What more can I say that the chart doesn’t already?
FRED recently updated the data I’ve been using for my fundamental and sentiment measures from Q1 to Q2 of this year so I thought it would be a good time to post an update to my market-timing model (see “Seeing The Forest For The Trees“).
First, there is the fundamental component for which I use Buffett’s favorite valuation yardstick: total market cap-to-GDP:
Over the past 60 years there has only been one time where stocks were more highly valued and that was during the height of the internet bubble.
This component is nearly 90% negatively correlated to future 10-year returns for stocks (higher readings are correlated with lower returns and vice versa). Right now it’s forecasting a -1.5% annual return over that time.
The sentiment measure, household allocation to stocks, is also now higher than it has ever been outside the peak of the internet bubble (though the 1968 occurrence comes close):
It’s even more highly negatively correlated to future returns and now forecasts a 2.4% annual return over the coming decade. The average of the two comes out to about 1.7% and amounts to one of the worst prospective returns in the history of the data. The risk-free rate, on the 10-year treasury, is almost a full percent higher.
So right now I’m watching the overall trend like a hawk. The S&P 500 and Nasdaq are still well above their 10-month moving averages. However, the Russell 2000 is now more than 1% below its 10-month moving average and the NYSE Composite flirting with its own.
Looking at individual sectors, those that are most in danger of losing their long-term MAs include consumer discretionary, industrials, retail and energy, perhaps the most cyclical sectors of the ten or so I follow. Consumer staples and healthcare, widely considered the most defensive sectors, remain the two strongest. It may be too early to read anything into this but it also may have implications for what’s currently going on in the land of profit margins and the economy.
Ultimately, the uptrend remains in place. However, there are signs that it could be at risk over the next couple of months. With both our fundamental and sentiment measures showing near-record extremes there’s real reason to worry about what the next down cycle will look like. Stay tuned.
The chart above plots corporate profit margins as a percent of GNP alongside the Wilshire 5000 Index. I ran it because I was curious to look at profit margin cycles and how they relate to stock market cycles.
It’s probably an understatement to say that the current bull market has had a strong profit margin tailwind. Margins in this cycle have surged to all-time record highs. This has amounted to what could be considered rocket fuel for valuations, which are now, according to some measures, as stretched as they have ever been:
Median stock in SP500 now sells at RECORD 2.08 Price/Sales (since 1964). Norm is 0.88 Source: @NDR_Research
— Meb Faber (@MebFaber) September 17, 2014
Looking back through history, it’s interesting to note that each of the major peaks in the stock market over the past 40 years has been preceded by a peak in profit margins.
In 2007, the stock market peaked about four quarters after profit margins did. In 2000, stocks peaked about ten quarters later. Stocks peaked about eight quarters after profit margins leading up to the 1987 crash. The only outlier I found (I admit it’s a small sample size) was the 1973-74 bear market where profit margins and stocks peaked simultaneously.
How is this relevant to today’s market? Well, profit margins peaked this cycle way back during Q3 of 2011 (eleven quarters ago) but they never really dropped much until Q1 of this year.
I have argued that a reversion in profit margins poses the “dominant risk” to stocks. Now that they seem to be reverting, we will soon find out if that’s true.
A good deal of attention has already been paid to the growing divergence between small cap and large cap stocks so far this year. The former have seen a small decline while the latter have risen about 8%. But I’ve seen very little commentary regarding WHY this might be happening. Of the many divergences the market has seen recently I think this one may be the most significant as the small caps could be the “canary in the coal mine” for the broader market.
It all comes back to what I have argued amounts to a bubble in corporate profit margins. Jeremy Grantham has used a 2-standard deviation event as one benchmark for a bubble. Using that definition, it’s hard to argue that profit margins are not currently in a bubble.
Warren Buffett also weighed in on unsustainably high margins back in 1999:
In my opinion, you have to be wildly optimistic to believe that corporate profits as a percent of GDP can, for any sustained period, hold much above 6%. One thing keeping the percentage down will be competition, which is alive and well. In addition, there’s a public-policy point: If corporate investors, in aggregate, are going to eat an ever-growing portion of the American economic pie, some other group will have to settle for a smaller portion. That would justifiably raise political problems—and in my view a major reslicing of the pie just isn’t going to happen.
Note that margins are now nearly twice the 6% level that Buffett considered a long-term upper threshold. Now I haven’t heard him say anything about current levels of profit margins (and I’d love for somebody to ask!) but I think his logic is still valid. At some point, the pendulum will have to swing the other way and profits will revert to some extent.
Like the price divergence between small and large caps, the forces behind the scenes here have also been the subject of much ink. It’s that 99% versus the 1% thing. You see over the past few years as the economy has slowly recovered in the wake of the financial crisis companies have seen their revenues grow but have been reluctant to add to their employee base. The result is that a larger and larger portion of these revenues fall to the bottom line. This goes on for a period of five years and, voila! Record profit margins. The 1% (owners of these companies) celebrate while the 99% stagnate.
There are signs recently that this dynamic is shifting. After all, you can only milk your current employee base so much before they become overextended and your product or service suffers or you can’t meet the growing demand, etc. At some point in the recovery or expansion process you have to start adding employees AND paying your current employees a little better in order to retain them.
And it’s beginning to look like this is exactly what’s starting to happen. As the BLS reported a couple of weeks ago, job openings are improving pretty dramatically. July saw a 22% gain year-over-year. And as we learned today, real wage growth spiked in August by the largest amount in years.
This is fantastic news for the 99%. It looks like more jobs and better pay are finally on the way. And it’s exactly the result the FOMC, with their albeit super-blunt tools, have been trying so hard to create. As Pimco’s Paul McCulley writes:
But as Martin Luther King intoned long ago, the arc of the universe does bend toward justice. And as I wrote in July, I think it will do so with the Fed letting the recovery/expansion rip for a long time, fostering real wage gains for Main Street. This implies that the dominant risk for Wall Street is not bursting bubbles, but rather a long slow grind down in profit’s share of GDP/national income.
But do bubbles usually unwind in a “long slow grind down”? Maybe. But sometimes they burst. Either way, this is not so good for the 1% and those record-high profit margins. And we’re seeing this happen already in what area of the market? You guessed it – the small caps and “middle market” companies. Sober Look reports:
While over 50% of [middle market] companies are seeing revenue growth, the fact that over 50% are experiencing EBITDA declines suggests margin compression. For the sixth consecutive quarter, more middle market companies experienced EBITDA declines than gains.
It’s been six consecutive quarters now that these smaller companies have experienced, “margin compression.” UBS recently confirmed this data noting the recent plunge in EBIT margins at small cap companies.
Chart via Business Insider
Make no mistake, this epic stock market rally has been built on the back of this profits boom. It’s been the source of much of the earnings growth we’ve seen and inspired investors to bid valuations to what has historically been rarified air. Should profits decline it would mean already extended valuations are even more inflated than they currently appear and would remove a major underpinning of the bull market.
What I worry about even more, however, is the amount of risk that has been assumed recently based upon the expectation that profit margins will remain at these record levels indefinitely. As Sober Look recently reported, leveraged buy out valuations are at heights not seen at any other time during the past 14 years. More importantly the amount of debt in relation to targets’ EBITDA is also at a record:
Chart via Sober Look
If EBITDA at more than half of these companies is actually declining now these multiples will soon look even more inflated than they already do and the massive amount of debt used in buying them is at risk even greater risk of becoming unsustainable than it originally appears.
Speaking of the “massive amount of debt,” It’s important to note that the volume of leveraged loans has far surpassed it’s highs of 2007…
Chart via Dallas Fed
…and the risk controls embedded in these loans has fallen dramatically as covenant-lite’s share of overall issuance is now twice what it was prior to the financial crisis.
Chart via Dallas Fed
So it looks as if we may have more built up risk on the debt side of things than we did prior to the financial crisis. If margins are actually beginning to revert, as the small cap/middle market is suggesting, at 2-standard deviations above their long-term average, they potentially have a very long way to fall. And with so much risk betting against this possibility the fallout could be dramatic.
Perhaps this is why spreads have finally begun to widen just a bit over the past few months in the high-yield market.
Chart via Charlie Bilello
All in all, this is clearly a very complex system with various intermarket relationships. But we are seeing some signals that point to the fact that the Fed may be close to achieving it’s goals of increasing employment and wages. While this is good news for the labor force, it’s bad news for companies and investors because the resulting margin compression would remove the main driving force of this bull market along with causing potential problems (defaults) in the high-yield bond market. So keep your eyes on the small caps; there are big implications in that divergence everyone’s looking at.
Three months ago I published a short piece titled, “Taking a shine to the gold miners,” and the ETF subsequently jumped a little over twenty percent. Recently, however, it has given back a good chunk of those gains and I’m once again intrigued by the trade.
Now I’m not going to try to make the fundamental case for owning gold. You can find great arguments being made on both sides of that debate (see “The Great Gold Debate“). Either you believe that gold is a unique store of value or that it’s just a shinier metal than most. What I’m interested in right now is the technical picture and general sentiment towards the asset after a three-year bear market that has seen the precious metal drop, in dollar terms, nearly forty percent.
On a long-term time frame gold is testing a pretty important uptrend right now. Actually, it’s fallen below the uptrend but I’ll be looking for a monthly close below the trend line and/or the 10-month moving average (both around 1275) to determine that the uptrend line is officially broken and there’s a lot of time left in the month of September for it to rally back and close above.
The MACD lines at the bottom of the chart are also getting very close to turning higher. Notice they last crossed down back in early 2012 which turned out to be a great signal to get out if you owned any gold at the time.
Looking at the gold ETF on a weekly time frame (a chart I posted at the end of last year on my public StockCharts chart book), it’s clearly breaking down out of a bearish pennant but key support lies just below at the 61.8% Fibonacci retracement around 113.68 (roughly 1200 for the precious metal).
So the metal is at a crucial juncture right here. Can it hold the monthly uptrend and the weekly 61.8% retracement level? Time will tell.
Should they both manage to hold these key leves, the best way to play it may be through the gold miners ETF. A look at the weekly chart here shows a potential variation of a head and shoulders bottom with a DeMark 13 buy signal. Technically, you could argue the price action is either basing for a reversal or flagging before continuing lower.
But at the bottom of the chart I’ve included the ratio between the miners ETF and the gold ETF. After underperforming the metal for the past three years or so the miners have recently begun to outperform. If their underperformance back in 2011 was a warning signal that gold’s bull run was coming to an end (and it was) this recent outperformance could mean the bearish trend of the past few years may soon be ending, as well.
And a longer term look at the ratio between the miners and the metal (at the bottom of the chart below) shows the miners have not been this cheap relative to the underlying metal at any point during the past twenty years.
Finally, not only are portfolio managers extremely underweight the precious metals right now, individual investors have pretty much abandoned them, too.
Chart via SentimenTrader
While the so-called dumb money gives up on the trade, the smart money is getting aggressive on the long side. George Soros nearly doubled his position in the gold miners ETF last quarter to over two million shares. He also bought call options on the gold ETF equivalent to about 1.33 million shares.
For those of you who fall in the gold bug camp, the technicals and sentiment may finally be aligning in your favor once again. Stay tuned.
Below is a compilation I put together using excerpts from Richard Fisher’s speeches this year (emphasis mine):
There is no greater gift to a financial market operator—or anyone, for that matter—than free and abundant money. It reduces the cost of taking risk. But it also burns a hole in the proverbial pocket. It enhances the appeal of things that might not otherwise look so comely. I have likened the effect to that of strapping on what students here at USC and campuses elsewhere call “beer goggles.” This phenomenon occurs when alcohol renders alluring what might otherwise appear less clever or attractive. And this is, indeed, what has happened to stocks and bonds and other financial investments as a result of the free-flowing liquidity we at the Fed have poured down the throat of the economy. Here are some of the developments that signal we have made for an intoxicating brew as we have continued pouring liquidity down the economy’s throat:
- Share buybacks financed by debt issuance that after tax treatment and inflation incur minimal, and in some cases negative, cost; this has a most pleasant effect on earnings per share apart from top-line revenue growth.
- Dividend payouts financed by cheap debt that bolster share prices.
- The “bull/bear spread” for equities now being higher than in October 2007.
- Stock market metrics such as price-to-sales ratios and market capitalization as a percentage of gross domestic product at eye-popping levels not seen since the dot-com boom of the late 1990s.
- The price-to-earnings (PE) ratio of stocks is among the highest decile of reported values since 1881. Bob Shiller’s inflation-adjusted PE ratio reached 26 this week as the Standard & Poor’s 500 hit yet another record high. For context, the measure hit 30 before Black Tuesday in 1929 and reached an all-time high of 44 before the dot-com implosion at the end of 1999….
- Margin debt that is pushing into all-time records.
- In the bond market, investment-grade yield spreads over “risk free” government bonds becoming abnormally tight.
- “Covenant lite” lending becoming robust – surpassing even the 2007 highs – and the spread between CCC credit and investment-grade credit or the risk-free rate historically narrow. I will note here that I am all for helping businesses get back on their feet so that they can expand employment and America’s prosperity: This is the root desire of the FOMC. But I worry when “junk” companies that should borrow at a premium reflecting their risk of failure are able to borrow (or have their shares priced) at rates that defy the odds of that risk. I may be too close to this given my background. I have been involved with the credit markets since 1975. I have never seen such ebullient credit markets. From 1989 through 1997, I was managing partner of a fund that bought distressed debt, used our positions to bring about changes in the companies we invested in, and made a handsome profit from the dividends, interest payments and stock price appreciation that flowed from the restructured companies. Today, I would have to hire Sherlock Holmes to find a single distressed company priced attractively enough to buy. The big banks are lending money on terms and at prices that any banker with a memory cell knows from experience usually end in tears.
The former funds manager in me sees these as yellow lights. The central banker in me is reminded of the mandate to safeguard financial stability. We must watch these developments carefully lest we become responsible for raising the ghost of irrational exuberance.
Why isn’t anyone listening?
Yesterday morning I came across a piece over at Harvard Business Review titled, “To Make Better Decisions, Combine Datasets.” I began reading it and realized that’s exactly the key to investment success and what I’ve tried to do with my market timing model: combine a variety of predictive datasets to create a holistic forecasting and timing model.
The stock market is driven not just by fundamentals or sentiment or technicals alone but by all of them in concert with one another. It follows then that an investor should try to incorporate each of them into her investment process in order to maximize its effectiveness.
And this is where I think many investors get lost. They try to focus on only one of these three. Fundamentals alone may work over the long run but cheap stocks can always get much cheaper in the short-term or they could just be cheap for a very good reason (I’ve learned this lesson more than a few times). Sentiment can also be very helpful but the crowd isn’t always wrong and markets can ‘stay irrational longer than you can stay solvent.’ And, as many traders know, the ‘trend is only your friend until it comes to an end.’
What I’ve found in my 20+ years of observing and trading markets is that looking at the forest, by putting all of these together, rather than the trees alone is absolutely crucial to making good decisions. So I thought it might be fun to look at the individual components of the model to see not only what they are saying about the markets but how they might be misleading when taken on their own.
For my fundamental component I use Buffett’s favorite valuation yardstick, total market capitalization-to-GDP. On its own it has roughly an 83% negative correlation with future 10-year returns in the stock market (based on 65 years worth of data). This means higher levels for this indicator are correlated with lower future returns and vice versa. Here’s what it looks like over the past 65 years or so:
Even considering the fact that the internet bubble has pushed the average higher over the past ten or fifteen years, this measure still suggests stocks are priced significantly above their historical range. Based on its high correlation with future returns this suggests investors should expect a very low return from present levels over the next decade.
BUT… this has been the case for most of the past 20 years! An investor looking at this measure alone might have sat out a couple of major bear markets but also would have missed a couple of the most massive bull markets in history! So it’s probably not smart to use this measure in isolation. Adding other related asset classes (like bonds – we’ll come back to that) and other, unrelated indicators should help give a bit more clarity.
My sentiment measure tracks the percent of household financial assets invested in equities. Believe it or not this measure is even more highly negatively correlated with future returns than Buffett’s valuation measure above (closer to 90% – hat tip, Jesse Livermore). Here’s what it looks like over the same time frame:
It’s also currently sitting significantly above its long run average suggesting returns should be far below average going forward. As I mentioned this is a better forecasting mechanism than the fundamental measure but even if the incredible euphoria of the internet bubble got you out of the stock market you may not have gotten back in over the past 15 years because we haven’t seen anything like the pessimism witnessed at the 1982 low.
Finally, I’ve added a third component to the model, inspired by Doug Short: a simple trend regression model based on Robert Shiller’s data going back nearly 150 years. With a negative correlation of roughly 74%, it’s not quite as effective at forecasting future returns as these other two but I think adding it, as a third independent component based on a very long-term trend, helps to make the model more robust. So here’s what the S&P 500 looks like relative to a regression trend line over the full time period:
Once again this indicator shows the stock market to be trading very close to the top of its historical range. Still, like the fundamental model this one might have had you sitting out of the stock market for perhaps the past 20 years!
So even though we have three independent models we need a way to put them together and then to put them into some sort of context. What I’ve done is used each indicator individually to create a 10-year forecasting model. Then I’ve simply averaged them together each quarter. All told, the combination results in a correlation to future 10-year returns of about 90%. Here’s a chart of the model’s forecast returns as compared to actual 10-year returns for the stock market:
Where the model is farthest off the mark (where you see the yellow line far above the blue line) is in the late 80’s early 90’s. Stocks surged further and faster during the internet bubble than the model forecast they would. Removing those years, the model’s correlation value rises to about 94%.
So we know what the individual readings look like. What’s the model saying about future returns from here? As the chart below shows, the model forecasts a return of just 1.2% per year over the next decade:
To add some context, in addition to the 10-year forecast I’ve put the yield of the 10-year treasury note on the chart, as well. Investors don’t look at potential returns in a vacuum; they compare potential returns of different opportunities, many times looking at the “risk-free” rate of treasury notes in the process. This next chart shows the difference between the model’s forecast return and the yield on the 10-year treasury note:
When the blue line is above zero, stocks offer the better return; when it’s below, bonds do. And as I’ve shown before in “How To Time The Market Like Warren Buffett” this timing model works very well. Just buy whatever asset class is more attractive – trading only once per year – and you’ll kill a buy-and-hold approach.
I think this alone is validation of a multi-disciplinary approach. But adding one more super-simple component makes it that much more effective: before we go and sell our stocks because bonds are more attractive, we want to make sure we don’t sell too early in a bull market or buy to early in a bear market. As the chart above shows this model would have had you sell your stocks and shift into bonds all the way back in April of 1996 and then miss all the gains of the next 3 1/2 years.
Adding a very simple trend-following approach solves this problem (hat tip, Meb Faber). Rather than sell right when bonds become more attractive it’s much more advantageous to wait for the trend to end. And as a representation of the trend, we can simply use a 10-month moving average. Below is a chart of the S&P 500 and this moving average:
To be clear we’re not trend followers all the time with this model. We buy-and-hold until the model tells us that stocks are not attractively priced and then we become pure trend followers. Once the model tells us stocks have become less attractive than bonds we wait for the S&P 500 to close at least 1% below its 10-month moving average at which point we sell our stocks and sit in cash, buy bonds or even short stocks (the latter generates the best returns over the period studied).
Should the index at any point close back above its 10-month moving average by at least 1% we buy stocks again. Like I said, so long as stocks are less attractive than bonds we are pure trend followers. Only when the model suggests stocks are once again more attractively priced than bonds AND the trend has turned up (as indicated by a monthly close above the 10-ma) do we buy stocks and abandon trend-following for buy-and-hold.
Ultimately what this produces is a combination buy-and-hold/trend-following model that owns stocks roughly 80% of the time and seeks to avoid major bear markets precipitated by high valuations, high levels of bullishness and prices extended far above their regression trend. It doesn’t avoid losses entirely, though.
The model didn’t recommend a move out of stocks prior to the 1987 crash which resulted in a decline of roughly 26% (its largest drawdown). It did, however manage to avoid the ’73-’74, ’00-’02 and ’08-’09 bear markets, the latter producing about a 50% decline. In fact, this is where all of the model’s outperformance is generated: in recognizing these major turning points fairly early on – essentially giving a warning signal – and then switching from buy-and-hold to trend-following when that strategy is more effective.
The next chart shows the results of three different investors. The first is a simple buy-and-hold strategy (blue line). The second goes to cash when the model indicates (red). The third, rather than going to cash, shorts the index (green):
Clearly there is significant benefit to abandoning buy-and-hold for a trend-following approach when our model suggests stocks are unattractively priced. Over the period the investor who just sits out major bear markets in cash ends up with twice as much as the investor who holds through the entire decline. And the investor who gets short, in turn, fares far better still.
I truly believe these superb results, hypothetical though they be, can be attributed to the holistic nature of the model. It combines datasets that are valuable independent of one another into something greater than its parts.
As of now, the model is telling us that stocks have once again become unattractive relative to bonds. However, the uptrend is still in tact. So it’s probably valid to be bearish for fundamental, sentiment and regression reasons. But the trend is also a valid reason to be bullish – even if it is the only reason. So I’m still looking at the market through a bearish lens right now but I’ll be watching for a monthly close at least 1% below the index’s 10-month moving average for the trend to validate the fundamentals and sentiment.
For reference I’ve put up all the spreadsheets, calculations and charts I used on a public Google Drive sheet here: Market Timing Model. I’ll be updating it as new data comes in.
Finally, I need to make the same disclaimer I’ve made over and over again during this series: because this is a hypothetical model that doesn’t incorporate taxes, transaction fees, etc. it is not representative of any real returns. It is merely for educational purposes. Clearly, past performance may not be indicative of any future results.