Have you ever run a back-test for net-nets in Japan? Buying profitable companies selling for less than cash and paying a dividend would make Graham's mouth water, it seems like a no-brainer. But given the continued depressed state of the Nikkei, has net-net investing worked there over a 10-year period? It's hard to come to terms with a poor macro environment like Japan — but on the same hand, the screaming deals would not exist if it were not for the problematic environment. How do you come to terms with that? I know net-net investing isn't a strategy based on macroeconomic timing or considerations, but doesn't it still make you question the viability of the strategy? Local evidence (i.e., backtesting) could help assuage that.
No. I haven’t backtested a net-net strategy in Japan. I don’t have access to good historical data in foreign countries.
No. It doesn’t make me question the viability of the strategy.
Net-net investing worked in actual practice (always better than a backtest) in the 1930s and 1940s in the U.S. The situation in Japan is similar, though worse (for investors).
I’m going to take some time to explain how bad Japan’s economic growth has been for the past two decades. People know it is bad. I think they underestimate just how bad it is – because they don’t realize how unusual a long-term poor performance like this is. A short-term poor performance – The Long Depression, The Great Depression, The Great Recession, etc. – is common. And for long-term investors, depressions are pretty forgettable. From a bird’s eye view showing a period of decades, these economic catastrophes barely register.
Japan’s economy has had it worse these last 20 years than the U.S. did from 1929-1949. You can see this in their much lower real GDP per capita growth rate. I think real GDP per capita growth in Japan has been about 0.9% a year over the last 20 years. The entire period from 1929 to 1949 was actually not bad for the U.S. Real GDP per capita rose about 2.2% a year. The last 20 years you just lived through was worse. Real GDP per capita only grew 1.5% a year from 1991 to 2011. So, the 1930s and 1940s were better than the 1990s and 2000s if you’re willing to put aside mass unemployment, genocide, world war, atomic bombs, etc.
Those things take center stage when you are writing world history. They are not that important when discussing the kind of returns investors will realize over a couple decades in the stock market. Only two things matter there: the price you pay and the value you get. We can break down the way price and value determine your returns by looking at four factors:
· Earnings yield (price)
· Return on investment (profitability)
· Sales growth (growth)
· Dividend yield (dividends)
From a long-term investment perspective, any period of a couple decades taken together doesn’t stand out much. We can’t look at the long-term performance of the economy starting from the most recent financial panic (since it's only been a short time since 2008). But we can take a look at a couple past economic catastrophes.
· The Great Depression (20-year real annual GDP per capita growth: 2.2%)
· The Long Depression (20-year real annual GDP per capita growth: 2.0%)
As you can see, measuring real GDP per capita for a full 20 years after starting in the supposedly peak years of 1929 and 1873 doesn’t do much to the long-term growth picture. It certainly doesn’t get us anywhere near Japan’s dismal last two decades.
Buying net-nets after the 1929 crash and before the 1950s certainly worked. We know that. And we have a very complete record of the kinds of things Ben Graham bought in the 1930s and 1940s. But, again, we need to remember that real GDP per capita growth was 2.2% a year over those two decades. In Japan, real GDP per capita hasn’t grown even half as fast.
And, of course, the rate of growth in capitas is even worse in Japan. Let’s take another look at The Great Depression and The Long Depression. But this time let’s look at the rate of growth in people rather than GDP per person:
· The Great Depression (20-Year annual population growth: 1.0%)
· The Long Depression (20-Year annual population growth: 2.3%)
And now Japan’s last two decades:
20-Year annual population growth: 0.2%.
So, the long-term growth numbers – before we even think about inflation – look something like this:
(Real GDP Per Capita + Population Growth = Real GDP Growth)
· The Long Depression (2.0% + 2.3% = 4.3%)
· The Great Depression (2.2% + 1.0% = 3.2%)
· The Lost Decades (0.9% + 0.2% = 1.1%)
So, yes, the business situation in Japan – over the last two decades – has been much, much worse than it ever was in the U.S. The U.S. never experienced a long-term slowdown as bad as Japan has experienced these last 20 years.
No growth habits have formed. There was never time for these to form in American depressions. Things got very bad. But they never stayed very bad for very long.
In Japan, perhaps things never got as bad. But they stayed much, much worse. The net result was obviously much worse for anyone young enough to live through the next couple decades.
So, Japan is stagnant in a way the U.S. was never stagnant.
That creates a problem. Remember, the return you get on a stock is going to be constrained by four things:
· Earnings yield (price)
· Return on investment (profitability)
· Sales growth (growth)
· Dividend yield (dividends)
Of these four things, dividends are the least necessary. A stock that pays no dividends can give you very good returns. And a stock that doesn’t grow – but pays a lot of dividends – can give you very good returns too. But a dividend is a return. So, the amount of value you need the retained earnings to add – through a combination of the initial earnings yield you buy the stock at, the sales growth and the return on investment – is lower to the extent a stock’s dividend is higher. And vice versa (the lower the yield on the stock – the higher its earnings yield, growth and ROI need to be).
It’s important to keep these four things in mind. But it’s also important to remember that it is the future of these items that really matters. I’ll be talking about today’s P/E ratio, today’s dividend yield, today’s return on investment, etc. But if earnings, dividends or growth is unsustainable, the numbers will give you a misleading impression of what the stock can return tomorrow.
Okay, let’s look at a group of five Japanese net-nets.
I’m going to just use Bloomberg data for this. So, the numbers may not be exactly right. But they will be consistent. The important thing is to look at these stocks as a group.
Earnings Yield: 9.84%
Dividend Yield: 2.22%
Earnings Yield: 13.31%
Dividend Yield: 3.92%
Earnings Yield: 11.88%
Dividend Yield: 2.83%
Earnings Yield: 10.15%
Dividend Yield: 3.77%
Earnings Yield: 7.91%
Dividend Yield: 2.62%
Now, the average earnings yield for these five Japanese net-nets is 10.62%. That’s a P/E of 9.42. The average dividend yield is 3.07%.
Japan’s inflation rate is negative 0.7%. The U.S.'s inflation rate is positive 2.9%. So, you need to add 3.6% to all Japanese yields to get the equivalent (in real terms) yields in the U.S.
In other words, for a group of U.S. stocks to have the same real dividend yield as Natoco, Excel, Kitakei, Chuokeizai-Sha and Otec, they would need to have dividend yields of 5.82%, 7.52%, 6.43%, 7.37% and 6.22%, respectively. The average real dividend yield of these five Japanese net-nets is 3.77%. For a U.S. stock to have a real dividend yield of 3.77% it would need to yield 6.67%.
Very few U.S. stocks currently yield 6.67%. And those that do are much more heavily indebted than these Japanese net-nets. Remember, some of these net-nets have negative enterprise values despite paying dividends. That means they don’t just have more cash than debt. They have so much more cash than debt that the surplus of their cash over their debt is greater than their stock price.
Okay. So, to deliver an equivalent real value to investors through its dividends as these Japanese net-nets do, a U.S. stock would need a dividend yield of 6.67%.
What would its P/E ratio need to look like?
The average earnings yield of these five Japanese net-nets is 10.62%. Again, Japan’s inflation rate is negative 0.7%. So, the real earnings yield of this group is actually 11.32%. The U.S. has a positive inflation rate of 2.9%. So, a U.S. stock would need an earnings yield of 14.22% to achieve a real earnings yield of 11.32%. Or to put it another way, a Japanese stock with a P/E of 9.42 has the same real earnings yield as a U.S. stock with a P/E of 7.03.
So, when we imagine what these Japanese net-nets would look like if we could move them to the U.S. while keeping the exact same real returns from dividends and earnings – we find that they would have a P/E of 7 and a dividend yield of 6% to 7%.
Does that sound impossible?
Only three or four decades ago, there were lots of U.S. stocks trading at those prices.
If you read “There’s Always Something to Do” about Peter Cundill you’ll remember that he loved magic six stocks: P/E of 6, dividend yield of 6% and 60% or less of book value.
In nominal terms, you won’t find a lot of Japanese stocks – even net-nets – fitting this formula. But, in real terms, you can find stocks in Japan that hit those numbers. Some of them are net-nets.
Of course, Peter Cundill was dealing with a period of much higher inflation. So his magic six-stock criteria was not nearly as demanding as it is today – when interest rates are so much lower.
Let’s look again at the four factors I mentioned as determining your return in a stock:
· Earnings yield (price)
· Return on investment (profitability)
· Sales growth (growth)
· Dividend yield (dividends)
A company’s real dividend yield is effectively a reduction in your hurdle rate. Think of it this way. If you are demanding 6.5%-a-year real returns before you buy a stock and that stock’s real dividend yield is 3.5%, you only need the earnings retained by the company to grow the value of your stock by another 3% (real) a year to provide you with the returns you want. Dividends are neither necessary nor are they some kind of bonus. They are a substitution. To the extent you get paid in cash you don’t need share price appreciation and vice versa.
The average dividend yield of the five Japanese net-nets I mentioned is 3.07%. Since Japanese inflation is negative 0.7%, this is a 3.77% real return.
Historically, U.S. stocks have returned 6.5% a year in real terms. I’m just using Shiller’s data here. It says real returns in U.S. stocks were 6.5% a year over the last 140 years.
So, if we want that kind of return in Japanese net-nets – today a 6.5% return would be about a 9.4% return in nominal U.S. dollars – we need to have share price appreciation in our Japanese net-nets that is at least equal to 6.5% minus the 3.77% real dividend yield we are getting.
That leaves a 2.73% a year gap that needs to be filled through share price appreciation. Can our Japanese net-net provide this?
Let’s look at our “retained earnings yield.” This is a stock’s earnings yield minus its dividend yield. These 5 Japanese net-nets are only paying out about one-third of their dividends in earnings. Over two-thirds of their earnings are being retained. For these net-nets that retention tends to be a buildup of cash not a reinvestment in receivables, inventory and property.
On average, these companies are retaining 7.55% of their stock price each year. So, if one yen retained by these stocks adds one yen in value to these stocks – we would expect nominal share price growth of around 7% to 8% a year.
But how likely is that? How likely is it that these Japanese net-nets actually get full value for their retained earnings?
How low can the value of each dollar – they’re yen, but we’ll pretend they’re dollars here – fall in value when it is retained by the company instead of being paid out?
Well, we need share price appreciation of 2.73% a year to make up for the gap between these stocks’ dividend yields and the kind of returns we want. They are retaining 7.55% of the stock price. So, we need each dollar of retained earnings to add at least 36 cents to the stock’s intrinsic value. Anything less than that – and the value destruction will be too great to deliver a 6.5% real return in our Japanese net-nets.
Do I think 100 yen retained by a Japanese net-net – or more accurately these five Japanese net-nets – will eventually add at least 36 yen to the share price?
If you look at these net-nets, they are adding to cash more than they are adding to property. You have miniscule growth in Japan’s economy. But it is not negative. Maybe we can assume 1% a year in real growth. So, reinvestment possibilities in Japan are very low. And the return on any reinvestment is low.
In that situation, it makes sense to either pay out earnings as dividends or retain earnings as cash. For the most part, those are the best net-nets to buy. The net-nets that:
· Have high dividend yields
· Have high earnings yield
· And keep their retained earnings in cash
Keeping retained earnings in cash avoids the risk of getting stuck in very low-return, long-term investment like additions to property, plant and equipment.
Let’s break down the issues surrounding our investment in Japanese net-nets – and really any investment – by ticking off each of the four factors that decide our annual returns in a stock.
Because we are comparing stocks in two countries – the U.S. and Japan – with very different inflation rates, I’m going to talk in real terms. Remember, that inflation is 2.9% in the U.S. So, when I say a 6.5% real return is our hurdle rate, that’s like saying a 9.4% nominal return is our hurdle rate.
Let’s look at each factor of these Japanese net-nets and how it compares to the rate of return we want to see:
Desired Real Rate of Return vs. Real Dividend Yield: 6.5% - 3.77% = 2.73%
So, we need our other factors – our return on retained earnings – to be at least 2.73% a year in real terms.
Real Retained Earnings Yield: 8.25% / 2.73% = 3.02 (Pass)
Real Return on Investment: ? / 2.73% = ?
Real Sales Per Share Growth: 1%/2.73% = 0.37 (Fail)
So, the amount of retained earnings we have being added to the company’s balance sheet is three times larger than it needs to be to deliver the kind of returns we want. But the real sales per share growth is only about a third of what we want here (it would have to be much higher, but low sales growth is okay when the dividend yield is high).
It seems to come down to the real return on investment. I’m not sure what that will be. I think 4% a year is possible. Not much more than that. In the U.S., real return on investment is much higher at many companies. It’s not unusual to see numbers in the 10% to 12% range right now.
How can I have any confidence in the returns on investment that will be achieved by Japanese net-nets?
I can’t. But two numbers make the ROI/growth hurdles very low. Low enough that I’m pretty sure even Japanese companies can clear them.
If you look at these five Japanese net-nets – they are paying you a dividend yield of 3.07% a year. And then they are retaining another 7.55% of your purchase price on their balance sheet. So, for every 100 yen you use to buy these stocks, you’re getting paid three yen a year in cash. And you’re getting seven to eight yen added to the balance sheet.
The issue of growth and return on investment only arises to the extent these things are necessary to provide you with a high enough return on your purchase price. If the purchasing power of the yen is stable, we are talking about a 10.6% real return in a combination of retained earnings and dividends.
That’s much higher than the real returns investors normally get in stocks.
Now, yes, the 7.5% part of that 10.6% real return is vulnerable to value destruction through reinvestment at very low rates of return.
If these low rates of return are, say, 4% a year – we’re fine. You’d still end up with real total returns in the stock that are higher than what investors in stocks have usually earned.
You can see this by assuming that each part of the return on retained earnings must be the same number. For example, a 6.5% earnings yield that is invested at a 6.5% return on investment and achieves 6.5% growth in sales per share will hit a target of growing intrinsic value by 6.5% a year.
A 4% ROI is low. But it’s enough to clear the less than 3% hurdle that’s left after we take these stocks’ dividends into account.
Let’s put dividends aside and just talk about why ROI, growth and a stock’s earnings yield all matter. But they don’t matter individually. They matter together.
So what if one of those numbers is lower than your hurdle rate?
The others have to be higher than your hurdle rate to offset this.
The margin of safety in Japanese net-nets is that the dividend yield is a payback unrelated to ROI. This reduces the hurdle you have to clear with retained earnings.
Think of it this way. The dividend yield on these Japanese net-nets is 3.07%. And the retained earnings yield is 7.55%. A stock with an earnings yield of 7.55% trades at a P/E just over 13.
So, when you buy a Japanese net-net you get two very bad features vs. U.S. stocks:
· Low growth
· Low return on investment
And you get three good features to offset those bad features:
· Dividend yield
· Excess cash
A Japanese net-net is retaining enough earnings that it would trade at a P/E of 13 even if you didn’t count the dividend as part of earnings. In that sense, you are buying a stock with a P/E of 13 and getting a 3% dividend yield for free. The purchasing power of the yen has risen 0.7% while the purchasing power of the dollar has fallen 2.9%. So, in a sense, you’re getting 3.6% there. (Of course, that’s the rear view mirror. Future purchasing power is what really matters.) U.S. stocks do not normally have a lot of surplus cash. Almost no profitable U.S. companies have so much more cash than debt that they actually have more net cash than their share price. Well, that happens in Japan. Some of these net-nets are proof.
So, the growth and return on investment picture is pathetic. That means retained earnings have less value in Japan than they do in the U.S.
But you don’t just buy a stream of retained earnings when you buy a Japanese net-net. You also get:
· A pile of cash
· A stream of dividends
· The change in purchasing power
In the U.S., it is not uncommon for a company to have more cash than debt. But it’s also not uncommon for a company to have more debt than cash. So, the first factor – the cash pile – is sometimes positive, sometimes negative and sometimes a wash in the U.S. The stream of dividends in the U.S. is nice on some stocks – but how often is it higher than the change in purchasing power?
Real dividend yields in the U.S. are pretty low right now. We’re talking 0% to 2% on what some people consider high-yielding stocks.
In Japan, you can find companies with net cash that have real dividend yields in the 3% to 4% range.
So, you get a dividend defense and a balance sheet defense in Japan that is often lacking in the U.S
In fact, you can find Japanese net-nets that sell for less than their net cash. That fact alone is not impressive. Bad companies sometimes sell for less than net cash.
But you can actually find Japanese net-nets that are priced to give adequate returns without their cash piles that also have really big cash piles.
A stock with no net cash, a P/E of 10 and a dividend yield of 3% is often going to do just fine.
In Japan, you can find stocks that have a P/E around 10, a dividend yield around 3% and a net cash position right around their market cap.
This splits your investment into three bets:
· A bet on the value of the stock’s future retained earnings stream
· A bet on the value of the stock’s future dividend stream
· A bet on the value of the stock’s future cash pile deployment
When you take these three bets together, I like the odds.
If anything is going to derail these Japanese net-net investments, it won’t be a bad economic environment in the sense that you mean it. It won’t be a continuation of the past. That’s baked into the cake. Stagnation as far as the eye can see is fine – you can still make money in Japan if that happens.
So what can derail Japanese net-nets?
A decline in the value of the yen.
A country that is accustomed to deflation could have poor-performing stocks if they were hit with a lot of sustained inflation. Remember, the nominal rates we are talking about here are not that impressive. They’re okay. They become very impressive when you realize they are being paid in a currency that has held its value for a while now.
Investors in the U.S. are rightly concerned about this. Many people tell me they won’t buy Japanese net-nets because they don’t know what the yen will do against the dollar.
That’s fine. In fact, I think it’s the best reason for passing on Japanese net-nets.
My own policy is that I’m always willing to put 50% of my money outside of U.S. dollars. I don’t have a view one way or another on currencies. But I feel that if half my money is in dollars and half is in something else and all 100% of my portfolio is in some of the cheapest stuff on earth – my results will be fine.
I’m not concerned about the fact that I have never backtested Japanese net-nets. Although I’ve talked about backtests of net-nets before, I’ve always really meant those backtests to be illustrations rather than evidence.
I think the U.S. in the 1930s is the best illustration of what net-net investing in Japan is like.
Obviously there are differences between the U.S. and Japan. But it’s not like changes of control were common in the U.S. in the 1930s. And many of the stocks Ben Graham bought were controlled net-nets that were not interested in selling the company at any price.
Also, one of the Japanese net-nets I bought in 2011 (Sanjo Machine Works) was bought out. So it’s hard for me to say Japan totally lacks a certain kind of catalyst that’s important in the U.S. I think net-net investors focus a bit too much on the whole idea of a catalyst. I prefer a lot of uncertain opportunities to make money over time to one seemingly certain exit strategy.
When stocks are as cheap as the five stocks mentioned in this article – there are a lot of different ways to get paid at some point in the future.
A lack of buyouts in Japan is definitely not a plus. But it’s hard to quantify how that changes net-net performance vs. the U.S.
Anyway, the answer is no, I haven’t backtested net-nets in Japan. And I don’t question the viability of the strategy.
If anything, the strategy seems a lot more viable to me in Japan than in the U.S. today – because the quality of net-nets in the U.S. is not as good as it is in Japan. Most U.S. net-nets are extremely unsafe. There are a couple dozen worth considering. The Ben Graham: Net-Net Newsletter probably owns half of them.
The American stock market isn’t exactly teeming with good net-nets right now.
That’s what happens after a few good years. After a few bad years, lists of net-nets look more like what you see in Japan right now.
Actually, in terms of the quality of the net-nets you can find in Japan, I’ve never seen anything like that here. Net-nets like the ones you can find in Japan just haven’t existed in the U.S. in the last couple decades.
Read Geoff’s Other Articles
Ask Geoff a Question about Japanese Net-Nets
Check out the Buffett/Munger: Bargain Newsletter
Check out the Ben Graham: Net-Net Newsletter