Even before Harvey and Irma were set to punish Texas and Florida, erasing at least 0.4% GDP from Q3 GDP according to BofA and costing hundreds of billions in damages (contrary to the best broken window fallacy, the lost invested capital more than offsets the "flow" benefits from new spending, which is why the US does not bomb itself every time there is a recession to "stimulate growth"), things were turning south for the US economy, so much so that according to the latest Deutsche Bank model, which looks at economic data that still has to incorporate the Irma/Harvey effects, the risk of a recession starting in the next 12 months is near the highest it has been since the last recession.
As Deutsche Bank's Dominic Konstam writes, at first glance, the modeled probability is admittedly low at about 8% as of the end of August (down a touch from near 10% in June), but it has been generally trending higher despite a brief post-election dip. As a result, the bank "sees appeal to buying SPX put spreads and bull flatteners in Eurodollars given the emergence of downside risks."
How does Deutsche estimate recession risk?
"We use a probit model to estimate the probability that a recession will start in the next 12 months using the 1s10s Treasury yield curve, the unemployment rate less CBO’s NAIRU, annual core CPI ex-shelter inflation, aggregate hours worked growth, and the year-on-year change in oil prices. Unemployment’s proximity to NAIRU and soft core inflation are the key factors contributing to the appearance of some recession risk currently. Aggregate hours worked remains on a relatively healthy trend and oil prices are slightly positive year-on-year, however. While it has flattened significantly, the yield curve is also relatively steep."
On the other hand, as we discussed two months ago when observing the imminent Y/Y contraction in C&I loans, traditionally a guaranteed leading indicator of future recessions, other metrics demonstrate a far higher recession risk:
Konstam admits as much, saying that "if we look elsewhere we can find reasons to believe our 8% estimate is too conservative. We noted the slowing in C&I loan growth last week, which has rolled over from a recent peak near 13% to just 1.6% y/y at the end of July. This type of rolling over is consistent with what is typically seen during recessions, not in the build up to them. As we’ve noted, Fed reserve draining against the backdrop of a flat yield curve and potentially tepid loan demand may simply result in an outright contraction of bank lending as banks choose cash assets over loans, which would push this indicator further into what would be “recession levels” by historical standards."
When considering the more practical recession indicators, the Deutsche economist concedes that when working with the bank's rates strategy team, who previously produced a recession probability model that used the yield curve adjusted for the level of yields, shown a few higher recession probability:
Regressing the curve on front end rates shows that the curve is quite flat versus the level of short rates, and when we re-estimate our recession probability model using this metric instead we find a recession risk closer to 20%, having been as high as 25% in the Brexit aftermath. Outside of the last several years, such divergence between the two recession probability estimates has been highly unusual.
So what does the above mean for risk asset returns? Here is Konstam's answer for equities:
Given its construction and purpose to predict recessions over the next 12 months, there should be some forward looking information for asset returns. There is some evidence of a bias in risk assets in the months following a recession probability of greater than 15% (as is currently reflected in the adjusted yield curve probit model). On a 6m look ahead, the S&P sells off 32% of the time since 1968, but that rises to 45% in the 6m following a recession probability of at least 15%, and the median return falls about 2%. A recession probability of 30% is consistent with the S&P selling off 50% of the time. In addition to the negative skew to returns, delivered volatility rises more frequently in instances of an elevated recession probability. We have previously discussed the risk of volatility/ risk-off feedback loops, which the modeled recession risk suggests are a higher likelihood in the months ahead.
Next, for junk bonds:
High yield widening increases in frequency from 47% to 65% (since 1985) after conditioning on a 15% recession probability, and the median 6m change is a 40bp widening (versus ~10bp tightening unconditionally). Note in high yield there is a significant increase in probability of widening up to +250 bps (to date recent widening is quite muted, around +30 bps). Despite these biases in risk assets, there is less evidence of any consistent behavior in yields or FX when the recession probability breaks above a given threshold.
The biggest take-home message, however, is what these rising recession odds mean for the Fed's upcoming tightening actions, and while there is a discrepancy between various measures and indicators of recession risk which in turn complicates the ability to draw a firm conclusion, there are enough warning signs for Deutsche Bank to say that this uncertainty in and of itself "furthers our argument that the Fed would do well to take a pause in its tightening for the time being."
In other words, a Fed Funds rate just above 1.00% may be all the massively levered US economy can take before rolling over into recession, something first suggested by the various R-star analyses conducted here in 2015. It also means that in just a few months the US may be discussing NIRP and QE4 all over again.
DB's conclusion: "while we are relatively optimistic in our medium term equity view – falling equity risk premia in a low inflation equilibrium world mean equities are more likely to rally to bonds than bonds sell-off to equities – we maintain our near-term caution. While we don’t see recession as imminent, the blinking yellow lights mean that upside may be contained for now."