Reflections [Expanded version]
Headwinds increasing for the stock market
NOTE : All images and charts displayed below are regularly updated and available to subscribers from the CHARTS menu.
The SP-500 has enjoyed an incredible 40% rally since the 23 March lows. This was mostly fueled by an unprecedented FED stimulus program, which is now coming to an end within the next few weeks:
Unless more stimulus is unleashed (and a further $1.5-$2 trillion seems likely) the risks of a stock market selloff are high, especially given that the market is some 12.7 to 22% overvalued according to our RAVI US Valuation model (Also see our bear market warning to clients in June 2019):
The unlocking of the US economy has now led to a 2nd wave of infections and the US economic mobility index (an excellent high frequency proxy as an economic indicator) is rolling over and falling back to 45% of the pre-recession level:
The stock market will not take kindly to this if the rollover persists, and there is every likelihood it will persist, since (1) the economic mobility rollover is broad and covers most US states that matter to the economy (log in to see the next 8 largest economic contributors):
…and (2) the Covid-19 infection spikes are also occurring in most US states, meaning mobility will remain under pressure for a while:
The traditional high-frequency macro-economic data (WLEI2) has been rolling over in the last 2 weeks in sympathy of the economic mobility data:
Interestingly, ECRI’s WLI keeps increasing and is no doubt contaminated with out-sized improvements from weekly claims coming off unprecedented lows which is masking the other components in their series. Since our WLEI2 limits component movement contributions to within 2 historic standard deviations for this exact reason, it does not succumb to this issue.
From a seasonality perspective, the month of July is not very bullish with a score of only 23.8 out of 100 according to our Seasonality Model:
If there is indeed a correction in July, then this will set-up August for a high confidence bullish seasonal month especially for the Presidential (4-year) cycle which suggests a rare 2x leverage month in August. We mention this since the 4-year presidential cycle has outperformed all the other cycles over the last decade:
The risk/rewards are therefore skewed toward the bearish side in the near term and a 5-10% correction is probable unless the conditions we mentioned above change materially. On the positive side we fully expect any dip to be bought hard as undoubtedly the FED will issue another round of stimulus and nobody has ever gone broke by buying the dip on the back of the start of any new stimulus.
Do not expect any pleasant low volatility long-term rallies though. Literally 3 days after we issued a future increased volatility trend warning on the Long-term Volatility Prediction Model chart update on 17 Jan 2020, the VIX average has been rocketing upwards which is likely to persists to roughly early 2022 (with VIX peaking roughly 12 months earlier around early 2021):
US enters 2nd wave of Covid19 infections
Note : Most charts shown below are available to subscribers in the COVID19 analytics section.
- Economic mobility in US increasing at a slower pace than Covid-19 infections, contrary to rest of G7.
- At a state level, New York, New Jersey, Massachusetts & Michigan are leading economic mobility recovery vs infections.
- US has moved from a peaked scenario to join a host of second-wave countries struggling to contain infections after lockdowns.
- Stock market is going to struggle to post new highs until daily infections decrease again or FED resumes stimulus.
US versus rest of G8
Instead of using traditional economic indicators (most of which have a month or more delay) we can examine high frequency (weekly) data from the Google community mobility indices to try and figure out how economies are recovering. These indices use Android smartphones and Google maps requests to anonymously track movement trends over time by geography, across different categories of places such as retail and recreation, groceries and pharmacies, parks, transit stations, workplaces, and residential.
During lockdown, residential movement typically goes up and workplace movement trends down. To adequately track traditional economic activity, we can subtract residential movement from workplace movement to derive an Economic Mobility Index and represent this as a % of the pre-lockdown trend.
Instead of plotting the Economic Mobility Indices versus time, we can plot them versus cumulative infections to achieve a two dimensional view to gauge mobility improvement versus infections, as we have done below in charts comparing the US to the G8 developed economies which represent over 50% of global economic output.
The charts have been designed so that we can use the 45-degree angle line as a guide for how well we are doing. If the mobility curves are trending above 45-degrees, mobility is improving at a faster pace than infections, which is a desirable outcome. If the curves are trending below 45-degree angles then infections are outpacing mobility improvements which is a less desirable outcome.
We can see from the above chart that the US mobility curve is at a 19.4 degree angle which means we are not getting bang for our buck with the lifting of restrictions, as infections are outpacing mobility improvements. The UK at 43.1 degrees is faring much better, with infections marginally outpacing mobility improvements whilst Canada at 46.7 degrees angle has mobility marginally outpacing infections.
France, Italy and Germany are doing very well with angles in excess of 73 degrees. They are getting great bang for their buck by relaxing restrictions and there appears to be no post-lockdown “2nd waves” of infections.
Whilst the national US figures are interesting, we can see that outcomes on a state-wide level vary significantly:
Some explanations on the charts above will prove useful. These weekly charts of the 12 main US statewide epicenters contributing 70% of total US infections and 60% of total US GDP are designed to measure the pace of economic re-opening and recovery as well as provide early warning of a potential 2nd wave of infections that could derail the recovery in question and send the stock market into another tailspin.
- The LEFT vertical y-axis = 7-day average of workplace less residential Google mobility indices as % relative to pre-lockdown baseline.
- The RIGHT vertical y-axis = Rt reproductive rate – the average number of people who become infected by an infectious person. If Rt is < 1.0, virus will stop spreading.
- The horizontal x-axis = cumulative Covid-19 reported infections (in thousands, measured daily).
FOR THE BLUE Mobility curve:
- Flatter slopes < 45-degrees are less desirable outcomes and indicative of slow mobility recovery and/or increasing infection rates (infections rise faster than mobility)
- Steeper slopes > 45-degree angles are better outcomes characterized by rapid mobility recovery and/or decreasing infection rates (mobility rising faster than infections)
FOR THE ORANGE Rt curve:
- Flatter slopes less than 1 are more desirable outcomes and indicative of lower growth in disease infectiousness (less likely 2nd wave of infections)
- Steeper slopes more than 1 are worse outcomes indicative of higher growth in disease infectiousness (more likely 2nd wave of infections)
- Rt Slopes < Mobility slopes are highly desirable, meaning mobility is increasing faster than infectiousness.
ABOUT THE SLOPES:
- Slopes for mobility & Rt curves are represented as angles in degrees, with horizontal being 0-degrees and vertical being 90-degrees. Angles below horizontal are negative
- Angles are calculated as the slopes of the trajectories of the curves as defined by the last 14 days data points.
US states seem to be either doing really well or really bad. There are very few in-between.
New York, New Jersey, Illinois, Massachusetts & Pennsylvania are all performing exceptionally well with:
- mobility outpacing infections,
- mobility outpacing Rt and
- Rt less than 1
Michigan is performing reasonably well with mobility outpacing infections, mobility outpacing Rt but Rt greater than 1. This could fall back into a 2nd wave of infections.
Maryland & Virginia are performing “OK”, with Rt less than 1 but infection growth still outpacing economic mobility improvements.
All the other 4 states are performing badly with daily infections outpacing mobility improvements, mobility improvements marginal and Rt > 1.
Overall US Covid19 Scorecard
Apart from a few large states mentioned above, the US National Covid19 performance peaked a week ago and appears on the decline as it enters a 2nd-wave status. We track 10 metrics daily to score the US National Covid19 situation, with the latest status below:
Our highest 7-day average of daily score of 8.1 was achieved on 28 May 2020 and has been on a steady decline since as various metrics deteriorated.
The US now joins the “second-wave” club – countries where daily infections peaked, but then started growing again as lockdowns were lifted:
Regardless if you consider this wave still being the first wave or part of a second wave, the result is the same – US stock markets will struggle to continue posting higher highs and volatility will increase until one of two things happen:
- Daily infections peak again nationally or at least in the big 12 states we monitor
- FED responds to increased infections with more stimulus
NBER declares 2020 recession dates
The National Bureau for Economic Research (NBER) has announced official start dates for the 2020 US recession. It is very rare for such quick pronouncements (they are normally made 9-12 months after the fact) but the fact that 90% of the economy came to a sudden halt, has led to such deep declines in their metrics that they could make an early pronunciation without risk of being proven wrong later.
The monthly economic peak was declared as February 2020 (first month of recession is March 2020), while the quarterly peak occurred in 2019Q4 (first quarter of recession is 2020Q1).
Note that most of the media and press is interpreting this incorrectly, saying the US entered recession in February 2020. This is of course not true, the economy made a peak in February (its best showing this business cycle) and fell into recession in March (March was first month of contraction).
The NBER declarations are consistent with the dates we have been using in all our charts/reports since mid-April already.
Our first proclamation was made in the 17 April 2020 weekly SuperIndex PDF report and has remained unchanged since it first appeared, namely:
Since this date you would have noticed all NBER recession shading in reports and charts to have commenced in March for monthly frequencies and 1Q2020 for quarterly frequencies.
You can read the NBER declaration over here
We are estimating the economy to trough in May or June, but more on that later.
Some updates & market observations
The STM Seasonality Model is a unique composite that looks at average monthly gains, gain-to-loss ratios and percentage of winning months for 1,2,3 and 4 year cycles to arrive at a composite seasonality score for each month. For the last 18 months, the model has been running at 80% directional accuracy on calls on the SP500 future direction which is rather remarkable given the strange times we are living in.
Even though May month was forecast as a non-leveraged long month, the gains for May are not expected to be that large, in fact they could even be slightly down. It is the months of June and especially July that are materially bearish though:
The STM chart is updated with quite a bit of breakdown detail once per month with a 3-month look-ahead, as shown below:
You can read about the STM methodology at the following research note.
You will notice that all charts, data-files, dashboards and SuperIndex reports continue to be updated through Saturdays early morning (4am) and by no later Sunday 5pm US eastern time.
We are also now tracking the current Seasonality Signal, the Yield Curve Complex Diffusion and US Covid-19 situation in the new DASHBOARD page:
You will note from the above that 1 of the 10 yield curve aggregate components has inverted, moving us out of the SAFE zone. You can see from the Yield Curve Chart that it is the two less one year that has inverted:
You can read about this recession forecasting model in our August 2019 warning of an impending recession in this link .
Whilst our August note seemed prescient in penning in an earliest recession date of April 2020, we were hopelessly wrong on the severity of the recession, since this was triggered by an exogenous (black swan) shock followed by a voluntary hard stop of the majority of the US economy due to the Covid-19 global pandemic (which we warned of in first week of March 2020, before WHO). As we now know, this recession is anything but shallow, breaking all the historical records on just about any measure you care to look at.
While we are on the Covid19 topic we are now tracking the daily US Coronavirus situation in the dashboard gauges as well, and this just indicates the last pane in the COVID19 dashboard which is the score out of 10 :
At the moment, 8 out of the possible 10 metrics we track are moving in the right direction and as you can see the scoring metric has steadily climbed from a reading of two to well above 6 now.
SP-500 and Recessions
We examined SP-500 behavior in the lead to and during US recessions a few years ago in an old research note (Recession – Just how much warning is useful anyway?) to conclude that more than 5-months warning before a recession was not constructive, and that you should focus on recession warning models that stuck to a 4-6 month historical lead time as close as possible.
Given the “voluntary” sudden-stop of the U.S economy due to Coronavirus lock-downs, we are faced with an “artificial” non-systemic recession that was not caused by the usual financial imbalances, but rather an exogenous event. In such situations, leading economic indicators offer little hope of warning and are currently operating as coincident indicators.
The statistics still offer some useful guidance though, even if we acknowledge we are comparing a non-systemic recession to historical systemic ones. For example, the recent correction delivered a peak-to-trough decline of 33.92% versus the average month-to-month recessionary decline of 30%. However, the recent 23-session correction compares to the historical duration of some 13 months!
We examine the historical behavior of the SP500 around US recessions with seven metrics as shown below, using monthly closes (as opposed to daily closes where draw-downs would be higher):
- P2NBER is the amount of months taken from the SP500 peak to the first month of recession.
- DROP1 is the draw-down in the P2NBER period (monthly closes)
- NBER2T is the amount of months from first month of recession to the SP500 trough
- DROP2 is the draw-down in the NBER2T period. (monthly closes)
- DROP3 is the total peak to trough draw-down experienced. (monthly closes)
- T2EXP is the months from SP500 trough to the first month of expansion.
- RISE1 is the size of the rally from the SP500 trough to the first month of expansion.
The metrics appear below in tabular form:
We conclude, on average, that the SP500 peaks some 7 months (AVG) before onset of recession, but with a wide standard deviation (DEV) of 4.58 months. This means the stock market can peak anywhere from 2.6 to 11.72 months before recession. Total declines average 30% but again with a wide 15.4% standard deviation meaning any draw-down of 14.6% to 45.4% is statistically possible.
The statistics for the post-trough metrics have far less variances though. If we take the standard deviation and divide it by the average we get a Coefficient of Variance (CoV) which is a dispersion metric. The statistical dispersion for the post-trough metrics are far less than those of the pre-trough metrics, as you can see in the last row of the above table.
The SP500 troughs some 5.29 months before the first post-recession expansionary month, with a small 1.16 month standard deviation. This means the stock market can trough anywhere from 4.13 to 6.45 months months before the economy resumes expansion again.
We can do a fun exercise with these low dispersion metrics. If we deem the 23rd March 2020 SP500 low of 2,237 as the ultimate trough for this business cycle, then we could expect to emerge from this recession anywhere between July 2020 and September 2020 (averaged at August 2020), with SP500 trough-to-peak gains of 17.4% to 37% (averaged at 27.2%) This gives us SP500 targets of 2,626 to 3,064 (averaged at 2,845.)
Interestingly enough, these figures are not too far off the RAVI 3Q2020 forecasts as shown below, meaning such a rally would theoretically remain within “sane valuations” territory. Unfortunately, we have already rallied close to this point, meaning any small term pullback would be welcome. However, with the market following the “FED put” and unprecedented quantitative easing ( economy and Cornavirus be damned) we should not be surprised to see the market rally way beyond 3,000 in the near future.
We will soon see if the WLEI2, SuperIndex and USMLEI leading indicators trough around the July/August time-frame to confirm the above assumptions.
COVID19 Recession Warning
Businesses are going to be shuttered in massive numbers as the U.S has to deal with the unavoidable nationwide lock-down that will be required to contain the highly contagious Coronavirus. From our Covid19 Dashboard we maintain for our subscribers, we can see that the number of cases is rising according to a quadratic equation that will yield over 100,000 cases by the end of this week and over 500,000 cases by 7th April (assuming trends hold.)
Hospitalizations are running at 9% of cases and given that the U.S has an estimated 46,500 ICU beds (maybe double in a “wartime crises”) we estimate 50,000 hospitalizations by 7th April at which time the health system is going to be under severe strain.
Notwithstanding the humanitarian crises, the unavoidable global and U.S lock-down is going to hurt the US growth story very badly and in very short order. We predict over 2 million unemployment claims will be filed this week which is going to decimate the upcoming WLEI print as our estimation below shows:
This will be the first recession trigger for the WLEI since the global financial crises and subsequent recession of 2008. The Weekly SuperIndex (which is a pseudo weekly index with weekly and monthly components) is likely to follow within 2-3 weeks after that.
As we have stated before, when we have a sudden and dramatic stoppage of all economic activity due to a humanitarian crises (or any other exogenous event) then precious few leading indicators will provide adequate warning especially if no financial imbalances were present to raise recessionary concerns in the first place.
Apart from our Long Leading Indicator (US-LONG) in the US Monthly Leading Index report (depicted below)
…the only other warning we had was from the RAVI which warned in 2Q19 that the stock market was dangerously overvalued and started predicting negative returns ahead. As with most valuation based models, the over-exuberance could remain in place a lot longer than your short-positions could remain liquid, but what these valuation models DO provide is a danger signal for a fragile overvalued market that will not be able to handle an exogenous event. Lo and behold, we had an exogenous event against the backdrop of a highly overvalued market and the most brutal market collapse since the great depression ensued. But be clear that sans the Covid-19 exogenous event, the RAVI model could have had divergent forecasts with actual stock market outcomes for potentially much longer, since bear markets rarely commence based on valuation concerns alone.
Given the sudden stop of the economy, we are treating all our leading indicators as very-short to coincident economic indicators until further notice.
With most of the damage already done on the stock market (we could see another 10-15% down-leg, but RAVI is seeing upside so we are at least “moderately priced” now), we turn our eyes to using the leading indicators to anticipate a turnaround in the economy. One of these will be the stock market itself which in turn will be looking for signs of peak-infection in the US, and that is why we have focused charts for the US Covid-19 outbreak.
When hospitalizations, daily new cases, days-to-double and active sick (cases less recovered) numbers peak, we will have our cues the worst is over and the U.S is winning the war on Covid-19. Once the market sees that, odds are very high a new bull market may commence.
COVID-19 Global Pandemic is here
Further to our March 5th 2020 warning on a looming Coronavirus (COVID19) global pandemic, the WHO has finally recognized as such and declared the outbreak an official global pandemic. It is not hard to see why, when one looks at the chart below:
Whilst China has managed to stabilize new infections (assuming their numbers are to be trusted) the rest of the world does not have the luxury of their socialist command-and-control government, hospital building productivity, general mobilization, hive-mind population and draconian quarantine ability. And it shows with the infections spreading to over 125 countries at an exponential rate and more worryingly, rising mortality rates and falling recovery percentages. Within the next 3 days China is likely to represent under 50% of total cases. And remember, these charts just depict reported, confirmed cases – who knows how many are either under-reported or even unreported (as they are not known).
Governments are waking up a bit late but catching up fast with wide-sweeping quarantine measures announced on the hour across the globe. Schools and universities are being shut down, entire states and even continents are banning travel among one another, major sporting events are being cancelled, religious gatherings and general public gatherings are being banned, courthouses, theaters, cinemas, gyms, children playgrounds and nightclubs being shuttered by governments and large (even global) companies closing doors and asking staff to self quarantine and work from home.
Since this outbreak is likely to be with us for several months, potentially even forever (as with influenza) we can say without a shadow of a doubt that this outbreak is a once in 100 year event and our (out-sized) social reaction to its spread is likely to have a profound effect on the way we live and interact with the environment and each other – at least in the short term until we know more about it and/or a vaccine is found. It is also going to have a profound (but short-lived we believe) effect on every single economy of significance on the planet.
Despite the huge policy/media impact this virus is having across the globe, it is still in the little leagues when compared to other diseases humans have become used to living with. The (incomplete) list below shows the worlds’ deadliest diseases, sorted by their mortality (death) rates and also showing the number of estimated global deaths:
How does a disease with “only” 135,000 infections, relatively slow infection rate and less than 5,000 deaths illicit such an earnest response from us humans, given its lowly stature in the above list? Most likely its due to its newness (its what we don’t know about it that’s scary), lack of a known vaccine, its rapid spread compared to the more recent Ebola/SARS/MERS scares, its relatively high mortality rate and of course the prevalence of online and social media giving daily blow-by-blow accounts of the spread of the disease. Also, deep down, many people realize that this virus could be with us for a long time, just like the common flu which has been around for 2,000 years, and knowing how common and easy it is to catch flu, the fear of the unknown aspects of the disease coupled with the stigma of catching it and the uncomfortable mortality rates, are enough to send anyone heading for the hills.
If we look comparatively at the flu and Covid19 statistics, one cannot help but wonder if the current market reaction is overdone. In fact our policy and sociological responses to Covid19 are likely to cause more economic damage than the virus itself.
There are no leading economic indicators for these “black swan” events, but you can be assured it will have out-sized effects on most, if not all, our leading economic indicators, US and global alike. Whilst some US recessions are triggered by weak leading and co-incident economic data, there are a fair share of US recessions caused by an exogenous event such as this pandemic, when accompanied by US economic data that is weak and vulnerable or when valuations are stretched to the extreme – exactly the position we are in now.
For this reason, it is quite feasible that the U.S recession probabilities currently painted by any macroeconomic models (our most pessimistic, apart from the RAVI model, being 35-50% odds) could be shy of the realities of the exogenous event currently being played out. As most of the models are a snapshot of December 2019, with some January 2020 and the Coronavirus only catching the attention of the world late January 2020, we will only now begin to start seeing the probabilities catching up with the reality.
Apart from the RAVI Valuation model which was warning of extreme market risk and negative future equity returns as early as 2Q19, we are currently witnessing reactions in two other indicators.
1. The Weekly Leading Index (with the SuperIndex likely to follow suit in ensuing weeks)
2. The Composite Market Health Index (CMHI) – which has just dropped recommended exposure from 100 to 67%, as a result of the number of long term falling trends in SP500 now exceeding long term rising trends. New annual lows average is also set to overtake new annual highs average in the next few days which will signal a further 33% reduction in long term exposure:
Also note the upcoming Weekly CMHI is likely to make a print below 0.3, which also serves as a market warning, as per this research note.
Having said all that, the stock market correction is now at extreme levels (SP500 trough probability model > 95%, Great Trough (GTR) model, Selling Pressure Diffusion (SPD) model and the Zweig Breadth Thrust (ZBT) model) and out-sized reaction rallies are highly probable from current levels shortly. The speculators can take advantage of these when the signals come or the longer-term folk who subscribe to the thesis that all things economical are going to get worse from here can sell into the rallies. There is going to be a lot of selling into the rallies by speculators caught flat-footed in this steep decline (dare we call it a crash?), and traditional buy-the-dip strategies may therefore no longer be fit for purpose.
COVID-19 starting to look like a global pandemic
The newly reported cases of Novel Coronavirus (COVID-19) in China appear to be tapering off, but it is the recent uptick of newly reported cases outside China that have reached alarming levels, resulting in total cases accelerating to just under 100,000:
The secondary round of infections, most likely from travelers from China before the largest quarantine in human history, is evident when one looks at the progress in the number of countries reporting confirmed infections, with a marked jump since 23rd February 2020, to just over 90 countries.
The time-lapse between China and rest-of-world infections is worrying scientists that the virus has gone largely undetected in the rest of the world, particularly 3rd world countries with immature health infrastructure. This may be due to delays associated with travel propagation, up to 14-day incubation period compounded by no visible symptoms during incubation, and delays in country preparedness and detection capability. Granted, many countries are reporting less than 5 cases, but the ballooning of infections recently in South Korea, Italy and Iran (apparently 8% of Iran parliament is infected) give an idea of the potential scale of this disease in individual 1st and 3rd world countries alike:
It would be foolish to assume reported confirmed cases constitute 100% of the sample set, so its a question of how much is being unintentionally unreported and intentionally under-reported. There are those that estimate China has purposely under-reported infection by 10-fold or even more.
The percentage of cases from China has been dropping steadily to just over 84% as infections spread faster outside China:
Given that China represents just 20% of the world population, and giving China the benefit of the doubt that they have indeed now contained the virus and we can believe their reported numbers and they top-out at 90,000 infections, we can make a best-case assumption that a global pandemic could easily surpass 500,000 infections.
With the current mortality rate at around 3.4%, this could imply 17,000 deaths globally:
The worst case assumptions are too ghastly to contemplate with many scientists saying up to 40% of the globe could be infected, resulting in over 100-million fatalities.
For market watchers, the concern is the out-sized economic effects of fear, panic, quarantine (almost 300 million students quarantined worldwide now), collapsed travel and tourism, plunging industrial production and world trade volumes and cascading disrupted supply chains.
With the global economy having looked like it was just emerging from a business cycle downturn in December 2019, the COVID-19 outbreak could be a real economic recovery spoiler, as shown from charts taken from our detailed December 2019 monthly Global Economy Report:
Even if the U.S manages to avoid an outbreak, their economy will not. Prior research shows that US leading economic data and indeed even stock market returns, have much higher-than-expected correlations to global economic conditions outside the U.S.
Also, with monthly leading US data looking vulnerable and future US recession probabilities in the 28-52% range, the US economy may not have enough buffer to avoid a local recession with a protracted continuation of the current global business cycle downturn:
If you are interested in tracking the daily progress of this virus outbreak, the best place we have found is here. Just be careful not to read anything into the current days numbers, as they adjust dramatically overnight as they are invariably incomplete. The prior days numbers are the ones to focus on.
Massive rebound in US housing market
All 8 components of our comprehensive US Housing Market Index have posted solid and sustained gains in the last 6 months:
Our detailed PDF report for Dec 2019 has been published to the REPORTS menu.
According to many market watchers, there is no better barometer on the health of the U.S. economy than housing. It’s an industry that encompasses a myriad of vital sectors — banking, manufacturing, commodities, construction, durable goods, international trade, transportation and, of course, consumer spending. So it’s not surprising the Federal Reserve closely monitors housing trends in the course of setting monetary policy.
Sound economic growth in the U.S. is not possible without a robust residential real-estate market and in fact 7 of the last 11 declines in the housing market has led to economic recession and with 11 U.S recessions since the end of World War II, all but two were preceded by a big decline in the housing market.
Even though housing does not account for all that much of the economy, its role in recessions is huge, because it is highly cyclical and sensitive to interest rates. Housing has never accounted for more than 7 percent of total US GDP, but it has on average accounted for about a quarter of the weakness in recessions since World War II, according to a 2007 paper by Mr. Leamer titled “Housing IS the Business Cycle.”
After housing, the sector that has historically been second most important to recessions is consumer durables, or expensive purchases like cars, furniture and appliances. Those are often connected to the housing market’s prosperity because people usually buy other things when they purchase a home.
We found some worrying signs in labor data
These days, its really hard to find worrying signs in US labor data. If one looks at the once famous Janet Yellen Labor Dashboard, apart from Job Openings, everything looks to be progressing fine, bar a small pullback here and there:
Sure, the employment-to-population ratio (participation) has not come close to peak achieved in the last business cycle but everything else has.Even the equal-weighted 52-state US national average unemployment rate has been falling nicely to multi-decade lows. But dig a bit deeper and lift the hood on this state unemployment data and you get a shock:
An increasing number of U.S states, in some cases more than half, are reporting rises in their unemployment rates by the 3 main measures we look at, namely:
- net percentage of states with rising unemployment,
- percentage of states with unemployment higher than low of the last 6 months
- percentage of states with unemployment higher than prior month.
The rising of these breadth metrics are strong leading indicators for the future direction of the aggregate nationwide unemployment rate and whilst multi-year rises below 50% are common, such as we have witnessed from 2014 to 2019, it’s when they persist in greater than 50% territory that one starts to have serious introspection about the U.S economy.
We’re not there yet but this bears watching closely.
IMPORTANT : Once in 4-year opportunity to secure a RecessionALERT annual subscription at 50% discount. See our SUBSCRIPTION page
U.S economy likely dodged a bullet
For two years, our comprehensive U.S monthly Leading Index (USMLEI) has been deteriorating, more recently to worrisome levels, with exactly half of the 23 components now in recession territory:
What was really alarming was that this was occurring against the background of elevated RAVI local stock market valuations, an inverted yield curve, a deteriorating Global LEI, a US housing market recession and a global trade recession.
We think the worst is likely over however. Notwithstanding a likely de-escalation in the US-China trade war and a massive $465Bn jump in liquidity via upcoming FED Open Market Repo Operations, all the other concerning factors bar valuations seemed to have turned corners.
Let us get to the elephant in the room first however – US stock market valuations. Our most accurate 1-year ahead forecast on the SP500 total return comes from the average annualized RAVI forecasts from the 1,2 and 3-year forecast models. This is forecasting muted short-run gains ahead for the SP-500:
This is the only model remaining bearish – however if you look at the prelude to the last two recessions you can see that a r-squared of 0.47, whilst remarkable for such a short run indicator, does not preclude the actual outcomes from diverging from the forecast for 1-3 quarters. The Sept ’06 scenario is a case in point. But for now, we have to be prudent and acknowledge that the current RAVI valuation model is not supporting favorable US stock market outcomes in the year ahead.
Next up is the yield curve inversion, which has for all intents and purposes UN-inverted after a short 10-week stay below zero:
As we have stated several times in the recent past, we were never convinced at the completeness of this inversion since at most 70% of the term-spreads inverted and history has shown that 90-100% need to invert before we can be assured it is a proper recession signal. Most importantly, the 10’s LESS 2’s failed to invert at all. We remain behind the hypothesis that we could likely be witnessing a false (incomplete) inversion as witnessed in 1999 as depicted in the chart below:
Next up is the all important Global LEI, more specifically the percentage of OECD countries with rising LEI’s, which improved considerably in October:
The hypothesis is that the direction of the blue line dictates the direction of the USMLEI and that the collective direction of LEI’s of countries outside the US can lead direction of the US leading data (see this note). We would be seeking a further improvement in the blue line for the month of November, potentially even a rise above 50% to signal a rebound in the leading US index.
Next up is the US Housing market which has staged a sharp recovery from a short recession and is a long leading indicator for the US economy. The theory is that there is no better barometer on the health of the U.S. economy than residential housing. It’s an industry that encompasses many vital sectors of the economy such as banking, manufacturing, commodities, construction, durable goods, international trade, transportation and, of course, consumer spending:
Finally, we can take some encouragement from the high-frequency leading data, where our Weekly Leading US index has been rising sharply of late – the theory being that this almost always leads to the Weekly Leading Super-index and Monthly Leading Index following suite:
There are very few guarantees in this business, just probabilities and likelihoods – and a betting man might conclude that things will be improving for the leading US economic data in the near future. There are a lot of tailwinds to support this theory at this juncture.
It is probably safer to be betting for the economy rather than against it right now.
New intraday charts
Most of our breadth and liquidity indices are updated end-of-day with the exception of
These 3 models above are updated every 15-min during the course of a trading day.
PRO subscription clients will from now on also be able to see the following new models updated every 15-minutes:
- New improved Short-Term liquidity index (STL2)
- New improved Medium-Term liquidity index (MTL2)
- New improved Average Liquidity index (ALIX2)
During the remainder of 2019 PRO clients will also see the inclusion of the following on an intra-day basis:
- Percent of shares above 50dma (MA50%)
- McClellan Breadth Family (MCOS)
- HiLo Breadth Indexes (HILO)
- Net New Highs & XOVER Diffusion (NEWHI)
- SP-500 Trendex
The intra-day charts will be available from the location depicted below:
Dramatic change in yield curve
The RecessionALERT yield-curve aggregate and diffusion has just made a dramatic reversal, with the percentage of 10 term-spreads that are inverted dropping from 70% to 40%:
The 10YR less the 2YR narrowly averted an inversion 8 weeks ago and the 10YR-5YR and 10YR-3YR never came close to inversion. If one looks at the latest history and with the benefit of hindsight, we can see that the yield spread aggregate and its 10 components seems to be on a trend upwards. The 10YR-1YR has “un-inverted” 2 weeks ago and this week the 5YR-3YR and 2YR-1YR managed to raise their heads above water. All the other components are threatening an inversion reversal.
For quite some time now we have been warning that the yield inversion we were witnessing may be a false positive as witnessed in late 1998 or a premature inversion as witnessed in 2006. Not only was the depth and duration of the latest inversion weak, but the diffusion representing the percentage of underlying term spreads that have inverted failed to stretch beyond 70%:
But let us assume we are wrong – and recession does occur as predicted by the yield curve inversion. Due to the mild depth, duration and completeness of the inversion, coupled with the inversion resulting from longer rates falling faster than shorter rates as opposed to shorter rates rising faster than longer rates as is the historical norm, we are still inclined to assume that it is likely to be a mild recession.
Again we would like to point out that the presence of many false positives on daily and weekly yield curve prints coupled with very long leads to recession and wide standard deviations on the average lead to recession can make basing market related actions on the yield curve less useful. At best the yield curve is a background signal. Below is a selection of RecessionALERT models with the yield curve characteristics very similar to the US-LONG category:
As prior research of ours has shown, getting out the stock market more than 5-6 months prior to recession is very counter-productive.
With the above in mind, its best to follow a battery of short-leading (20-30 weeks) indicators with very few or zero false positives and low standards of deviation in their lead times. The SuperIndex RFE-6 that is displayed in every weeks’ SuperIndex PDF report fits that bill perfectly. Adding the US Monthly Leading Index (USMLEI) to the RFE-6 ensemble, gives you the RFE-7 which also provides for zero false positives historically for signals greater than 1:
The ability to hard-wire RFE-5 (trigger =0), RFE-6 (trigger =1) and RFE-7 (trigger=1) into the monthly data-file will be made available from next months report in the OVER-RIDES input section as shown below:
We have also added to the user selection section the average recession lead, standard deviation, false positive count and coefficient of variance (CV) statistics for every one of the 20 or so models we track. Ignoring false positives, if one takes the coefficient of variance as the main performance metric (lower=better) then here are the 5 most/least powerful RecessionALERT models:
The models in green are the most powerful ones when one measures the standard deviation (variance) of average historical lead times to recession as a percentage of the average lead to recession. RFE-6 and RFE-7 have zero false positives (triggers > 1), low CV’s and fit into that golden window of 20-28 weeks lead to recession (even accounting for reporting lags).
Are trade war concerns valid?
It appears U.S investors’ concerns with global trade wars are dominating U.S stock market direction for the last two years:
This is with valid reason, as prior research of ours (Global Economy affects U.S stock market returns) has pointed out that whilst a global recession does not necessarily result in a U.S recession, it can certainly lead to one if the U.S economy is vulnerable. Additionally that research pointed out much bigger than expected correlations between U.S stock market returns and the state of the leading data for the rest of the world.
As far as we are concerned, a global business cycle downturn has been underway since December 2017 and it is certainly being accompanied by a global-trade (imports+exports) recession as shown by one section of our Global Economic Report:
But let us get back to what keeps us awake at night. The global LEI’s are struggling to complete a rebound that commenced in late 2018, and this is occurring against the backdrop of a very vulnerable RecessionALERT US leading indicator.
Failure for a global recovery in the country LEI’s will almost certainly drag the US LEI underwater if past decades of history are anything to go by. If the U.S and China can reach some kind of compromise before the year is out, then the global LEIs are likely to pick up and the US stock market is almost certain to rally to new highs, despite stretched valuations.
About the new TRENDEX SP-500 model
There have been numerous queries about the new TRENDEX chart in the PRO Charts section.
This model supersedes the Demark and the Demark+ trend counting models as it is a far superior methodology focusing on support and resistance levels as opposed to closes 4 days ago and moving averages.
The methodology was a by-product of the research done on the SP500 Probability Model when initially using DEM+ as one of the six model components and realizing there was a better way to track downtrends more persistently without whipsaws. The superiority on the downtrends applied to the up-trends as well, meaning we could issue probabilities of market tops as well as market bottoms.
- STOP = cancel level for current trade (arrow depicts movement from prior day)
- p(n) = probability of last day depicted on chart
- p(n+1) = probability tomorrow if we don’t cross the black line
Apart from using support and resistance in counting trend duration as opposed to arbitrary closes 4 days ago and moving averages subject to whipsaws, the TRENDEX also differs from DEM and DEM+ in that we dispense with standard daily counts of a trend and dynamically map these to actual probabilities these counts represent from the historical record of success of a trend reversal. This is far more intuitive than daily counts especially if it means the historical record is changing all the time with new daily data.
We fully expect to retire DEM and DEM+ in their entirety with the TRENDEX once it has been fully documented and potentially refined further in months ahead. We are scheduling a detailed research note on the TRENDEX and we will maintain the PRO-Chart daily until then.
Note that currently the SP500 Trendex is a component in the multi-factor Market TOP probability model found in the “SP500 TOP” tab. We have yet to roll out multi-factor peak-detection models for the various other ETF we maintain probability models for.
Yield Curve inversion suggests mild recession
There has been acute interest in the inversions currently taking place on the term-spreads around the world:
And this comes as no surprise, since more than half of the world’s sovereign yield curves have now inverted…
Right now, 70% of the U.S yield-curve cluster comprising the 10/5/3/2/1 year bond yields are inverted as shown below. In prior research we have advocated using greater than the 60% threshold to have a “guaranteed lock-in” of the inversion:
A closer inspection shows that the 10-year less the 2-year is on the cusp of inverting, which will mean 80% of the yield curve complex will have inverted when this happens:
It is interesting to note that the current inversion process is very different from the past. In the past we have inflation and short rates rising as the economy overheated whereas now we are seeing both the long and short rates falling rapidly. The past trend of the last 7 inversions had the short term rates rising faster than the long term rates. This time around, the long term rates are collapsing faster than the short-term rates:
We do not think this changes any of the historical inversion connotations. The distortions and behaviors driven by short rates being higher than long rates still applies in our view. However, this difference in itself is more accommodating to the economy since rates are falling (as opposed to rising in the past). Even if the US economy falls into a recession, it’s likely to be less severe than recessions in the past.
There are a lot of theories on why a yield curve inversion causes a recession. This one is our favorite purely because of its simplicity:
As we have stated before, there is not much use in knowing a recession is coming in 12-24 months time since the post-inversion stock market peaks and variances in the leads to recession are too widely dispersed (wide standard deviations) to be of practical use. However, for those that are curious, here are the leads to recession after a 10’s vs. 1’s yield curve inversion:
The two smallest lead-times to recession average 8 months, the median lead-time is 12 months and the two longest lead-times average 22 months. This puts forward the following dates for recession:
We likely avoided a full yield-curve inversion
In our early June post “Is the U.S Yield Curve Inversion locked in?” we mused that only portions of the term-spread complex had inverted and most likely would remain that way, allowing us to avoid a full scale term-spread inversion. As the chart below shows, this is indeed the case – with only 50% of the term-spread complex having inverted as of 26 July 2019 and a maximum of only 60% of the term-spreads having inverted so far:
Literally all 10 of the term-spreads seem to be rising now. Furthermore, the average of all 10 term-spreads, our preferred term-spread metric, has inverted only briefly for a 1-week and 2 week spell and also seems to be trending upwards away from inversion.
Our message remains that focusing on (1) the percentage of term-spreads that have inverted together with (2) the average yield of all term spreads together, is a more robust mechanism of evaluating the term spread structure for recession risk. In this respect, the following two conditions need to both be present before proclaiming a full-blown yield curve inversion:
- The average yield of all 10 term-spreads need to be negative for at least 4 consecutive weeks
- At least 7 of the 10 term-spreads need to be inverted (70%)
The historical deployment of these conditions can be examined on the chart below:
It is not outside the realms of possibility that we are witnessing a similar false positive to that of 1998, and dare we say it, but perhaps the U.S economy is in for a soft landing? If this is indeed the case then all we have to be really concerned about is the current dangerously high (subscription required to open) stock market valuations.
It is going to be interesting to see the reaction of all these term spreads to a looming Fed rate-cut. Conventional wisdom is that they will all move even further away from inversion.
RAVI Warning issued
This subscriber-only client alert has now been unlocked for public viewing.
NOTE : The RecessionALERT Valuation Index (RAVI) now warrants its own, more detailed, dedicated PDF report which you can now find in the REPORTS>RAVI menu tab:
Well, it has finally happened, we have a recession and stock market bear warning from just about every RAVI indicator as at 1Q2019. All of the 8Q average of the 2YR forecast, the 4Q average of the 1YR forecast and the average annualized 1/2/3 YR forecast are signalling negative annual returns for the stock market, and given that 2Q2019 stock market returns were positive, this is likely to make the 2Q RAVI report we will see in the 1st week of September 2019 even worse:
Here are the prior two occasions these indicators signaled negative stock market returns:
We like to treat the RAVI signals as long-leading indicators with the 8Q average of the 2YR forecast giving 4 quarters warning to the last two recessions (3 quarters real-time warning when accounting for the 1-quarter lag). The warning to stock market peak is far less forgiving though, and by the time you see the 8Q average signal the stock market has likely peaked or about to peak in the next quarter, assuming the model and paradigms it is premised on, hold in this business cycle.
We have found it rather instructive to examine the behavior of the average of the 1-year forecast and the 2 and 3-year forecasts annualized. In fact, whilst the 1-year forecast has an r-squared of 0.352 with future annual stock market returns:
…the average of the 1-year forecast, the 2-year forecast annualized and the 3-year forecast annualized has an incredible r-squared of 0.46 to 1-year ahead SP500 total returns:
On a one-year look-ahead basis, this is an incredible result. Here is the average annualized forecast together with the US Long Leading Index taken from the monthly USMLEI report, with lead times to recession depicted in quarters (subtract 1 for real-time lag):
We must point out at this juncture that using RAVI forecasts to warn of recession does not work every time. It only works when forecast negative returns for the stock market are of significant magnitude and persistent enough – i.e when stock market valuations have gotten way ahead of themselves. To wit, the prior 3 recessions before those shown above were not accompanied by significant enough stock market declines to render the RAVI forecasts of any use. To this end, the RAVI remains foremost a stock market returns predictor as opposed to an economic recession predictor.
We see from the above chart that the average lead of US-LONG and the RAVI annualized average is around 7-8 quarters (6-7 quarters real-time) and given that both have been inverted now for an average 3.5 quarters, the implication is that the business cycle peak will be in 3.5 to 4.5 (lets say 4) quarters time from 31 March 2019, or 31 March 2020.
Another benefit of using the RAVI annualized average is that it gives us more warning to the likely stock market peak, namely 4 quarters which implies the top is already in, or will be shortly.
We wish to leave you with one final thought. Below is the 10-year forecast from RAVI. We are on the exact 10-year anniversary of the March 2009 stock-market bottom that birthed the current bull market. Ten years ago, the RAVI forecast we would see 361% returns till 1Q2019. We actually achieved 338% which is remarkable:
In conclusion we have a major shot across the bows that valuations have got ahead of themselves and that odds of benign future stock market returns are low. This is not to say sentiment doesn’t drive the markets higher, it just says headwinds are becoming non-trivial and that volatility will most likely increase significantly. With future short-run stock market returns predicted to be negative according to the model, it also means the RAVI joins the other long-leading signals that have turned bearish on future economic outcomes.
Is the U.S Yield Curve Inversion locked in?
In our February 2019 commentary we forecast that the 10’s and 1’s yield-curve would invert in May. The data below is taken from that market commentary, with a warning that the indicated dates of recession have a very wide standard deviation over the historical record:
The 10’s and 1’s term spread has been inverted since 23rd May, for six sessions now. The question that naturally comes to mind is “how long must the daily yield curve be inverted for us to have a “latch“?”
The chart below shows daily 10’s less 1’s yield curve prints since 1962:
Apart from brief 1-day inversions on 21/09/1998 and 05/10/1998 there have been no inversion spells that have not ultimately led to full-blown sustained inversion very shortly thereafter. Our preferred granularity is on the weekly level, which at six days inversion has now been met.
Assuming prior daily history holds, one would be inclined to assume the 10’s and 1’s yield-curve will remain inverted or if not, will remain inverted in the very near future.
However, in a past commentary we observed that once more than 60% of the treasury yield-curve spectrum inverts, the remainder invariably all invert in very short order. It is probably more useful to examine the entire treasury term-spread spectrum to try and assess where we are:
At this point in time we are bouncing on that 60% level with just the 10’s and 2’s, the 10’s and 3’s, the 10’s and 5’s and the 5’s and 3’s (the traditional inversion laggards) above water:
And herein lies a clue as to why this time things may be different – the ones above water and some of those already below water are actually widening their spreads! This could quite well shape up to be a false positive as witnessed in 1999 when only 60% of the treasury spectrum inverted.
The longer the 10’s and 1’s (and all the others that currently are inverted) remain below zero, the higher the odds of a sustained inversion but for now we are still 40% short of a full spectrum treasury term-spread inversion.
There is of course one more question we need to start answering and that is assuming the yield curve remains inverted or does so in short order, will this lead to recession? There are two camps at the moment:
- Those that ridicule the “this time is different” arguments, noting that every time in the past things we proclaimed different they were anything but;
- Those that argue that the negative real interest rate policy of the Fed and other forces causing today’s inverted yield curve are entirely different from those in the past.
We have seen some pretty compelling arguments from the “this time is different” yield-curve camp and of course there have been prior inversions that have NOT led to recession (see 1966) and we will therefore be observing many other long-leading indicators less influenced by the Fed and artificially low interest rates to come to our determinations. Any recovery in the residential housing market index (USHMI) should be watched as well as the current situation of all the other long leading indicators in the Monthly Leading Economic Index (USMLI).
Conclusion : A definite yellow flag for the treasury yield curve spectrum is up, but given only a 60% inversion, we do not think the red flag for the full yield curve spectrum is up yet.
Unemployment is worse than it looks
The U.S civilian unemployment rate reached new lows of 3.6% in April – numbers last seen 51 years ago in 1968:
There are a number of ways to use the national unemployment rate to signal recession, but almost all of them are co-incident to slightly lagging in the warning they provide.
We have the most commonly used method which is annual growth of the unemployment rate, which has provided about six false positives since 1950 and can lag on occasion due to the long time frame it uses – although it provided timely warnings in the last 3 recessions:
When national unemployment rises more than 0.5% in an economic expansion, it usually signals the start of a U.S recession without any pesky false positives. If we assume 3.6% was the low then this implies a unemployment rate greater than 3.6%+0.5% = 4.1% will be our warning signal:
We prefer to examine the state-level unemployment data which tells us about what is going on “under the hood” of the national unemployment rate. The three metrics we prefer to look at all provide earlier warning signs than the traditional methods and are covered in our monthly Labor Report for subscribers. Here is last months’ reading:
We notice that all three metrics are showing a lot more underlying deterioration and loss of momentum than is currently observable from the national unemployment rate itself. In fact, since March 2014 these metrics have all been climbing steadily, as they did in the prior 5 recessions.
It’s unfortunate that accurate state-level unemployment data is only available going back to 1978 to provide longer term history but this does not detract from the usefulness of these indicators to get a more sensitive reading into what is happening with unemployment.
Despite the visible deterioration in the state-level data it is not panic time yet. More states have to have rising unemployment than those with falling unemployment for the one metric (green line) and both the other two metrics need to stay above 50% before we start hitting the panic buttons.