The past few weeks have brought not just strong market volatility, but lots of troubling economic reports that have many fearing a recession in 2019.
While predicting economic downturns is far from an exact science, there are some time-tested approaches that offer useful insights.
The best recession predictor ever discovered is currently predicting a recession is likely to begin in 16 to 23 months (most likely in 20 months).
While bear markets are even harder to pin down, the average market top averages 8 months before a recession, meaning the next big decline is likely to start in 4 to 21 months (most likely 12 months).
These are all merely probabilistic forecasts based on ever changing data and investors should always focus on maintaining the right asset allocation to meet their long-term goals no matter what the economy or market is doing.
Note that due to reader requests, I’ve decided to break up my weekly portfolio updates into three parts: commentary, economic update, watch lists/best stocks to buy now. This is to avoid excessively long articles and maximize the utility to my readers.
This week’s commentary points out the three easiest ways to make money in the stock market.
Note that I offer these weekly economic updates purely because I believe that investors should always take a holistic “big picture view” of the world. That means knowing the state of the economy and what the short- and medium-term recession risks likely are. However, as I’ll explain later in this article (recession risk section), macroeconomic analysis has historically proven to be a terrible tool for stock market timing (SPY) (DIA) (QQQ). Which is why I only offer these analyses so that readers will likely be able to see a recession coming about a year or so away.
That will hopefully allow you the time to prepare yourself emotionally and financially for the downturn. It will also hopefully allow you to adjust your portfolio’s capital allocation to a more defensive stance, such as with defensive sectors, or potentially greater allocation to bonds (for lower risk-tolerant investors).
How I Use Probabilistic Models To Estimate When The Next Recession And Bear Market Will Begin
Ask five economists/analysts when the next recession is likely to start and you’re likely to get 10 answers. Ask when the next bear market is likely to begin (stocks top and begin falling) and you’re likely to get even more responses and wide ranges of dates.
That’s because the economy is massive, complex and the stock market is driven in the short term by fickle investor sentiment that swings from euphoria to absolute panic.
However, while nailing down the precise start date of the next recession and bear market is impossible there are still some time-tested probabilistic models we can use. These tell us the most likely (though far from guaranteed) start date of both economic and market downturns based on the latest macroeconomic data and trends.
Since 1929, 55% of bear markets have come during recessions, and since recessionary bear markets are typically the most severe, this is where we want to start building our model.
According to a March 2018 study from the San Francisco Federal Reserve, the most accurate recession predictor in history is the yield curve. And according to a follow-up August 2018 study, the most accurate yield-curve of all is the 10y-3m.
Banks were also asked how their lending policies would potentially change in response to a hypothetical moderate inversion of the (10y-3m) yield curve prevailing over the next year. Significant shares of banks indicated that they would tighten their standards or price terms across every major loan category if the yield curve were to invert…Those banks that indicated they would tighten their lending policies because of an inversion of the yield curve were asked to provide reasons for their response. Major shares of banks indicated that their bank would interpret this scenario as signaling a less favorable or more uncertain economic outlook and as likely being followed by a deterioration in the quality of their existing loan portfolio. In addition, major shares of banks reported lending would become less profitable and their bank’s risk tolerance would decrease in this scenario. ” – Dallas Federal Reserve Bank Survey (emphasis added)
The reason for that is because the 10y-3m curve is the one that, according to a Dallas Fed banking survey, is what banks are using to predict recessions and determine lending policy. In other words, there is a direct link between this yield curve and the actual reduction in credit that actually causes recessions. Ultimately, the 10y-3m curve can be thought of as a self-fulfilling prophecy.
If banks think that an inverted (negative) 10y-3m curve means a recession is looming, then they will hunker down and reduce credit to consumers and businesses and thus trigger the very economic downturn they were expecting.
Thus we can use the historical trend line of the “banker’s yield curve” to estimate when the next recession is most likely to start (confirmed by other data sources).
What about the next bear market? Well, that’s far harder to determine since non-recessionary (cyclical) bear markets are possible. In fact, since WWII, the stock market has predicted 13 of the past 7 recessions.
The average warning time is 253 days or 8.4 months. As measured by the Dow, the average bear market over the past 80 years begins with a 9% decline in the first 3 months (not the sudden plunge we saw this time).
Of course, averages are only a rough guide (all downturns are different) and the range of lead times from market tops and recessions is also very wide. It ranges from 53 days (2007) to 367 days (1969).
The best we can do is estimate when a recession is likely to start (based on 10y-3m trend and inversion) and then assume that the market will top and begin sliding (most likely relatively slowly) two to 12 months before then.
For example, say the 10y-3m curve inverted on Monday, January 7th. Then our historical analysis would estimate the next recession begins 9 to 16 months later (September 2019 to April 2020, most likely February 2020). That would then mean a bear market is most likely to begin between September 2018 and February 2020 (most likely June 2019). The range on the most likely market top is 17 months or about 1.5 years. That kind of wide range makes traditional market timing (going 100% to cash) all but impossible.
However, even such wide range probabilistic estimates can be useful for capital allocation purposes because they allow investors to either adjust their mix of cash/bonds/stocks as well as shift into more defensive stock allocations (for that part of your portfolio).
So what does the latest data and trends indicate in terms of the most likely start dates for the next recession and bear market?
When The Next Recession And Bear Market Are Most Likely To Begin
Since the end of the Fed’s Quantitative Easing program the “banker’s yield curve” has been steadily falling. While at times it has risen above its long-term trend line, as you can see it tends to mean revert to the trend and sometimes with alarming speed. Today, this curve stands at 25 basis points.
While it’s possible that the curve could invert very soon (due to worse-than-expected economic reports and crashing inflation expectations) for my model I assume the long-term trend will hold. That would mean the most likely 10y-3m inversion is 7.4 months away (mid-July 2019).
Since the 10y-3m month inversion gives a historical lead time of 9 to 16 months (average 13 months), that puts the next recession start date range at mid-April 2020 to mid-November 2020 (most likely mid-August 2020). Note that this range now precludes the earlier estimate of 2020 or 2021.
The next bear market start date (market top) estimates thus range from mid-April 2019 to mid-June 2020 (most likely mid-January 2020).
But of course, since these are merely probabilistic forecasts of what’s likely to happen, we need to carefully track the actual macroeconomic data (the facts) rather than obsess over forecasts that are constantly changing as trends change over time.
Current Economic Growth
- Q3: 3.4% (final estimate)
- Q4: About 2.5% to 2.6%
- Full Year 2018: 2.9% to 3.0%
- 2019: 2.1% to 2.7%
(Source: Atlanta Federal Reserve)
Every major GDP model uses slightly different combinations and weightings on leading indicators to estimate the current growth rate of the economy. Thus, the actual weekly figure is far less important than the trend of the estimate.
The Atlanta Fed’s model is the most volatile one I track but recently it has stabilized right at the consensus economist estimate of 2.6% for Q4’s growth rate.
(Source: New York Federal Reserve)
The New York Fed’s more stable model estimates similar 2.5% GDP growth in Q4 and just 2.1% growth for Q1. That’s below the full-year GDP growth estimates ranging from the Fed’s 2.3% to the IMF’s 2.7%. Due to the shutdown now being in its third week and reducing Q1’s GDP by about 0.05% per week, I think that overall 2019 growth might come in as low as 2.1% as well, if current negative trends continue.
Nowcasting has recently had the most bullish GDP growth estimates but is also showing a trend of rapidly falling expectations.
Once per month, James Picerno at Capital Spectator compiles all the current GDP Nowcast estimates. The median consensus is for 2.7% growth in Q4.
Basically, the combination of these forecast models is confirming what all the other data (and the bond market and recession models) are telling us. We’re headed for far slower growth in 2019 which will put a lid on inflation and give the Fed plenty of room to end its tightening policy.
In fact, the bond futures market is currently pricing in zero rate hikes in 2019 and one cut in 2020. And the probability of a rate cut in 2019 has increased to 25.4%.
Recession Risk: Very Low
- The probability that we’re in a recession right now: 0.8%
- The probability of a recession starting in the next three months: 3.11%
- The probability of a recession starting in the next nine months: 22%
I use eight key meta-analyses to track the health of the economy. That includes those which have historically proven to be good predictors of recessions:
- The 10y-2y yield curve;
- The 10y-3m yield curve (most accurate)
- The Base Line and Rate of Change or BaR economic graph;
- Jeff Miller’s meta-analysis of leading economic indicators;
- The St. Louis Fed’s smoothed-out recession risk indicator; and
- The New York, Atlanta Fed’s and now-casting.com‘s real-time GDP growth trackers.
(Source: Business Insider)
The yield curve has proven the single most accurate predictor of recessions over the past 80 years. Specifically, when the curve inverts or goes below 0 (because short-term rates rise above long-term rates), then a recession becomes highly likely. It usually begins within 12-18 months.
|Yield Curve Inversion Date||Recession Start Date||Months To Recession Once Curve Inverts|
|August 1978||January 1980||17|
|September 1980||July 1981||10|
|December 1988||July 1990||19|
|February 2000||March 2001||13|
|December 2005||December 2007||24|
(Source: St. Louis Federal Reserve, Ben Carlson)
According to a March 2018 report from the San Francisco Fed, an inverted yield curve has “correctly signaled all nine recessions since 1955 and had only one false positive, in the mid-1960s, when an inversion was followed by an economic slowdown but not an official recession.”
In other words, if the yield curve goes negative, there is probably a 90% chance of a recession starting within the next 17 months or so.
Unfortunately, investors hoping to use the yield curve to time market tops are out of luck. While a yield curve inversion is very accurate at predicting recessions with long lead times, its track record on predicting bear markets is far less impressive.
10y-2y Yield Curve Inversion Vs. Bear Market Starts
(Source: Wealth Of Common Sense)
The lag time between market tops and yield curve inversions is all over the map, ranging from just 2 months in 2000 to nearly 2 years in 2005.
And if we go back to 1956 (using the 10y-1y yield curve), we can also see that yield curve inversions are largely useless for timing bear market starts. In fact, on three occasions, the forward-looking market has actually peaked before the curve inverted. This means that the yield curve should not be used as a market timing mechanism but rather purely as a good recession risk indicator.
Current 10y-2y Yield Curve: 0.17% (up from 0.17% three weeks ago)
On December 3rd, the 5y-3y warning curve (its inversion has heralded the 10y-2y inversion by a few months for 40 years) inverted but on December 14th, it uninverted (currently 0.01%). This shows why it’s important not to panic about a warning curve inversion because yield curves can go up as well as down.
But more important than the 10y-2y curve (the most popularly followed one in the media) is the 10y-3m curve.
(Source: San Francisco Federal Reserve)
According to an August 2018 study by the San Fran Fed, “The best summary measure is the spread between the ten-year and three-month yields.”
10y-3m yield curve: 0.25% (down from 0.47% three weeks ago)
According to a Dallas Fed bank survey, banks are closely watching the 10y-3m curve and plan to pull back on lending (and thus cause the recession they fear) when this curve inverts. Fortunately, that curve, the most accurate recession predictor of all, has been flattening at a highly predictable rate.
But it’s important to remember that you shouldn’t fear a flat yield curve as a sign of poor short- to medium-term stock performance.
During the strongest bond market in US history (tech boom), the yield curve was as low or even lower than it is now. Of course, that was also an epic bubble, but the point is that a flat, but positive yield curve is not a sign of poor returns ahead.
Average Monthly Stock Market Returns By 10y-2y Yield Curve Slope (Since 1976)
In fact, over the past 42 years, the period when monthly stock returns were at their highest and volatility was at its lowest was when the yield curve was flat but positive. This means that we’re likely in the sweet spot right now, and investors should avoid using fears of yield curve inversion as a reason for market timing.
That’s because even after an inversion occurs, stocks tend to continue rising for quite some time and tend to generate strong returns before the next bear market begins.
Basically, the yield curve is a totally binary indicator.
- positive = very low recession risk (carry on with long-term investing plans)
- negative = 90% chance recession is coming within 6 to 24 months (most likely 18 months) – consider getting more defensive
The second economic indicator I watch is David Rice’s (aka Economic PI) baseline and rate of change, or BaR economic analysis grid. This is another meta-analysis incorporating 19 leading indicators that track every aspect of the US economy. That includes the yield curve, though a different version of it. I consider it the best overall indicator of fundamental economic health (because it’s so granular).
(Source: Economic PI)
The BaR grid has shown to be a reliable indicator, predicting the 1980, 1990, 2001, and 2007 recessions.
(Source: David Rice)
Currently, 10 out of 19 economic indicators are pointing to positive economic growth with eight indicating negative growth. That number of negative indicators is up from three a month ago.
(Source: David Rice)
Note that over the past 35 weeks, the number of leading indicators in the decline quadrant has ranged from three to 10. There is a lot of volatility between the number of indicators showing decelerating or accelerating growth. This is just statistical noise, and only long-term trends should be used as recession risk warning signs. However, the steady climb in negative indicators and the MOC shifting into the decline quadrant (rate of change now negative) is not a good sign.
(Source: David Rice)
The mean of coordinates or MOC continues to remain high, though it is now down from 35.3% three weeks ago. The average of the leading indicators is now below the MoC though indicating that growth has likely peaked and will be trending lower (confirming what other models are expecting and forecasting).
(Source: David Rice)
As you can see, the MOC has at times been in the decline quadrant most notably in 2016 when the oil crash significantly slowed US GDP growth (US is now the world’s largest oil producer, so very low oil prices are not good for our economy).
According to Mr. Rice, the peaking of the MoC usually happens about two years from the start of a recession, which backs up what the yield curve and other models are saying.
Next, there’s Jeff Miller’s excellent economic indicator snapshot, a rich source of numerous useful market/economic data. It also provides an actual percentage probability estimate for how likely a recession is to start in the next few months.
(Source: Jeff Miller)
What I’m looking at here is the quantitative estimates of short-term recession risks. In this case, the four-month recession risk is about 3.11%, while the probability of a recession starting within nine months is 22%. The short-term recession risk is highly volatile, ranging from 0.24% to 3.32% since I began tracking the economy over that past nine months. Thus, the more important thing to focus on isn’t the absolute figure but the trends in both short-term and medium-term recession risks.
Both have shown low risks, with the 9-month recession risk being highly stable at 24% all year, and now falling to 22%. Inflation expectations are falling in line with slower growth expectations giving credence to the idea that the Fed is done hiking this cycle. Meanwhile, stocks are cheap with a historically high equity risk premium (forward earnings yield minus 10-year yield). So if you have spare cash to put to work in the equity portion of your portfolio, now is likely a good time to do so.
For a final look at recession risk, I like to use the St. Louis Fed’s smoothed-out recession risk indicator. This looks at the risk of a recession beginning in the current month (it’s actually delayed two months). It uses a four-month running average of leading economic indicators.
(Source: St. Louis Federal Reserve)
Since 1967, this smoothed-out recession probability estimator has predicted five of the last seven recessions before they have started. The key is that as long as the recession risk is at 3.9% or below, the economy is very unlikely to be in a recession. At 0.8% risk right now, this confirms that the US economy is likely to keep expanding for the foreseeable future.
Bottom Line: The Next Recession And Bear Market Are Getting Closer But Not Yet Around The Corner
Again, I’m not a market timer, just a macroeconomics nerd (my major in college) who wants to ensure I and my readers see the big picture. Any estimates of when recessions and bear markets are likely to begin are purely based on historical averages and the most time-tested models we have. They are probabilistic and not definition predictions that should be used for market-timing purposes.
Basically, these weekly economic updates are not meant for market-timing purposes, but rather to allow you to prepare yourself emotionally and financially for when a recession does inevitably happen. It’s also meant to give you around a year’s warning (hopefully longer) to adapt your portfolio’s capital allocation strategy.
That might mean:
- Stockpiling some cash (to take advantage of future bargains during a bear market);
- Putting new capital to work in more defensive companies (utilities, healthcare, telecom, consumer staples); or
- For the most risk-averse investors, potentially moving some money into bonds or cash equivalents (asset allocation changes).
Personally, I’m now in recession prep mode, which means I’m focused on paying off all margin ($126,000 worth) ahead of the next bear market, which is still a long way off (but it will take me about 12 months to fully deleverage which is why I’m starting now). Note that I don’t plan to ever use margin again (it was my biggest investing mistake of 2018) and recommend that most people don’t as well.
Here are my best estimates for when the next economic and major market downturn are likely to begin.
- The most likely recession start date: mid-April 2020 to mid-November 2020 (most likely mid-August 2020)
- The most likely next bear market start date: mid-April 2019 to mid-June 2020 (most likely mid-January 2020).
I’m not actually being defensive since I’m still 100% in stocks and have no plans to sell before the next market downturn starts. Nor do I recommend most investors switch from their current long-term investment plans, which should already have you in the right asset allocation that best meets your personal needs and risk profile.
Rest assured that if the macro trends (both economic and earnings) shift and point to a recession (and bear market) starting earlier than you will read about it here.