Stock-Bond Positive Correlation

Warning: this might be a self-serving post. A juice of pure confirmation bias. The comfy blanket I look for when I am scared.

You probably know my model portfolio (and if do not, click on the link ;)). 66% of it, or 100% depending on how you perceive it, is linked to a mix of stocks and bonds, the infamous 60/40. The combination of stocks and bonds, a cornerstone of many investment portfolios, allows investors to balance risk and reward, generate income, preserve capital, and potentially increase their wealth over the long term.

So they say.

But what truly forms the foundation of this mix? Why bonds should provide an effective protection to stock volatility while having a positive expected return? Many thinks it was just a fluke of the recent past, a time when stocks and bonds were negatively correlated. A party that might be over:

If I look at how the “synthetic NTSX”

has performed in the past, in periods of positive and negative S&B correlation, there is not much to be optimistic about.

Positive correlation:

Negative correlation:

Positive correlation:

Granted, our backtest includes 8 years of positive correlation and 21 of negative one. Not the greatest sample. But not a great omen either looking forward?

BankerOnWheels included this research piece in one of their weekend reading lists (yes, this blog is included in their list, so I understand if you do not consider my opinion as objective…still, I strongly suggest you check it out every Saturday Friday Thursday now?!? ;P) and I found it amazing. Mostly because I agree with their outlook – stocks and bonds will continue to be positively correlated from now on – and because they reassured me that the future of a stock-bond portfolio might not be that bleak.

[Quick, relevant, reminder. Here at The Italian Leather Sofa we are well aware of the shortcomings of the stock-bond combo. That’s why the Model Portfolio includes those bloody trend ETFs]

I’ll post some excerpts from the research and add my comments.

Fiscal sustainability concerns tend to put upward pressure on interest rates and downward pressure on economic growth, which translates into positive stock-bond correlation. Despite a long post-COVID expansion, US debt-to-GDP is elevated and is poised to continue its climb. Given the current political climate, there seems to be little will to address these issues in the near term, with fiscal worries likely to extend and continue to provide support for positive stock-bond correlation. It has been several
decades since investors have had to put US fiscal policy near the top of their list of concerns – perhaps not since the early Reagan years. But that seems to be the current direction of travel, with the UK’s recent experience a cautionary tale of how rapidly such worries can intensify.

Assiduous readers of this blog should recall my mentions of Cem Karsan’s worry about populism. It might be hard to predict who will be the next president in the US or elsewhere (well, not exactly everywhere…) but for sure the Gov purses will stay open in any scenario.

The current conduct of fiscal and monetary policy has led to an increase in interest rate uncertainty, pushing rate volatility to multi-year highs (in contrast to other measures of uncertainty). To the extent that fiscal sustainability and Fed independence and discretionary policymaking remain on the minds of investors, interest rate uncertainty may remain elevated too. This has direct implications for stock-bond correlation as short-term interest rate volatility is an input into our models of stock-bond correlation. Indeed, elevated interest rate volatility is responsible for a good deal of the uptick in stock-bond correlation and its shift from negative to positive.

This is quite interesting and they explain why it is the case in another paper:

The volatility of changes in risk-free rates: This component has a positive effect on stock-bond correlation because risk-free rates are part of the discount factor of both stocks and bonds. Unsustainable fiscal policy and procyclical monetary policy will likely lead to large swings in policy rates, greater rate volatility and positive stock-bond correlation. Sustainable fiscal policy and countercyclical monetary policy (with monetary authorities predictably raising rates in response to strong growth and cutting rates in the face of weak growth) support negative stock-bond correlation.

I would say that, as risk-free rate levels increase, the discount factor becomes more important in determining the value of stocks and bonds. And also, as the discount factor moves, stocks and bonds move in the same direction.

In other words, the denominator in both equations gains relevance compared to the numerator.

It is impressive to notice that bonds are maybe the only market where volatility has been elevated in recent years. That’s a market where I would like to be a volatility-seller, not like stupid $JEPI & friends.

The venerable 60/40 portfolio had a record bad year in 2022, falling 17% as correlation turned positive; stocks fell 18% and bonds fell 16% (Figure 9). While painful, it is important not to learn the wrong lessons from that annus horribilis. We cannot emphasize enough that positive correlation, in and of itself, did not cause dismal portfolio performance – indeed, 2023 was a strong year for the 60/40 portfolio despite the prevailing positive correlation regime. Both history and theory point to the enduring value of a balanced portfolio regardless of correlation regime, with little to suggest that periods of positive stock-bond correlation are particularly challenging for multi-asset investing.

Ben Carlson’s “intuition” might be valuable here: stocks and bonds have both positive EV (and negative skew), therefore it is understandable to expect they are up, alone but also together, most of the time.

Year over year, the 60/40 returns have been more volatile in the positive correlation regime but overall the situation doesn’t seem to have dramatically changed.

PortfolioCharts allows us to get the same picture but on a real basis:

Again, I do not think you would be able to spot the moment correlation changed if I presented you only the above picture.

What if we zoom out even further?

Before the 70s, bonds and stocks spent most of the time in a positive correlation regime: 25 years vs 10 with a negative correlation. I wanted to check this because the original study by Cliff Asness on what would later become NTSX (with some differences that we will see later) covered 1926-1993.

Here is the study made by WisomTree, the issuer of NTSX. They first replicated Asness’s results and then extended the period by 25 years:

Sure, the 94-18 period offered even better results for the strategy but still, the 26-93 time frame, when stocks and bonds were mainly positively correlated, was not a total bust. Quite the contrary.

NTSX differs from Cliff’s portfolio in 3 ways:

• lower leverage
• no exposure to credit (they use only Treasuries)
• lower duration

Unfortunately, WisdomTree did not do an extended backtest on their strategy as they did for the Asness one but we can still infer some conclusions. The leverage part is immaterial (155% vs 150%). Credit Risk is highly correlated with stocks, so not a real diversifier: its removal should reduce returns but also drawdowns, since credit spreads typically increase when stocks crash. The lower duration should also help when the correlation is positive, as described above in the risk-free volatility bit: a lower duration should mean the NTSX components should be less correlated compared to Asness ones. For all these factors, NTSX performs worse than Asness portfolio when the correlation is negative, as it happened in the first three years since the fund was launched:

The following two figures explain soundly the consequences at portfolio level of the change in correlation:

While moving from negative to positive correlation might feel like a tectonic shift, the investor does not have to make radical portfolio changes to achieve the previous level of returns (or risk).

This is the most important conclusion of the study:

Bonds still provide a tail hedge to steep stock declines. Positive correlation between stock and bond returns means that when one asset class experiences above (below) average returns it is likely – depending on the strength of the correlation – that the other asset class will as well. Positive correlation does not imply that both asset classes will experience simultaneous declines, despite the 2022 experience (Figure 12). In fact, over the last 50y and when correlation is positive:

• Stocks and bonds are far more likely to climb in tandem on a 12m basis (67% of the time) than to decline in tandem (3% of the time). • Stocks and bonds still frequently hedge each other. Their returns have opposite signs 30% of the time (i.e., 13% + 17%, as shown in Figure 12).

• When stocks do decline, bonds tend to go up far more frequently (i.e., 85% of the time = 17% ÷ (17%+3%)) than down. This 85% figure is about the same regardless of correlation regime!

• Even when correlation was positive, during extreme equity declines, like Black Monday (1987) and the Russian Crisis (1998), bonds rallied, asserting their tail-hedging characteristics, irrespective of correlation regime (Figure 13).

What I am reading now:

Follow me on Twitter @nprotasoni