Thoughts On Business Cycles, Recessions, and Financial Markets
As as student of economics, I am eminently fascinated with how such disastrous occurrences such as recessions cannot reliably be spotted in real-time. How can some of the smartest, most well compensated individuals in the world analyzing reams of data fail to spot economic weakness as it occurs? It is an amazing question.
Furthermore, how can I reconcile the fact that I believe short term returns are completely unpredictable (a random walk), while factoring in the work of Shiller et al that demonstrate that the overwhelming predictor of long term returns is starting valuation?
The implications are obviously profound, as late cycle, the current valuations suggest that the next recession is likely to be rather nasty with regards to equity prices.
Over the past several years I have thought about this, and have looked to the data, particularly the 2001 recession. I have come to have a bit of an epiphany relating to the nature of economics and financial markets in general which answers some of the longstanding questions that I have personally held for many years about the nature of the business cycle and how to position a portfolio, and as I find organizing these thoughts by writing them to be somewhat revealing in and of itself, thought I would share it on this blog.
To me, it’s a bit of a unifying theory that marries together a lot of my unsolved economic theories.
These thoughts may turn out to be completely incorrect, so I’ll mention the disclaimer that none of this is investment advice nor do I recommend any course of action for anyone else. You are free to take these thoughts as worth exactly what you paid for them: nothing.
I will break this into four separate posts with the following topics:
- Economic data during the 2001 recession
- On the nature of recessions
- My exit point: the yield curve
- My entry point: the NBER recession announcement
Economic data during the 2001 recession
Particularly for the 2001 recession, it appears to me that much of the weakness in the economic data occurred simultaneously with the drop in the stock market. The S&P sets a peak in October 2000, and by December’s low, it is already down about 15%. The trouble with most economic data sets is that they are not viewed contemporaneously: there is a delay to the data itself, the data is subject to revision after the fact, and downtrends are seen even in healthy ‘correction’ type events. I will try to summarize the key question:
If you needed to be able to sell the October top to avoid a 4Q 15% draw down, what could you look at that would differentiate something in the August or September of 2000 that was materially different than from what you might have observed in 1996 or in 2011? The latter 2 instances were situations in which the ‘typical’ measurements of economic strength would yield false alarms.
Frequently cited indicators are initial unemployment claims, the Conference Board Leading Index Indicators, and the Chicago Fed National Labor conditions. However, to me, each of this appears somewhat misleading. The CB LEI numbers themselves have more than a one month delay to them, so the latest observable value in October 2000 would have been the September 2000 number. This is how the data looked in the Fall of 2000 and in the Fall of 1995 respectively:
This is data as you would have seen it in December of 2000, by which point the market was already down 15%. Materially, the leading indicators are not materially worse than at other times.
Nor is the Chicago Fed National Labor conditions number, which reported the first parts of 1995 as weaker than the first parts of 2000 (on average).
And you would have seen a material increase in unemployment claims in 1995 as well.
Furthermore, an indicator based on the Chemical Activity Barometer shows some weakness, but nothing materially different than 2011, which was another year where the Chicago Fed said the labor market was worse than 2000. Again, by the time you are viewing this data, it is too late to miss the first 15% of damage to the portfolio. It was only with the benefit of an additional 3-4 months where it became evident this was no routine softspot but a full fledged recession:
It was only with the benefit of an additional 3-4 months where it became evident this was no routine softspot but a full fledged recession:
So, how does one differentiate between a 2000 and a 2011? How do you identify the false positives in the system? Can you? I have concluded, as I describe in the next post, that to a large extent you cannot.
On the Nature Of Recessions
- I believe the economy is an extremely dynamic and interrelated system. An economy is balancing millions of different markets, each with their own fundamental characteristics, their own lead and lag times to ramp up or ramp down production, and adapting and evolving at different rates. And each responding to incentives and stimulus to tax policy, financial markets, and geopolitics in their own ways at their own rates.
- I believe it is therefore more appropriate to view a recession as a stochastic event. That is to say, at any point in time, the economy has a certain percentage of chance of crashing into the mountains, so to speak. Risk factors that might increase the risk of a recession would be bad debt loads, overvalued assets, overbuilt capacity, geopolitical uncertainty, increasing commodity prices, etc.
- I believe the economy is at its heart, a system run and controlled by humans. Corporations are run by humans, as are stock markets, capital markets, mortgage decisions, and investment banks. And humans are, of course, subject to basic emotions of greed, fear, and a particularly nasty herd mentality.
- Accordingly recessions are unavoidable. At a micro scale you see this in every commodity sector. The boom/bust nature of oil, shipping, gas, mining. The price rises, people get overly excited and leverage up, and build out too much capacity. Supply crushes demand, firms go insolvent. The cycle continues.
- I believe capital markets are too large of a part of life for individuals, companies, and governments to ignore. A CEO who has just watched his share price fall 30% will behave differently than if his share price had risen slightly. A family of 4 who has just had their wealth cut in half will spend differently than if they had not.
I therefore have come to believe the start date of a recession is not something decreed a priori on stone tablets from Mount Sinai but is something that is itself endogenous in the system.
- If we accept that interest rates have an impact on the real economy and also have an impact on financial markets, it therefore becomes clear that financial markets and the economy are themselves correlated bidirectionally.
- I have begun to think that perhaps 2001 was like World War I. By the 1910’s, a systemic European-wide conflagration was pretty much inevitable as the arms races, the clashes of colonialism, and the German challenge to the British hegemony assured that sooner or later war would come to pass. But would it be something in Africa in 1912 that lit the spark, or something in the Middle East in 1916, or as history would have it something in the Balkans in 1914?
- By 2000, for example, the excesses had built themselves up to such an extent that recession was inevitable. But why did it have to be March 2001? Why not June 2001 or January 2001, or the summer of 2000? All of the pieces were pretty much in place throughout the late 90s.
- Indeed, the yield curve first inverted in June 1998 before recovering to invert again in February 2000, about 1 year before the recession. In December 2005, it inverts more than 2 full years before the recession. In 1988, there was nearly 2 full years before the 1990 recession. Why in 2000 couldn’t conditions have lasted an additional 1 year in line with the other recessions?
To me, therefore, the answer to the question of why does the stock market peak 6 months before a recession is not because the recession can be predicted but rather because the cause of both the stock market drop and the subsequent recession is a loss of confidence in the system and capitulation by the herd.
If the underlying decay of the economy has reached a point where any disturbance to it, be it a bank failure, some pronounced layoff, a Fed rate hike, or merely a 10 percent equity correction like the one we just had sets off the panic, then I think it plausible this not only impacts stocks but cascades into a full blown recession. A loss of confidence starts to show as projects are delayed and workers are gradually furloughed. The effect snowballs. And that loss of confidence will manifest itself in the economic data only ex-post, much as we see with the fall 2000 and early 2001 data.
Again, for clarity I am not suggesting that the stock market collapse itself causes the recession. Rather I am supposing it to be plausible that both the 2000 stock market selloff and the 2001 recession were completely unavoidable events, but were triggered by a concurrent collapse in economic confidence that occurred somewhere in the third quarter of 2000, which was when the layoffs began to turn strongly upwards, the stock market peaked, and the data got much softer.
The thing that this way of thinking solves for me is the 2011 conundrum. The prior post showed some data from the Chemical Board that the numbers in 2011 did not look good. Unemployment claims jumped 10% in the middle part of the year, and US GDP printed its worst quarter since the recession with a -1.5% Q1 number. And Europe looked to be falling absolutely apart. I think in this context there was a fairly high chance for the economy to crash into the mountains, and its not a coincidence that this corresponds to a 22% drop in the S&P between the May highs and the October lows, the most severe correction of this bull market.
To use the football metaphor, if we play that game 100 times, I think in 30-35 of those times, Europe tears apart and the world economy slides into recession. Certainly it was an event where one false move from one single leader could have sent things into the abyss, and I don’t know that type of outcome can ever be predicted.
So what, if anything, can we do about this? While I truly believe at least SOME recessions can be spotted in real time, given the personal stakes, I’m looking for a harder trigger.
Exiting Equities: The Yield Curve
One of the metrics getting much discussion of late, with a great deal of merit, is about the spread in the yield curve: the difference between long and short term interest rates. The Fed tightens monetary policy by setting the very short term ‘prime’ rate, which is what you see in a money market account. Longer term rates are set by the sentiment of the investor community at large based on their views on inflation, economic performance, etc. Usually, investors require additional interest to get them to commit for longer durations, and therefore the shape of the yield curve is upward sloping. In some instances though, short term rates are higher than long term rates. This is called an inverted yield curve.
Philosophically it means that the market is not worried about longterm inflation but is instead concerned with longterm return of principal. It also means the bond market is pricing in a period where the implied forward interest rates are zero or negative. Historically, the yield curve has been a very solid indicator of future economic trouble, although as a timing mechanism, it is rather poor.
Today, I spent some time looking at how the yield curve spread works as a predictor of future equity market returns. For purposes of clarity, I am defining the ‘Yield Curve Spread’ as the difference between the 10 year and the 2 year Treasury rate, and equity market returns to be the total nominal market returns, inclusive of dividends, for the S&P 500.
If I break the yield curve spread down and its subsequent returns into deciles, something very noteworthy emerges in the 1st and 2nd deciles: low levels of the yield spread are associated with extremely poor 3 year equity returns.
If I combine this information with the belief that the ultimate triggering event for a recession is itself random, in my view, the prudent conclusion is that the time to reduce (or perhaps eliminate entirely) equity exposure is when the yield curve spread is very weak. Certainly negative, but perhaps setting a threshold of 0.2 depending on tolerance for risk. For clarity, at the time this work was authored, we were at a range of about .66, solidly in the fourth quartile and not consistent with an impending recession. That has sense narrowed to a band of 0.25 (as of August 2018), alarmingly in the area where 3 year returns have historically been rather terrible.
The Re-Entry Point: The NBER Recession Pronouncement
So when do I purchase again? Rather than subjectively trying to declare a bottom of the recession, I thought about what is the most observable, bulletproof way to determine whether a recession is occurring or has happened? And what happens if you just buy the market at the time that you are 100% sure that the recession is underway?
It turns out we can look to the National Bureau of Economic Research to see when their business cycle pronouncements have been made.
Typically, the announcement of a recession is made several months to a year after the recession. In April 1991, they announced the recession began in July 1990. For March 2001, the announcement was made November 26, 2001. And for December 2007, the announcement came in December 1, 2008. So, I decided to back test how a strategy would have performed the past 30 years. The parameters of the strategy are:
- Sell stocks, buy bonds with negative yield curve. This sells stocks (and does sit out some very nice gains in 1999, 2000, 2006, and 2007) effectively in:
- Summer 1998
- December 2005
- Sell bonds, buy stocks in 12 monthly installments over 1 year after NBER pronouncement is made:
- Repurchase – 11/01 – 11/02
- Repurchase – 12/08 – 12/09
Performance of this strategy relative to the S&P 500 for a $10k initial investment:
Importantly, this delivers better returns with lower volatility:
Certainly, one should take backcasts with a tremendous grain of salt, but philosophically this line of strategy is extremely attractive to me in that it unifies a great many of my beliefs:
- It reduces equity allocation when the bond market is saying all is not normal with the economy, but allows for the exact timing of the recession to be random.
- Its entry and exit mechanisms are very clear and unambiguous and require no interpretation whatsoever.
- It allows me to be more bullish when I have reason to believe valuations will be extremely depressed (post recession) and less bullish when I have reason to believe the economy is behaving weirdly (yield curve).
I therefore view it somewhat as a gambler leaving the table when he knows the cards are no longer in his favor. On the exit path, it sacrifices the absolute peak late cycle gains that you might see in a 1999 or 2000 in exchange for a more certain indicator that things are highly probable to end in a recession. On the entry path, it will not catch the absolute lows but instead dollar cost averages when you are definitely sure the economy has been in recession for quite some time.