Earlier this week, when recapping the latest bearish outlooks from most large Wall Street banks, we touched on Deutsche Bank's latest House View take in which the bank - along with most of its peers - said that its strategists expects an "imminent correction" without giving much context for the bank's gloomy take.
So fast forwarding to today, the bank's chief equity strategist Binky Chadha expounded on this bearish take, and in a note pointing out that "equity valuations are extremely rich", he first explains "what's keeping equity valuations high", and then goes on to warn that "the risk the correction is hard is growing" adding that "there has been a clear negative relationship between initial valuations and subsequent forward returns, with the relationship strengthening as the return horizon is increased from 1 to 3 to 5 years out. For the highest decile of initial valuations, which is where we currently are, 5-year forward returns have on average been slightly negative."
Starting at the top, Chadha lays out what everyone - even the NY Fed president - is aware of, namely that equity market valuations are "historically extreme" on almost any metric: "trailing and forward price to earnings (P/E), enterprise value to EBIT, EBITDA or operating cash flow are all well into the 90s in percentile terms. High valuations are broad based across sectors and median company valuations are high."
Furthermore, he notes that "equities are very expensive (36%) relative to the drivers of valuation which have explained the bulk (70%) of historical variation in them." As a result, the cyclically adjusted P/E at 26.2x is well outside its historical 10-20x range "and easily the highest outside of the Tech bubble."
Some more details on the current extreme valuations:
- Valuations are high across sectors. Of the 11 sectors in the S&P 500, 7 are trading 10%-85% above their (ex-recession) average trailing multiples since 1995; 3 are trading in line (Materials, Health Care and Financials); and Energy the lone sector that is trading at a discount (-35%).
- Median company valuations are very high. DB looks at the valuation of the median company in the S&P 500 to reduce the impacts of outliers or particular sectors in the aggregate. For the median company, the traditional trailing P/E multiple is near a record high and well above Tech bubble peaks, while the multiple on forward consensus estimates is down from its peaks but at levels comparable to Tech bubble peaks.
- For the median company, EV to EBIT, EBITDA, or OCF are above their Tech bubble peaks. Unlike for the aggregate (37th percentile), for the median company EV/FCF is much higher (97th percentile) relative to history.
- Equities are very expensive (36%) relative to the historical drivers of valuation. According to his market model, which is based on payout ratios (+), earnings deviation from trend (-), inflation (-), real rates (+) and inflation volatility (-), which captures the large majority (70%) of the variation in P/E multiples over the last 85 years, BofA sees fair value of 17x. That leaves a 6.1 multiple point gap (36%) between the current P/E at 23.1x and fair value of 17x.
- The cyclically adjusted P/E (CAPE) is very high. From a long perspective, it is useful to look at cyclically adjusted P/Es. Arguably the best-known measure is the Schiller CAPE. But in to Chadha, that measure has a number of shortcomings as it relies on long moving averages and adjusts for inflation and is not comparable to other measures. Instead, he notes that there has been a very clear trend in S&P 500 EPS since the mid-1930s (when the US left the gold standard, a very different macro regime) and cyclical adjustment of S&P 500 earnings is therefore straightforward. Using this trend as a cyclically adjusted earnings measure, on current trend EPS of $172, the P/E is 26.2x, easily the highest outside the late 1990s bubble when it peaked at 35x.
Ok fine, everything is expensive, but if the Fed continues to inject liquidity there is no reason why stocks can't get even more expensive as every single dip is bought? In his response to this key question, Chadha asks rhetorically "why are valuations high" and debunks one by one all the widely circulated, popular narratives why the S&P is trading at 4,500. As the DB strategist writes, "extreme valuations are naturally fertile ground for new narratives and the list is long." Clearly, this is the case now on the up side, the opposite of what happened post the GFC when the S&P 500 multiple hit a low of 12x and stayed below its historical average for several years and below fair value for even longer. In his view, the list includes:
- low interest rates;
- the pandemic accelerating the adoption of technologies that raise productivity, margins and the level and/or trend growth in earnings;
- large firms (which dominate the major equity indices) gaining market share at the expense of small, raising the trend level of earnings;
- the share of high growth companies in the market has grown;
- the cash flow generated from earnings has risen; a paradigm shift in monetary and fiscal policies has reduced downside risks to macro growth;
- the increased participation of retail investors.
He discusses some of these in more details below, starting with why he doesn't think it is due to low interest rates:
- Rates have been here before, but multiples were much lower then. Looking across time, rates are where they were post the GFC when multiples were low and severely depressed relative to fair value. Similarly, looking across countries, interest rates are lower in Europe and Japan than in the US, but so are equity multiples.
- The equity risk premium is a much bigger component of the equity discount rate and inversely related to rates. Rates are a small part (10y yield currently at 1.3%) of the equity discount rate (average 10.3%), with the larger part being the equity risk premium (currently 8%). The equity risk premium has historically moved inversely with rates. That is to say, when rates fell (say because of low growth) the equity risk premium rose, for the same reason that rates fell in the first place. Historically there is a very strong negative correlation between rates and the equity risk premium (beta near -1), implying equities were priced largely independently of rates.
- The historical correlation between interest rates and equity multiples has been strong but shifted between long periods of positive and negative correlations. This ambiguity disappears when rates are broken into inflation and real rates. While inflation has robustly been negative for equity multiples, real rates have actually been positive.
- In the current context, rates are low because real rates are low, while breakeven inflation rates look in line with history. Low real rates argue for lower not the high multiples currently prevailing, as do inflation risks
Chadha then digs into the heart of the matter and looking at the recent surge in earnings relative to trendline, notes that he is "skeptical the 85-year trend in earnings has changed."
Since there are cyclical reasons for earnings to have risen above trend levels with the economic recovery, it is too early to tell whether either the trend level or growth rate has risen, either because of the adoption of technologies or gains in market share of large companies. This will only become evident over time, if and when earnings start to slow toward trend or that they do not. The 85-year trend has held through numerous regime changes and cycles that included the inflationary 1970s, the Tech boom of the 1990s and through and after the GFC. It held across various sector and industry group rises and reversals: as such Chadha remains "skeptical that the trend (normalized) level of earnings or the underlying growth rate has risen."
Going down the list, Chadha writes that the share of faster-growing companies is well within its historical range. Defining high growth companies as those with 3-year annualized growth in earnings above 15%, their share of S&P 500 companies, at 30%, is at the lower end of the historical range of 20%-55%. Their share in earnings at 45% is in the middle of the historical 20%-65% range, while in terms of market cap at 50% it is on the high side but well below the prior peaks of 60% hit pre-GFC and at the tail end of the Tech bubble. Nor has the earnings growth of high growth companies been higher than in it has historically.
No evidence that earnings are generating higher free cash flows (in fact one can argue the opposite). The ratio of free cash flows to earnings peaked recently at 1.1 and is down to 1.0. The recent peak was comparable to the peaks reached after the Tech bubble burst and after the GFC.
What about the recent surge of retail daytraders? According to Chadha, from a demand-supply perspective, retail investor participation in equity markets has clearly increased, and has been a driver of high valuations. This increased retail investor participation has been spurred by working from and staying home and encouraged by stimulus checks; retail participation had closely followed (inversely) the Covid time line, rising during lockdowns and falling with reopening, while retail investor surveys have shown the role played by stimulus checks.
Retail participation peaked back in January this year, fell and has been noisily going sideways since.
So if it the widely accepted narratives are not behind record valuations, what is it? Chadha next lays out his own view on what is behind the relentless market surge, starting with the surprisingly rapid rebound in earnings which "has been key":
S&P 500 earnings have beaten the bottom-up analyst consensus estimate by an unprecedented 15-20pp for 5 quarters running. Since the analyst consensus is informed by a dialogue with company managements, large recurring beats suggests that companies themselves were surprised by the strength and speed of the recovery. Meanwhile, the market multiple on forward consensus earnings has been in a relatively narrow range. So price returns have been about equal to changes in forward consensus estimates, which in turn have been rising rapidly, buoyed by the big beats.
But after a torrid 18 months, the analyst consensus looks to have caught up and Deutsche bank - like us - look for earnings upgrades to slow or even end (note the stark recent profit warnings from the likes of PPG). According to Chadha, earnings beats and upgrades will slow, "diminishing if not ending what has been a key driver of equity upside, particularly year to date and supported multiples." The bottom up consensus for S&P 500 EPS in 2021 has been rising rapidly, from its lows last summer of $160 to $202 presently (+26%). This compares with DB's estimate of $210, which suggests 4% upgrade upside. But this is actually about the typical 4-5% historical beat, suggesting less likelihood of upgrades. The analyst consensus for 2022 has also been rising and at $220 now, about in line with DB's estimate of $222. Historically, Chadha observes, "there is a positive correlation between S&P 500 analyst estimate changes and macro data surprises, which after remaining positive for a little over a year turned negative in late July this year."
And while lofty expectations are set to disappoint, the earnings cycle is also very advanced. S&P 500 earnings had already recovered back up to trend levels by Q4 of 2020; are now 10% above; and could rise to 20% above by year end. With typical expansionary phases of the cycle seeing earnings go 20% above and past records of 30%, there is room for them to go further above trend, but the cycle is clearly very advanced according to DB.
So cutting to the chase, Chadha argues that at a fundamental level, "the key reason multiples are high is market confusion over where we are in the earnings cycle, in part reflecting the speed and surprise with which the economic recovery has unfolded and the large persistent beats this generated." Given that GDP is still well below trend, there is a popular view the better part of the recovery is yet to come. However, the parts of the economy that S&P 500 companies are exposed to are already significantly above trend. Indeed 2/3 of S&P 500 industries have activity levels 1sd or more above trend; about half of those, so about 1/3 of the S&P 500, 5sd above! So - and this probably is the best summary of the DB bear case - "the cycle is much more advanced and the risk is that activity begins to slow, while the market is priced for most of the recovery as yet to come and large beats to continue."
Which finally brings us to the punchline: the coming "hard correction."
According to Chadha, valuation corrections don’t always require market pullbacks - indeed multiples can shrink at roughly the same pace as earnings are growing resulting in unchanged stock prices - but they do constrain returns, or as Chadha puts it, "while equities are very expensive, by itself this does not necessitate a large market correction as valuations compress when price increases are smaller than earnings growth." Of note, historically valuation “corrections” in equities happened in slow motion, i.e., equity prices rose by less than earnings grew; and took place slowly, on average over 3 years. That is certainly not the case this time.
Furthermore, while valuation corrections are likely to be softer, i.e., not entail market drawdowns, the earlier in the recovery earnings are as growth is rapid. But with the current cycle advancing very quickly - recall this is also Morgan Stanley's core thesis, as the bank sees the current cycle burning "hotter but shorter" - Deutsche Bank warns that "the risk that the correction is hard" i.e., one requiring a market drawdown - "is growing" for the simple reason that the current cycle "has been advancing very quickly and so the risk that the valuation correction is harder is growing. "
If this is the case, and if a correction is imminent, it is also intuitive that higher valuations will lower returns longer term, since the "market has pulled forward much of the returns of this recovery."
Historically, looking at forward price returns vs initial valuations, there has been a clear negative relationship between initial valuations and subsequent forward returns, with the strength and magnitude of the negative relationship increasing as the return horizon increases from 1 to 3 to 5 years out. Looking at price to trend EPS as the valuation metric, the average and median price return is highest for low initial valuations, then goes mostly sideways to gently down for intermediate valuations, before falling notably for high valuations. For the highest decile of initial valuations, which is where we are currently, 5- year forward annual returns were slightly negative on average. There is a wide dispersion in returns around this average, ranging between -10% (10th percentile) and +5% (90th percentile). But even an outcome at the higher end would mean lower than average returns.