In recent weeks, I've noted that our estimate of the prospective market return/risk profile has shifted to the most negative 1% of instances we've observed in the historical data. Most of the time, a given set of market conditions is associated with some mix of positive and negative outcomes, so we focus on the average of those outcomes in the expectation that doing so will produce good results over the complete market cycle even if we are incorrect in specific instances. With regard to current conditions, there is an absence of redeeming instances where things worked out well, coupled with an abundance of starkly negative market outcomes that have accompanied similar conditions. This uniformity of bad outcomes is why I keep using the word "warning" lately. The market's prospective return/risk tradeoff here is highly unbalanced toward the risk side.
This isn't just a matter of advisory bullishness being high in one week or another, or even valuations being rich, or just economic risks appearing high. Rather, what concerns us most is the syndrome of evidence: the fact that we observe so many red flags at the same time - rich valuations, overbullish sentiment, heavy institutional saturation in "risk-on" trades, near-panic levels of insider selling, a burst of new stock issuance, overbought conditions (focusing on intermediate-term horizons), a two-tiered market that couples speculation in a handful of momentum stocks with broadly deteriorating market internals, a variety of historically hostile syndromes (see An Angry Army of Aunt Minnies), and increasing likelihood of oncoming recession.
Various observers will undoubtedly take issue with each of these measures. One can look at the Investors Intelligence bullish sentiment figure, which has eased back to 43% from over 50% in early April, but ignore that bearish sentiment is down to 20.4%, less than half of the bullish sentiment figure, and the lowest level since just before the 2011 market rout. One can look the market's price-to-forward-operating earnings multiple, which seems to be in an acceptable range, but ignore the stratospheric profit margins baked into earnings estimates. One can take issue with our recession concerns, choosing one rule-of-thumb or another that has gone "quiet" out of the broad ensemble of measures that we've presented over time, but ignore everything else we've written on the subject.
For example, our Recession Warning Composite is "quiet" here, as it usually becomes active only after a market loss of about 10-15%, yet still generally well before a recession is obvious to all (as was true both in late-2000 and late-2007). Strictly defined, the composite would require the manufacturing PMI to decline by a fraction of a point, year-over-year payroll growth to slow another 0.08%, and credit spreads to widen by about 0.25% here. The composite is generally a useful and early signal of recession, and it's clear that the signal last August was either false or more likely just deferred by monetary interventions. Still, we've always advised against focusing on any single indicator, and there's certainly no lack of additional evidence that I've presented on the subject of recession risk in recent months, so that shareholders can see the same things that I'm looking at. The value of research is that it constantly gives you better tools. Our research in areas like ensemble methods and noise reduction (including what we developed through our work in autism genetics) contributes firepower to that arsenal, and we try to approach economic and market issues with everything we've got.
Investors wishing to wait for a fresh negative signal from our Recession Warning Composite can do so, but should again recognize that it typically goes negative only after the market has lost some significant ground already. More often than not, stock market weakness continues well beyond those signals, but it's not a "market timing" tool and isn't intended for that purpose. It's worth noting that aside from the S&P 500 - which has benefited from monetary-driven risk-on speculation, the other components of that composite are still fluctuating within a hair of their respective trigger points. Meanwhile, however, it isn't helpful to ignore that we've never seen the components of economic activity as uniformly weak as they are today on a year-over-year basis except in association with recession (e.g. real final sales, real personal income, real personal consumption, employment growth, etc).
Just an analytical sidenote while we're on the subject: when evaluating economic risk, it isn't enough to show that some indicator has a high correlation with GDP growth. You also have to test that the indicator leads that growth rather than lags it. Otherwise it's not a suitable way to identify a turn. We've seen a lot of charts lately that fail to make that distinction.
Continue reading.
This isn't just a matter of advisory bullishness being high in one week or another, or even valuations being rich, or just economic risks appearing high. Rather, what concerns us most is the syndrome of evidence: the fact that we observe so many red flags at the same time - rich valuations, overbullish sentiment, heavy institutional saturation in "risk-on" trades, near-panic levels of insider selling, a burst of new stock issuance, overbought conditions (focusing on intermediate-term horizons), a two-tiered market that couples speculation in a handful of momentum stocks with broadly deteriorating market internals, a variety of historically hostile syndromes (see An Angry Army of Aunt Minnies), and increasing likelihood of oncoming recession.
Various observers will undoubtedly take issue with each of these measures. One can look at the Investors Intelligence bullish sentiment figure, which has eased back to 43% from over 50% in early April, but ignore that bearish sentiment is down to 20.4%, less than half of the bullish sentiment figure, and the lowest level since just before the 2011 market rout. One can look the market's price-to-forward-operating earnings multiple, which seems to be in an acceptable range, but ignore the stratospheric profit margins baked into earnings estimates. One can take issue with our recession concerns, choosing one rule-of-thumb or another that has gone "quiet" out of the broad ensemble of measures that we've presented over time, but ignore everything else we've written on the subject.
For example, our Recession Warning Composite is "quiet" here, as it usually becomes active only after a market loss of about 10-15%, yet still generally well before a recession is obvious to all (as was true both in late-2000 and late-2007). Strictly defined, the composite would require the manufacturing PMI to decline by a fraction of a point, year-over-year payroll growth to slow another 0.08%, and credit spreads to widen by about 0.25% here. The composite is generally a useful and early signal of recession, and it's clear that the signal last August was either false or more likely just deferred by monetary interventions. Still, we've always advised against focusing on any single indicator, and there's certainly no lack of additional evidence that I've presented on the subject of recession risk in recent months, so that shareholders can see the same things that I'm looking at. The value of research is that it constantly gives you better tools. Our research in areas like ensemble methods and noise reduction (including what we developed through our work in autism genetics) contributes firepower to that arsenal, and we try to approach economic and market issues with everything we've got.
Investors wishing to wait for a fresh negative signal from our Recession Warning Composite can do so, but should again recognize that it typically goes negative only after the market has lost some significant ground already. More often than not, stock market weakness continues well beyond those signals, but it's not a "market timing" tool and isn't intended for that purpose. It's worth noting that aside from the S&P 500 - which has benefited from monetary-driven risk-on speculation, the other components of that composite are still fluctuating within a hair of their respective trigger points. Meanwhile, however, it isn't helpful to ignore that we've never seen the components of economic activity as uniformly weak as they are today on a year-over-year basis except in association with recession (e.g. real final sales, real personal income, real personal consumption, employment growth, etc).
Just an analytical sidenote while we're on the subject: when evaluating economic risk, it isn't enough to show that some indicator has a high correlation with GDP growth. You also have to test that the indicator leads that growth rather than lags it. Otherwise it's not a suitable way to identify a turn. We've seen a lot of charts lately that fail to make that distinction.
Continue reading.
