The S&P 500 now sits less than 2% away from new all-time highs. It’s also completely unchanged over the last 15 months, and during this time experienced what I would classify as medium/high volatility. In particular, 4Q2018, although the volatility in January 2018 was notable in that it prompted the “blow up” of a few volatility-oriented investment funds.
Conditions over the last year have been dynamic, but most recently markets enjoy particularly positive performance. Given this recent investment environment, what is the prognosis going forward? I’ll use a framework of valuation, fundamentals, and sentiment to try and build a usable road map and set of investment expectations.
In short, markets appear to be in the upper half of historical valuations after reaching particularly high valuation levels in mid 2018. However, the level at a particular point in time seems to be less important than the duration of stay at the elevated level. Mid-2018 appears to be the first time this bull market breached the barrier into high valuation territory, suggesting the bull market may not be over.
With that said, the factors influencing the market constantly change and therefore, with enough change, an investment outlook can shift at any point.
Over the past few years, analysts seem to unanimously indicate that the U.S. stock market, on a long-term basis, is overvalued. They cite the cyclically-adjusted P/E ratio (Shiller P/E), the Q ratio, or market cap/GDP. I believe this analysis to be incorrect, or too simplistic and therefore misleading and unusable. I say “too simplistic” because the market and its drivers, over the long-run, experience a constant state of change. I say this with confidence, while also acknowledging that some things never change (e.g. fear, greed, herding behavior, etc.).
A better way, in my opinion, to view long-term valuation methods is to acknowledge what has changed that may impact traditional value measures. When I ask this question, dividends and interest rates come to mind. Dividend yields have been in a secular bear market for 100 years. More notably, bonds have experienced a widely-known (and profitable) bull market since 1980 (link to 10-year treasury bond yield chart). If a stock’s present value is the sum of all future cash flows, discounted to present value (using a currently-available interest rate), how do major and long-term fluctuations in interest rates not directly impact values? They do, so the goal then is to create a new means of valuation that adjusts with the impact of rate fluctuations. Unfortunately, this task is riddled with complexity and potential issues.
Before I propose a solution, let me state the following: I believe that robust, sustainable, and long-term investing should be kept as simple as possible. Many investors and analysts instinctively go for precision and incorporate huge sums of data. A discounted cash flow analysis can take up massive Excel spreadsheets and AI/big data solutions can be utilized for intermarket analysis and market timing models. These models require assumptions and estimates that are, by definition, wrong – leading to a compounding problem. Instead, I’ll follow Mr. Einstein’s advice and keep things as simple as possible, but no more.
Interest Rate-adjusted CAPE Ratio
Just as the Federal Reserve Board never knows what the exact natural rate should be, they can estimate using robust tools. So too can investors judge the over/under valuation on equity markets by adjusting the Shiller P/E by interest rates and dividends. See chart below:
The original Shiller P/E and my adjusted version both show an upward inclination over the decades. I don’t know what causes this but I have some thoughts.
- It seems to me that over time, as data slowly (but in an accelerated manner) circulates on a wider scale and with more depth, valuations have slowly drifted to their truer level (higher). This line of thinking means that equities may have been chronically undervalued until the 1980s.
- Population growth and/or emerging markets “coming on line” (in a financial and commercialization sense) may skew the demand side of the supply/demand relationship for equities. This would have occurred over time and may account for the gradual upward trend in both models above.
Whatever the reason, I don’t think it matters much. Even if I knew for sure, the relationship is too slow to profit from. Forgetting the narrative and focusing on the problem at hand prompts me to simply detrend the data set. I use a 40 year moving average (really long-term investing here!) to mirror an investing career and multiple boom/bust cycles. This detrended valuation method provides the below chart. I’ve removed the Y axis since it’s irrelevant and includes negative numbers, which are nonsensical. The value in the below chart is in comparing the current level to historical ones.
As an investor, I must operate in a manner where I plan for the worst outcome and am rewarded in (most) others. When I work with a custom valuation model like this, it needs to fulfill the logic test. The original Shiller P/E makes an inexcusable error in my opinion; it fails to show the 2009 low in equities as a tremendous and historically cheap buying opportunity. Undervalued, but only equivalent to equities in 1989? No, I’m sorry, that is a flaw. This adjusted model addresses the issue and indicates the 2009 low as substantially undervalued, similar to 1974 and the 1930s. It also still shows the tech bubble as what it was, the 1929 bubble as severely overvalued and today’s market as moderately valued. Check, check, check and check. I know that it is also flawed to some degree, but it looks and feels like an improvement. Above all else it makes sense and seems logical. Investing is about working with the best information available. Also, the point here is to develop a tool for approximating value for a very large number of individual stocks. To do this properly, one is probably better served to use a sledge hammer rather than a scalpel.
What does the adjusted valuation tell investors about current conditions and the likely path forward for the broad indices? From a big picture standpoint, I see that 2018 valuations crept to lofty levels – a level that warranted a narrowing of the spread between prices and earnings. The late-2018 low brought valuations squarely back to the median; we have since made up about half of that decline.
Today’s environment, from a number of different vantages (including valuation), reminds me of the 1960s. Rarely will a bull market reach overvaluation and then immediately end. It’s a gradual process involving investor “anchoring” where market participants gradually get accustomed to more recent valuation levels (this is not an exact science). The 1960s experienced a series of equity advances with each one taking valuations to token new highs. Eventually equities reached an unsustainable extreme and secular bear market of the 1970s ensued.
Another reason I would lean more toward the 1960s as an analog model is because the current environment seems markedly different than the late 1920s or 1990s. Both were punctuated with new technology and innovation. Admittedly, I’m not a market historian, although I enjoy reading market history and have done quite a bit of it. That said, the current economic environment just seems different. I know it’s different than the 1990s, a period I lived through while living in San Jose. Hindsight is 20/20, but the profiles of each period make a blow-off style of market more understandable. Sure, we have innovation today, but it seems to focus on efficiency and middlemen. Society had taxis before Uber; we could even call 777-7777 from our smart phones to request one. Uber just reduced friction and provided an improved UX, certainly valuable but not comparable to inventing the internet or the personal motor vehicle.
There is a very small sample size (three secular market tops found in ’29, ’70, and ’00), so any conclusions need to be questioned at a minimum. However, of the secular market tops available for study, they either ended in dramatic over-valuation (blow-offs) or repeated testing of nominal new valuation highs. I have no idea what this current bull market will do, but it would appear, if we follow one of the two established tendencies, that there is still time on the clock. See below table:
Valuation is just one part of crafting a forecast and, for me, it’s the least useful. Valuations can remain high for years (as seen above) or low for years, like in the 1930s and 1940s. Just for fun though, the mean lag time would indicate a market top of early 2023 and the median lag time indicates mid 2022.