Should You Really Care If Stocks Are Cheap?
| November 7,
2011 |
Are stocks cheap right now? Are they expensive? Should you even care?
It's almost axiomatic in this business that investors should buy stocks when they're cheap and avoid them when they're expensive. All of us accept this principle without question, even if we're hardcore technical analysts.
But let me run you through a few charts that may raise some questions about your blind faith in the Price-to-Earnings multiple, the most popular way of measuring value.
What this first chart does is take a snapshot of the market's P/E ratio on January 1, 1900. Then it fast-forwards three years to see how much money you made or lost if you made an investment in the Dow. Those two pieces of data become one little dot on our scatter point chart. I've plotted points for every month in the 20th and 21st centuries.
click to enlarge
In this chart we're relating present valuations to future returns. We're checking to see if buying the stock market when it's cheap translates into profits down the road.
Over a three year window, the data tell a fairly convincing story. Valutions matter a lot less than we're comfortable admitting. As indicated by our regression line, there is a modest correlation between present valuation and future return, but with an R-squared of 0.03, it should be ignored. So there you go.
Over a 3-year window, there is no correlation between ttm P/E's and the degree of future profitability.
Below average P/E's do not imply above average returns over a 3 year window, nor do above average P/E's imply below average returns. Some of you are probably excited to learn that, while others are thinking I've fipped my lid. But it's not me that's saying this. It's the data.
3 years isn't a very big window, and while it may not be too far off the average holding period of a long-term investment, it's not what most people have in mind when they think about investing for the "long run."
Let's broaden our window to 5 years. Same methodology. Current valuations versus 5-year subsequent returns.
Huh. Very similar story.
No correlation between present valuation and the degree future profitability. Our R-squared even went down a little bit. Not that a reading of 0.0295 matters in a statistical sense.
Just looking at the chart visually, you can see that when the earnings multiple gets really cheap, it suggests that your odds of getting burned really bad go down and your odds of coming out OK 5 years hence are a little better. That's assuming you can hang on for 5 years.
But in the middle of the curve, between a 10x and 20x multiple, it's a crapshoot. Even over 5 years. Here, let me provide some visual aid and get rid of all the extreme points.
Sure, there are more windows of gains than losses, but that's just because over the last century the market has tended to rise, not fall.
How confident are you -- really -- that any ttm P/E ratio between 10x and 20x tells you anything about what's to come over the next five years?
Let's broaden our window a little further. Let's widen our scope to 10 years. Does buying stocks when they are particularly cheap translate into more profitability for the coming decade?
The visual correlation is a little more clear. Our R-squared is a bit higher but is it even at the level where it's statistically meaningful? I'm still not sure I'm prepared to say that buying at lower multiples translates to higher profitability.
I feel good about saying one thing, though. There are few guarantees in this business, but -- surprise! -- if youy buy stocks when they are craaazy cheap and can hold onto them for 10 years, the odds suggest you'll make money.
For the record, those data points in the lower left quadrant, those that produced negative subsequent 10-year returns and started from very low valuations all began in mid-1920s. A decade later in the middle of the Great Depression, you had technically broken even on your purchase. But if you did happen to buy, say, in 1924, surely you must have taken something off the table at some point before the crash? At the peak your investment had quadrupled.
In practice, the answer to that question was probably "not bloody likely!" Given what we know about human psychology and patterns of behavior, you were probably loading your boat with even more stock in early 1929. Nevermind that by that point, the ttm price/earnings multiple had risen from just under 10 to over 17. That was almost two standard deviations above the average P/E multiple for the previous three decades. At that point, you should have been swapping out your stock for as much "imported" Canadian Whisky as your boat could fit.
Anyway, the point is that this analysis cuts both ways. If you buy when stocks are cheap relative to their ttm earnings, it's not quite a guarantee that you'll make money over the coming decade. Though the odds are definitely in your favor. The catch is that you shouldn't necessarily wait 10 years (or 5, or 3). You should remain mindful of that P/E multiple and reduce exposure accordingly as the ratio rises.
I know that this is pretty obvious to any disciple of Graham & Dodd or even a casual practitioner of fundamental analysis. But it's worth re-iterating because so few investors actually seem to do this. Human psychology is broken when it comes to investing, and when dealing with long windows of time, all sorts of noise comes into play. Still, the enduring concept is known to pretty much everyone. So let me wrap this section up with a few points that are substantially less obvious.
Three less-obvious conclusions
1. Correlation between earnings multiples and subsequent returns is loose. Statistically speaking, it's almost non-existent.
2. Unless you're dealing with long holding periods, pretty much any outcome is in play. Markets that are historically cheap can get a lot cheaper and take a lot longer than most humans are comfortable waiting before swinging back into profitability.
3. In truth, this kind of analysis is only really useful during a few select moments in history. It's only helpful at identifying extremes when the probabilities start to skew in the rational investor's favor. The problem is that during those extremes is when the fewest number of investors are bothering with this type of analysis. They're lost in all the other noise of the moment. They're buying tech stocks in 2001 because, it's the future, maaan! Or they're cowering in their bunker in 1948. Or broke in 1933. Or high on the Nifty Fifty in 1965, among other interesting things.
You don't need to look at an earnings multiple to know that if you buy some stocks and hold them long enough there will be periods where you will be in the red and also in the black.
Finally, let me add a twist to this story. There are other ways of assessing earnings. Some people like to normalize them. Personally, I like to look at the average earnings over the last decade. This is what Robert Shiller does. John Hussman relates prices to peak earnings. That's cool too.
Here's a chart relating normalized earnings (10 year average) to the return over the coming decade.
Now we're talking.
Over the course of a decade, low normalized PEs not only imply a greater chance of future profitability, they also suggest a greater degree of profitability.
Again, that's not me saying that. It's the data. If you buy the market when stocks are fundamentally cheap on a normalized basis and hold them for a long time, the odds are pretty good you'll make money, and make more of it than if you'd bought when stocks were expensive.
That chart carries one important footnote. You'll notice that in that chart there are two distinct curves. The top curve is pretty much everything from the 1990's through today. The lower curve represents the rest of history. You're welcome to completely disregard either curve, but the take home point is the same. There's a definite correlation between normalized valuation and future price appreciation.
Where are we at today?
All this research has been fun. But this is Seeking Alpha and at some point we need to talk turkey.
As of right now, the S&P 500 is priced right around 15x the last twelve months' earnings. You know what this means. This means that any outcome is in play.
On a normalized basis, the S&P 500 is currently priced just under 21x. Depending on whether you think the coming decade will look like the 90's/00's or the rest of 20th century, two very different baskets of outcomes are in play.
Last week I did a big walkthrough of the major economic fundamentals. Let me spoil the conclusion: pretty much all the data right now suggests an outcome where anything could happen. One of the secrets about professional portfolio management is that it's environments like these that are particularly hostile for mainstream investors. It's environments where anything can happen and investors aren't getting appropriately compensated with risk-adjusted returns for entering the market.
If you buy stocks today with the intention of holding them for at least 3 years, you should be prepared to at some point have to deal with losses of 30-60%. You could also be up double in 3 years. Why not hit the track instead? Or better yet, come visit our casinos here in my home state of Nevada!
Sorry, I don't mean to be cynical or flippant. The real take home point is that if you are trying to figure out how to invest today, the valuation of the broad market doesn't really need to be a part of your analysis. You should be doing different kinds of research to get you through the next decade. You've got to construct clever strategies and you've got to know a few tricks and tactics.
Earlier in the year I did an example of that here at Seeking Alpha. We're doing other things right now with our hedge fund clients. I even wrote an entire book about how average investors can construct some snazzy long-term strategies that should tilt the odds in your favor over the next decade. Because, God knows, the market valuation isn't doing anything right now to boost your probability of success.
How to play it
If you think that all this talk of historical valuation is bunk & hooey and you've got your own reasons for getting long the market, go ahead and load up on the SPDRs (SPY) or Diamonds (DIA). Then again, if you believe that these valuations suggest trouble on the horizon, feel free to short those ETFs as well. Generally speaking, I prefer shorting ETFs to individual stocks to express macro views because it eliminates the idiosyncratic risk associated with specific companies (aka "Volkswagen Risk").
If I was forced to buy something, I'd rather have a portfolio of systemically important U.S. stocks with clean balance sheets, defensible business models, and a high likelihood that they'll grow at a rate greater than GDP over the next decade. All the usual names apply -- Johnson & Johnson (JNJ), Exxon (XOM), Intel, (INTC), Wal-Mart (WMT), Proctor & Gamble (PG), and the like. This is still an environment where investors will be better off with active strategies and picking specific stocks than with passive strategies and positioning oneself like an index. I don't see that environment changing for a while.
Someday we'll find ourselves at another extreme. I'm looking forward to it. If market valuations do get down a ways below 10, on both a trailing and normalized basis, it'll be time to load the boat. It'll be a once-in-a-career trade and there will be so much noise and fear that most people will miss it.
An ability to look at this stuff rationally will ensure that you won't.
It's almost axiomatic in this business that investors should buy stocks when they're cheap and avoid them when they're expensive. All of us accept this principle without question, even if we're hardcore technical analysts.
But let me run you through a few charts that may raise some questions about your blind faith in the Price-to-Earnings multiple, the most popular way of measuring value.
What this first chart does is take a snapshot of the market's P/E ratio on January 1, 1900. Then it fast-forwards three years to see how much money you made or lost if you made an investment in the Dow. Those two pieces of data become one little dot on our scatter point chart. I've plotted points for every month in the 20th and 21st centuries.
click to enlarge
In this chart we're relating present valuations to future returns. We're checking to see if buying the stock market when it's cheap translates into profits down the road.
Over a three year window, the data tell a fairly convincing story. Valutions matter a lot less than we're comfortable admitting. As indicated by our regression line, there is a modest correlation between present valuation and future return, but with an R-squared of 0.03, it should be ignored. So there you go.
Over a 3-year window, there is no correlation between ttm P/E's and the degree of future profitability.
Below average P/E's do not imply above average returns over a 3 year window, nor do above average P/E's imply below average returns. Some of you are probably excited to learn that, while others are thinking I've fipped my lid. But it's not me that's saying this. It's the data.
3 years isn't a very big window, and while it may not be too far off the average holding period of a long-term investment, it's not what most people have in mind when they think about investing for the "long run."
Let's broaden our window to 5 years. Same methodology. Current valuations versus 5-year subsequent returns.
Huh. Very similar story.
No correlation between present valuation and the degree future profitability. Our R-squared even went down a little bit. Not that a reading of 0.0295 matters in a statistical sense.
Just looking at the chart visually, you can see that when the earnings multiple gets really cheap, it suggests that your odds of getting burned really bad go down and your odds of coming out OK 5 years hence are a little better. That's assuming you can hang on for 5 years.
But in the middle of the curve, between a 10x and 20x multiple, it's a crapshoot. Even over 5 years. Here, let me provide some visual aid and get rid of all the extreme points.
Sure, there are more windows of gains than losses, but that's just because over the last century the market has tended to rise, not fall.
How confident are you -- really -- that any ttm P/E ratio between 10x and 20x tells you anything about what's to come over the next five years?
Let's broaden our window a little further. Let's widen our scope to 10 years. Does buying stocks when they are particularly cheap translate into more profitability for the coming decade?
The visual correlation is a little more clear. Our R-squared is a bit higher but is it even at the level where it's statistically meaningful? I'm still not sure I'm prepared to say that buying at lower multiples translates to higher profitability.
I feel good about saying one thing, though. There are few guarantees in this business, but -- surprise! -- if youy buy stocks when they are craaazy cheap and can hold onto them for 10 years, the odds suggest you'll make money.
For the record, those data points in the lower left quadrant, those that produced negative subsequent 10-year returns and started from very low valuations all began in mid-1920s. A decade later in the middle of the Great Depression, you had technically broken even on your purchase. But if you did happen to buy, say, in 1924, surely you must have taken something off the table at some point before the crash? At the peak your investment had quadrupled.
In practice, the answer to that question was probably "not bloody likely!" Given what we know about human psychology and patterns of behavior, you were probably loading your boat with even more stock in early 1929. Nevermind that by that point, the ttm price/earnings multiple had risen from just under 10 to over 17. That was almost two standard deviations above the average P/E multiple for the previous three decades. At that point, you should have been swapping out your stock for as much "imported" Canadian Whisky as your boat could fit.
Anyway, the point is that this analysis cuts both ways. If you buy when stocks are cheap relative to their ttm earnings, it's not quite a guarantee that you'll make money over the coming decade. Though the odds are definitely in your favor. The catch is that you shouldn't necessarily wait 10 years (or 5, or 3). You should remain mindful of that P/E multiple and reduce exposure accordingly as the ratio rises.
I know that this is pretty obvious to any disciple of Graham & Dodd or even a casual practitioner of fundamental analysis. But it's worth re-iterating because so few investors actually seem to do this. Human psychology is broken when it comes to investing, and when dealing with long windows of time, all sorts of noise comes into play. Still, the enduring concept is known to pretty much everyone. So let me wrap this section up with a few points that are substantially less obvious.
Three less-obvious conclusions
1. Correlation between earnings multiples and subsequent returns is loose. Statistically speaking, it's almost non-existent.
2. Unless you're dealing with long holding periods, pretty much any outcome is in play. Markets that are historically cheap can get a lot cheaper and take a lot longer than most humans are comfortable waiting before swinging back into profitability.
3. In truth, this kind of analysis is only really useful during a few select moments in history. It's only helpful at identifying extremes when the probabilities start to skew in the rational investor's favor. The problem is that during those extremes is when the fewest number of investors are bothering with this type of analysis. They're lost in all the other noise of the moment. They're buying tech stocks in 2001 because, it's the future, maaan! Or they're cowering in their bunker in 1948. Or broke in 1933. Or high on the Nifty Fifty in 1965, among other interesting things.
You don't need to look at an earnings multiple to know that if you buy some stocks and hold them long enough there will be periods where you will be in the red and also in the black.
Finally, let me add a twist to this story. There are other ways of assessing earnings. Some people like to normalize them. Personally, I like to look at the average earnings over the last decade. This is what Robert Shiller does. John Hussman relates prices to peak earnings. That's cool too.
Here's a chart relating normalized earnings (10 year average) to the return over the coming decade.
Now we're talking.
Over the course of a decade, low normalized PEs not only imply a greater chance of future profitability, they also suggest a greater degree of profitability.
Again, that's not me saying that. It's the data. If you buy the market when stocks are fundamentally cheap on a normalized basis and hold them for a long time, the odds are pretty good you'll make money, and make more of it than if you'd bought when stocks were expensive.
That chart carries one important footnote. You'll notice that in that chart there are two distinct curves. The top curve is pretty much everything from the 1990's through today. The lower curve represents the rest of history. You're welcome to completely disregard either curve, but the take home point is the same. There's a definite correlation between normalized valuation and future price appreciation.
Where are we at today?
All this research has been fun. But this is Seeking Alpha and at some point we need to talk turkey.
As of right now, the S&P 500 is priced right around 15x the last twelve months' earnings. You know what this means. This means that any outcome is in play.
On a normalized basis, the S&P 500 is currently priced just under 21x. Depending on whether you think the coming decade will look like the 90's/00's or the rest of 20th century, two very different baskets of outcomes are in play.
Last week I did a big walkthrough of the major economic fundamentals. Let me spoil the conclusion: pretty much all the data right now suggests an outcome where anything could happen. One of the secrets about professional portfolio management is that it's environments like these that are particularly hostile for mainstream investors. It's environments where anything can happen and investors aren't getting appropriately compensated with risk-adjusted returns for entering the market.
If you buy stocks today with the intention of holding them for at least 3 years, you should be prepared to at some point have to deal with losses of 30-60%. You could also be up double in 3 years. Why not hit the track instead? Or better yet, come visit our casinos here in my home state of Nevada!
Sorry, I don't mean to be cynical or flippant. The real take home point is that if you are trying to figure out how to invest today, the valuation of the broad market doesn't really need to be a part of your analysis. You should be doing different kinds of research to get you through the next decade. You've got to construct clever strategies and you've got to know a few tricks and tactics.
Earlier in the year I did an example of that here at Seeking Alpha. We're doing other things right now with our hedge fund clients. I even wrote an entire book about how average investors can construct some snazzy long-term strategies that should tilt the odds in your favor over the next decade. Because, God knows, the market valuation isn't doing anything right now to boost your probability of success.
How to play it
If you think that all this talk of historical valuation is bunk & hooey and you've got your own reasons for getting long the market, go ahead and load up on the SPDRs (SPY) or Diamonds (DIA). Then again, if you believe that these valuations suggest trouble on the horizon, feel free to short those ETFs as well. Generally speaking, I prefer shorting ETFs to individual stocks to express macro views because it eliminates the idiosyncratic risk associated with specific companies (aka "Volkswagen Risk").
If I was forced to buy something, I'd rather have a portfolio of systemically important U.S. stocks with clean balance sheets, defensible business models, and a high likelihood that they'll grow at a rate greater than GDP over the next decade. All the usual names apply -- Johnson & Johnson (JNJ), Exxon (XOM), Intel, (INTC), Wal-Mart (WMT), Proctor & Gamble (PG), and the like. This is still an environment where investors will be better off with active strategies and picking specific stocks than with passive strategies and positioning oneself like an index. I don't see that environment changing for a while.
Someday we'll find ourselves at another extreme. I'm looking forward to it. If market valuations do get down a ways below 10, on both a trailing and normalized basis, it'll be time to load the boat. It'll be a once-in-a-career trade and there will be so much noise and fear that most people will miss it.
An ability to look at this stuff rationally will ensure that you won't.
No comments:
Post a Comment