Has this time been different?

One of the main tenets at this space has been to cut away the extremes in your [equity] investing strategy. There are those who see market tops and imminent crashes everywhere, and then there are those who believe debt-financed share buybacks and dividend payments can continue to propel the market higher forever. They are both wrong, but the persistence of these two narratives and their interaction has been a key story in this cycle. It is my firm belief that the oscillations between these two positions have created a huge middle ground, which allows investors to make money. In the peanut gallery we talk about "sector rotation," but it's more than that. It's also about different themes which cut across traditional equity sectors and allow for price movements of key industries—even country indices— in opposite directions. I suspect most market geeks would be able to agree on this over a pint in the pub. But can we quantify it? I suspect the first port of call for analysts would be to look at cross-asset correlations, and it certainly seems like something happened after the election last year. But I am not sure low cross-asset correlation is the smoking gun I am looking for. This indicator is mainly used as an argument for why macro traders should do well—they haven't been able to exploit it so far—and in any case, it doesn't appear as if cross-asset correlation has been systematically lower in this cycle. Rather, I would frame the key questions as follows; 

1) Do equity sectors in this cycle exhibit lower—or higher—return variations than normal in this cycle? 

2) Has individual sectors have been able to rise even as the overall market has declined in this cycle? 

3) The flip-side of number 2 is whether the bull market in this cycle been more resilient to value destruction in individual sectors?

In order to answer those questions—which isn't easy mind—I ran a number of studies using the S&P 500 as the benchmark index, and its GICS 2 sectors as my main sample. These exercises tend to be US centric simply because that is where the best data are. I think it would be worth examining these questions in other markets too. Crucially, I also think the evidence of divergence in global business cycles across regions and countries hints that this is not only an equity sector story. If you're not into quantitative finance, leave immediately. Otherwise, please read on. 


Have sector rotation been more prevalent in this cycle? 

For the first question, I created a diffusion index of three-month and 12-month returns on the GICS 2 sectors. The indices take a value of 1 if the absolute value of the returns is below 5% and 10% on a three-month and 12-month basis respectively. The chart below shows the fruit of that labour. 

At a first glance, it doesn't really suggest that much has changed in this cycle. In a broad sense, the return variability of sectors appears to have declined since the financial crisis in tandem with the maturing bull market. But the variations—the second derivative effectively—have been significant which provides a hint that this cycle is slightly different than in the run-up to the financial crisis. What's particularly interesting I think is that the movement in my diffusion index appears counterintuitive. If we take the bull market as given—which so far has been a sound bet—you would expect sell-offs to be associated with higher variation in sector returns. But the opposite appears to have happened since 2009. We have had many sell-offs since 2010, but two have been especially severe; the debt-ceiling crash in 2011, which was the financial market precursor to the EZ debt crisis, and the Chinese devaluation scare in the first quarter on 2015. Both of the periods, however, were associated with less return variability among sectors. This is to say that if you had been looking only at a sector dashboard, the sell-off would have looked relatively benign. By contrast, the almost uninterrupted increase in the S&P 500, which began at the end of 2012 was associated with comparatively high variation in sector returns. Similarly, the most recent march higher has also been associated with higher sector return variability. 

I am not completely sure that this vindicates my thesis, but it does indicate that this cycle is different from the bull run which preceded the 2008 crash. The chart shows that sector return variability fell sharply between 2004 and late 2006, which was the final stage in the bull market. In 2007, return variability increased slightly but remained fairly low even as the market started to sell off. It was only after Bear Stearns and Lehman that sector return variability soared, as everything got swept away in the crash. Indeed, the transition from record-low sector return variability in 2005 to record-high return volatility at the nadir in 2009 appears to have a been a key underlying driver of the deep dive in stocks. Finally, it is fascinating to see that the dot-com bubble had almost no impact on the variation in sector returns. From 1996 to 2003 when the market bottomed after the tech deluge the diffusion oscillated around a fairly constant mean and didn't seem to care much about either the bubble or the subsequent decline. 


Has sector rotation protected investors during sell-offs?  

Questions 2 and 3 are probably the key tests for my thesis. If sector rotation is supposed to protect investors during sell-offs, we should be able to show this relatively simple. To do this, I created two diffusion indices. The first tracks the number of sectors that positive returns on a three-month and 12-month basis even as the market falls. The second shows the number of sectors, which have negative returns even as the overall market is up.

chart 3.jpg

These models are sensitive to the thresholds employed, but they provide some support for my thesis. The first chart shows that the number of sectors with positive returns increased during the two major sell-offs in 2011 and 2015, providing a clearer picture of market movements hinted by the indices tracking return variability. This is in stark contrast to the bull market from 2003 to 2008, where this just didn't happen. Some sectors had positive returns in the initial stages of the sell-off in 2007, but in hindsight, that doesn't count. We know now that this was a bear trap. It's slightly different if we look at the bull run that ended in the Dot-com crash. In 1994-to-1995, when the S&P 500 had a soft patch, plenty of sectors enjoyed positive returns. Since then, however, the main index pretty much dictated everything until the selloff in the tech high fliers began in earnest. In conclusion, the evidence appears to suggest that individual sectors have been more resilient to sell-offs in this cycle. Specifically, we can make the point that major selloffs in the overall market appears to have coincided with—or even triggered by—a rotation between the major sectors. 

The final chart shows occurrences when the number of sectors that are falling on a three-month and 12-month basis rising as the market goes up. It's noisy but does not appear to show any substantial change in behaviour since the crisis, even though the average value of this index is up since 2008. The most interesting aspect of this chart is that it exhibits clusters. Since 2014, we appear to have been in a cluster of significant sector individual sector weakness even as the market has pushed higher. This is partly because this period includes the sharp sell-off in Q1 2016, but the number of sectors exhibiting weakness actually peaked in August 2016. Readers with well-curated hindsight will remember that this coincided with a big internal rotation in markets toward "reflation trades", a rotation which arguably reversed at some point in Q1 2017. 


What next? 

If you've made it this far, you deserve a badge of honour not least because it isn't clear that any of the analysis above is forward looking. The most dangerous thing about a study like the one above is that you spend an awful lot of time identifying a certain regime, and how to invest in it, only to realise that it has changed just as you're getting used to it. It's a bit like the difference between trading a market that is trending or ranging. We have ironclad rules for both, but deciding which regime we're in is the tricky part. I often think that this exercise is the finance/economics proof of Heisenberg's uncertainty principle. 

I think we can say with some validity that this cycle is different from the pre-crisis run up. Sector divergence has been more significant, which has been an important pressure valve for the market during sell-offs. This suggests that a major drawdown in the future doesn't have to be similar to the gut-wrenching crash in 2008. In other words, the next recession in the U.S. does not have to morph into a global financial crisis. The market behaviour during the dot-com crash, for example, offers hope. The charts above show that the run-up to the sell-off was characterised by significant sector rotation. This really is a euphemism for the fact that the tech sector was running the show, but that other industries were trying to break away from its singularity. Eventually, everything was correlated in the final stages of the sell-off, and I don't think that would ever change. 

I don't think investors should kid themselves into believing that this time is different; it never is. This bull market is, to a large extent, driven by excess liquidity and when that tide turns—and it will eventually—investors will get stung. But I stand on my point. We have seen cases of significant value destruction in retail, energy, financials and most recently telecoms, even as the market has ground higher. Anyone trading single names will also nod approvingly to the idea that many firms—even in strong sectors—have suffered. By contrast, secular growth stories such as big data, e-commerce, e-sport etc have propelled the market higher via the infamous FANGs and their brethren. Flows between these themes invariably mean-revert, and investors have been able to exploit that.

It would be criminal to expose you to this analysis without offering a concrete example. The 17 largest constituents of the MSCI World are Apple, Microsoft, Facebook, Johnson & Johnson, Amazon, Exxon, JPM, General Electric, Wells Fargo, Bank of America, AT&T, VZ, Pfizer, Nestle, Chevron, Home Depot and Comcast. Out of these mega caps, three currently stand out as being relatively attractive: General Electric, Verizon, and AT&T. John Hempton recently threw in the towel on telecoms, which is a red flag if there ever was one. That said, I leave you with the following chart, which shows the relative year-over-year performance of an equally weighted position of GE the S&P 500 telecoms index vis-a-vis the S&P 500. The portfolio is positioned accordingly, in hope that the past, at least for now, remains an accurate predictor of the future investing regime.