”
I do use predictability in a general sense. I don’t want to talk about GuruFocus’s measure of predictability specifically – because GuruFocus publishes some of the stuff I write. So, I’m biased. If I say something good about a GuruFocus feature – you won’t believe me. And I wouldn’t want to say something bad because they’re kind enough to publish my writing. I’ll just say that “predictability” is a good measure to look at. And that is what the GuruFocus predictability rank is trying to do. So, it’s trying to do a good thing. And you’re not going to come to any harm by looking at it. And you might get something positive out of looking at GuruFocus’s predictability score for a company. So, yes, I like the metric of predictability.
Now, what do I use personally? I enter a company’s financial data into Excel myself. I’ll look at the data at places like GuruFocus. But, I want to go to the historical 10-Ks and pull the numbers out myself and adjust them as I see fit. This is to get comfortable with the numbers. There’s a difference between when you are relying on something a computer has done and when you are doing the data entry yourself. So, I set up an Excel sheet with 15 years or 20 years or 30 years of financial data. I prefer 30 years where 30 years of data exists. I also like to read the very oldest and very newest annual report from the company. Sometimes I read other specific past annual reports (like the 2008 financial crisis year) that were unusual for the company, the industry, or the country it operates in.
One thing I have Excel calculate is the variation in EBIT margin. So, if you have say 30 years of financial data for EBIT, and the EBIT margin is positive in every single year – you’ll be able to calculate both the standard deviation and the (arithmetic) mean in the series. Excel can give you other types of averages too. It does geometric and harmonic means which investors use rarely. But something like the harmonic mean is actually a useful measure for the very long-term return on capital, because as a rule – a company’s compounding in its intrinsic value will not be less than the long-term harmonic mean of its return on capital. It could – especially if it’s in a very cyclical industry – be much lower than the arithmetic mean. I bring this up because it’s sort of related to “predictability”. Investors don’t pay much attention – except to “recency” – to the order a company’s results come in. But, the order of results is important for things like compounding. It’s not important if the variation in ROC is close to the mean. Let’s say I’m looking at ROC and I have a 30-year series of ROC figures with – Minimum: 10%, Maximum: 20%, Median: 15%, Mean: 15%, Standard Deviation: 5%. That’s a very predictable company. The median and mean are the same. We can scale the standard deviation to the mean (5% / 15% = 0.33). Variation of 0.33 in ROC is low. Not many companies have such low variation in ROC.
I use the same calculations I just showed you for the company’s ROC history and apply it to the company’s EBIT margin history. I like to look at both return on sales (EBIT/Sales) and return on capital (EBIT/Net Tangible Assets). When you are buying into a company – you must remember that earnings aren’t actually that stable. As a rule, things like assets and sales are going to vary less than earnings. That’s as a rule. It’s not totally true. There are cases where a company has several good business units and one money-losing or breakeven business it’s just running as a legacy. In that case, it could dispose of a lot of assets and eliminate a lot of sales and yet not make much of a dent in earnings. But, if we’re talking about the core business – the good business – a change in assets or sales from year to year is going to be smaller than a change in earnings.
Let’s talk about the danger of using the two ratios value investors mention most: P/E and P/B. Earnings are volatile. So, P/E isn’t a good measure. And then P/B is a net number. Book value – or tangible book value (you should always be using tangible numbers for equity and assets) – is net of the liability situation. So, it’s a heavily leveraged number. When a company has low liabilities, P/B could be a pretty stable and solid indicator because it’s basically like Price/Tangible Assets. But, in cases where the company has an “iceberg” type balance sheet where it has $100 a share in assets and $90 a share in liabilities and a stock price of $8 a share – yes, it’s selling for a price-to-book ratio of 0.8. But, an 11% decrease in assets would wipe out all of the company’s equity. You see the problem. If we look at things like sales and assets – those are much more stable.
I also do something else people think is really weird. I look at gross profit. Gross profitability is an important number for me. When I say gross profitability, I mean Gross Profits / Tangible Assets. So, it’s possible a supermarket – which tends to have a ton of operating leverage, it’s a high unit volume business – could have a 120% gross return on assets (Gross Profit/Total Assets = 1.2) and yet it only has a 20% pre-tax net return on assets (EBIT/Total Assets = 0.2). That can actually happen with a supermarket. So, why does gross profit matter? It has to do with things like scale, business organization, who is running the company at the top, how much are they getting paid. Things like that. For example, Village Supermarket (VLGEA) is a relatively small (by number of stores) Shop-Rite operator in New Jersey. It is run by members of the family that control the company’s stock. They are well compensated. It’s possible that there are years where the company makes say $30 million in EBIT and yet pays $5 million to top executives of a company with only about 30 stores. Now, if another New Jersey Shop-Rite operator (Wakefern co-op member) acquires Village, it’s not going to pay management $5 million a year to run 30 supermarkets. I’m not sure it’s even going to pay a combined $1 million to however many people you have running a 30-store operation. An acquirer already has general counsel, a CFO, a CEO, a COO, a board, etc. So, there’s $5 million right there that can be severely trimmed. And I don’t know how lean Village runs its head office, its IT, etc. I can better judge the stores. So, collecting data on the store level – and the gross profit level – is more useful. I’m not sure that high gross profitability is ever a “green flag”. But low gross profitability absolutely is a “red flag”. Companies talk about synergies all the time. They talk about plans to turn a business around and improve margins and so on. Sometimes these plans are realistic – sometimes they aren’t. But, I’ll tell you right now – they’re a lot more realistic if they plan to increase operating profit relative to gross profit. They are a lot less realistic if they think they are going to raise gross profit as a percent of total assets or total sales. If you have 10 firms in an industry with a combined $10 billion of assets and $2 billion of gross profit – that’s a bad industry. And it’s likely to be a bad industry even if you consolidate down to just 8 firms, 6 firms, or even 4 firms. The economics of that business have something severely wrong with them. The ratio of just 20 cents of gross profit for every 1 dollar of assets in service is simply unacceptable.
What does this have to do with predictability? It’s easy to predict an industry like that will have long-term problems it can’t fix. On the flip side, if you have a company with good gross profitability year after year for decades and yet net profitability is only now looking good – that’s something you might want to buy. The reason for this is that when we’re talking about predictability – we’re talking about persistency. How durable are sales, assets, gross profits, net income, book value, etc. Things like sales per share, assets per share, and gross profits per share are more persistent and more predictable from year to year. Now, some investors are going to say: so what? As an investor, you make money from free cash flow. Free cash flow is usually somewhat approximated by net income (it can be a very approximate relationship) and so the P/E ratio is telling you what kind of dividends and share buybacks a company can give you. That’s why a P/E ratio can tell you what kind of returns to expect in the future. I’d agree with that line of logic if you applied it to an entire nationwide market. If we’re talking the S&P 500. Then yes. First of all, the P/E ratio still doesn’t matter. But a normalized P/E – like the Shiller P/E does matter. But that’s because we know the quality of the S&P 500. It doesn’t get better or worse. It’s too generic a mix of businesses to be worth buying on anything but price.
A company can be worth buying for its predictability though. Some industries have more persistent profits than others. When I was writing my newsletter, Quan (my co-writer) and I gave every single stock we looked at an industry category code. So, we’d code an ad agency like Omnicom as “business services” and a fast food joint like Greggs as “restaurant” and a helicopter hoist maker like Breeze-Eastern as “capital goods”. You can find research on the relative persistency of profits in industries like business services, restaurants, and capital goods. Generally speaking, the lower your customer retention, the lower the purchase frequency, and the further you are from the end consumer – the worse the persistency of your profitability is. Now, this isn’t always true. Market structure matters too. Breeze-Eastern was a duopolist with a 50% or higher share of search and rescue helicopter hoists. It was serving an oligopoly (about 5-6 helicopter makers). As a rule, a duopoly that serves an oligopoly is going to have highly persistent profits compared to some other market structures. For example, the beverage can manufacturing business has persistent profits. In a given market, you generally have like 3 companies: Crown (CCK), Ball (BLL), and someone else serving a small number of customers (Coke, Pepsi, Coors, etc.). There’s probably a huge market for smaller plants serving smaller customers. But the biggest customers need someone who can do huge volumes at one plant. So, you have only a small number of customers who need huge volumes at a single plant and only a small number of competitors who are willing to produce huge volumes at a single plant for a single customer. So, you have persistency in that kind of business. I think BWX Technologies (BWXT) should have persistent profits. It’s the monopoly provider of all but one (Curtis-Wright makes the one) nuclear components for reactors in U.S. Navy submarines and aircraft carriers. It also has a monopoly position in uranium down blending and some other stuff related to the U.S. nuclear weapon program. I think it’s persistent because I don’t think the U.S. Navy will – within the next few decades at least – want to stop producing nuclear powered submarines or aircraft carriers and I don’t think they can possibly choose any other provider besides Babcock for what they need. So, Babcock has a monopoly position.
The best kind of persistence comes from high customer retention. I like the ad agency business, the auto insurance business (although I’m convinced driverless cars will obsolete this business over the next few decades), and banks because the customer retention rates for an ad agency can be 95%, for a bank they can be 90%, and for an insurer they can be 80%. If ad agencies, banks, and insurers didn’t retain the same clients – if they had to market aggressively to win new business, I’d have no interest in them. The lack of price competition to retain customers is what attracts me to this business. Now, there is very intense price competition in insurance – especially to win a new customer. In fact, auto insurers lose money on new customers at first because they spend so much on marketing. Once they have a customer though, retention rates are high. There are people who have been renewing with the same insurer for 25-50 years. Ad agencies are even better. There are tons of big brands that have been with the same creative agency for more than 25-50 years. A couple have been with the same agency for 100 years.
So, yes. I do look for predictable companies. I look at long-term variation (standard deviation / mean) in return on capital and EBIT margin. GuruFocus also has predictability rankings. You can use those too. Think a lot about how persistent profits tend to be in each kind of industry. Know that EBIT margins at a restaurant are a lot more predictable than EBIT margins at a capital goods company. And remember that things like assets and sales are more persistent than earnings. Gross profit is also more persistent than net profit. So, start by looking at long-term trends in sales, assets, and gross profit. For example, has asset growth been consistent? Or has it been accelerating in recent years? Try not to buy into a company that is growing its assets more rapidly than everything else. High asset growth right now is usually a bad sign for ROA in future years. Growing profits faster than assets in recent years is usually a better sign.
I’d focus on persistence on measures like ROC and margins rather than growth itself. Consistent growth is good. But, a 7% annual growth rate that is consistent and accompanied by a high ROE is all you need. You don’t need 12% annual growth. What you need is consistent growth accompanied by a high ROE.