If you believe the bond market speaks to us through the curve, the recession clock started months ago, when the inversions came calling.
Of course, not everyone agrees the yield curve is an infallible Nostradamus. The lag between inversion and recession varies, which is problematic because eventually, economic expansions always give way to contractions. If the only test for recession predictors is whether they show up at some point prior to a downturn, then virtually anything could qualify. When you throw in the laundry list of factors which, together, make it very difficult to take the long-end at face value, there’s always a case to be made that curve inversions are false positives.
In any case, what about stocks? What are stocks saying? Well, recently, they’re saying something’s amiss. And that something is rising real rates, which have bled the S&P’s forward multiple to the tune of almost 25% (figure below).
As it turns out, the drop from the highs counts as the fifth biggest since the 1960s, according to UBS.
It’s not obvious that multiple compression, on its own, is indicative of stocks pricing a recession. I’ve variously argued that the real test will come when the guidedowns roll in and earnings estimates get cut. Target and Walmart offered a preview of what that might entail for equity prices.
But the decline in stocks is already indicative of elevated recession odds. For example, UBS’s logit model “suggests the equity market is pricing a 40% chance of a recession by year-end,” the bank’s Keith Parker said (see the figure below).
That’s 85th%ile. And 98th%ile outside of recessionary periods.
By contrast, the odds of a recession implied by US economic data are incalculably small (despite a negative Q1 GDP print, apparently), which makes for a rather glaring juxtaposition.
Who’s right? Well, it’s impossible to know, but suffice to say the outcome for stocks is binary. When the equity-implied odds of a recession rise to the 80th%ile or higher, and no recession occurs, stocks rally 12% on average with an 80% hit rate, according to UBS’s Parker. If, on the other hand, stocks were right to price a recession, they (stocks) fall 9% on average over the ensuing year, almost never managing to dodge a negative return.
You might suggest this is self-evident, and you wouldn’t be wrong. Stocks don’t do as well when there’s a recession as they do when there isn’t one. But, Parker’s analysis does speak to the likelihood of double-digit returns in the event recession fears end up being overblown.
Pessimism is pervasive, positioning is extremely bearish and cash levels are high. If the worst case doesn’t materialize, it’s probably not a stretch to suggest the recovery in risk assets could be even more pronounced than the historical average during “false alarms,” as it were.
Finally, have a look at the figure (below), which shows the S&P’s performance in and around the onset of Fed tightening cycles.
Referencing the chart, UBS noted that the average one-year performance for the S&P following Fed liftoff implies greater than 25% upside from current levels after this year’s 16% drawdown, which is pronounced by historical standards.
“The worst case scenario in previous Fed tightening cycles has been a 30% max drawdown,” Parker wrote. That would imply US equities have, at most, another 10% downside over the medium-term.
A max draw down of 30% may be a selective interpretation of historical data by UBS. On June 30, 1999, the fed started hiking rates in 25 bps steps and at the time the S&P 500 was around 1400. Through mid-2000, the overnight rate went from 5.0% to 6.5% but the S&P 500 essentially kept climbing (certainly not a straight line) to about 1525. By later 2000, the S&P started a downward trend and by early 2001, the Fed started cutting rates (went from 6.5% down to 1.75% by the end of 2001). The S&P 500 didn’t bottom until the end of 2002 at around 800. This is a nearly 43% drop from when the Fed initially starting hiking rates and a nearly 48% drop from the S&P 500s prior peak (again, which came after the Fed started hiking rates).
This, in a nutshell. I’ll add that analysts are constantly making two kinds of mistakes. First, overly focusing on the average historical performance in a given scnario, without attention to the size of deviations from the average. Second, presuming that historical outcomes provide the boundaries for future events.
The 10 year monthly chart in the SPY has us right at the regression trend-line, suggesting fair price right here.
The 2020 COVID sell-off was an overreaction to the downside. The subsequent Fed induced bubble was an overreaction to the upside. We are now right back at trend.
QQQ is modestly below trend. IWM is well below trend.
The 10 year monthly chart in the TLT has us well below the bottom of the regression channel, suggesting very over-sold levels.
10 years with a happy Fed tailwind. The wind direction has reversed. Don’t fight the Fed.
Their models say there is little or no recession risk. That will only change when the dubious job openings data plunges.
Remember what we were saying about 1Q being the valuation compression qtr, and 2Q the estimate cut qtr . . .
TGT NTM PE has crashed to 10.6X, less than a point above its NTM PE at the worst of the GFC. Estimates haven’t been cut yet. In coming weeks we’ll see where valuation is on updated estimates. On one valuation tool I look at, TGT might be discounting zero to negative FCF growth . . . forever.
Blown-up names like this, after they have seasoned a while, might be not the worst buys.
Time to do the work.
I’m not as conversant in the jargon and acronyms but which models are based on Stagflation? (last seen in the US 45 years ago?)
Rather than a clean up or down – energy/oil prices stay elevated (or rising) which also heavily impacts food – and that inflation occurs when the US (and global) economies slow down?
The gray area in the curve makes it seem like the paths taken by the market under consideration are many, enough to have confidence in the extremes’ ability to act as limits/constraints on the future path.
With a sample count lower than 100, I would argue this plot under-represents the likelihood of the current path’s breaching a previous maximum/minimum.
Plus, most people think of log-normal processes when they think of stocks: a histogram of returns shows a Gaussian distribution, and future returns are uncorrelated with previous returns.
Reality is that the random process that models stock paths best do have memory. In practice and for this case, it means that if the path has breached the lower limit (as it has), it’s more likely than we would think to do it again.