#310 | The Proprietary Index Duration Bomb
The conventional wisdom, at the moment, is that bonds are garbage. In a rising rate environment, bonds are set to deliver the worst of both worlds – sub-optimal yields and capital devaluations. Savers who crave the income and stability of fixed income will get neither. What’s the solution? The life insurance industry is quick to step up, pointing out that its fixed products aren’t marked-to-market to the same degree as bonds, providing something of a safe haven in a rising rate environment while still delivering potentially attractive returns via indexed upside. As usual, insurance products – and particularly indexed insurance products – are positioned as the balm for every ailment.
The problem, as I’ve written before, is that indexed insurance products face some unique challenges in a rising interest rate environment, particularly in the form of higher option costs. Indexed UL products rely on portfolio yields to generate attractive caps in a low-rate environment. But when interest rates increase, not only will portfolio yields lag market interest rates but option prices will also increase, leading to a reduction of caps when interest rates go up (all else being equal). That effect, coupled with sky-high volatility skew that is increasing the cost to hedge caps, has forced life insurers to begin to look to proprietary indices for illustrated – and hopefully real-world – performance for indexed insurance products.
Life insurers want proprietary indices that are cheap to hedge and deliver fantastic lookback performance. That’s the recipe for sales success. And life insurers, in my view, have willingly made a deal with the devil to get what they want. They know that these proprietary indices are highly complex, often unpredictable, generally unexplainable by the agents who sell the them and universally incomprehensible to the clients who buy them – they know these things and they’re willing to make the trade in order to get sales.
But what life insurers very likely didn’t know is that the deal involved something else – long-duration fixed income exposure. Every proprietary index in the market has a volatility control mechanism that generally works to blend high volatility equity exposure with low volatility fixed income exposure to meet the stated volatility target for the index, which is what allows for cheap and stable options. Volatility control is the buy-in to play the high-stakes game of placing indices at insurance companies.
In the early days, proprietary indices used the simplest, most liquid and most logical low volatility asset to counterbalance equity exposure – cash. Obviously. The problem with cash, though, is that interest rates have been so low over the past couple of decades that cash doesn’t have a return. Blending cash into the index to control volatility essentially requires using a dead asset, at least for backtesting purposes.
Enter long-duration fixed income, the magical asset class. Over the past couple of decades, long-duration fixed income has notched strong and steady returns courtesy of a consistently declining interest rate environment. Long-duration fixed income is, as one major index creator said to me a few weeks, the only asset class with a Shape ratio of 2 in recent history. Substituting long-duration fixed income for cash in a proprietary index is like a cyclist going on a doping regimen. From outward appearances, nothing has changed – and yet, suddenly, that cyclist is passing all of his peers. What do the other cyclists do? They start to dope too. Why? Because they have to in order to keep up.
What long-duration fixed income gives in a falling interest rate environment, it takes in a rising interest rate environment. This isn’t an abstract issue. The Fed has made it abundantly clear that they intend to systematically tighten rates in order to combat rampant inflation in the market. 10 Year Treasuries have responded apace, nearly doubling since January of 2021 and sitting at their highest levels since 2019 with ample upside to go.
What do rising interest rates mean for proprietary indices that are reliant on long-duration fixed income to generate attractive backtested performance? It ain’t pretty. To give you a sense for how these indices interact with changes in interest rates, I’ve put together a very simple analysis that looks at the daily changes in index values relative to daily changes in the 10 Year Treasury (TNX) yield. In theory, there should be a very slight positive correlation between daily changes in the TNX and an equity index like the S&P 500. Why? Because rising yields in the TNX signal a risk-on trade that benefits equities and falling yields in TNX signal a risk-off trade as demand for safe havens increase.
That’s exactly what we see in the data going from 1/4/2021 until now, with daily observations ranked from the best daily S&P 500 index return to the worst S&P 500 index return. In generally, there is almost no correlation between daily changes in the TNX and the S&P 500, but what little correlation exists is positive (10.21%). Take a look at the results below:
Now, let’s take one step into the world of volatility-controlled indices by looking at the S&P 500 Daily Risk Control 5% index, which blends the S&P 500 with cash to hit a volatility target of 5%. Take a look at the results using the exact same analysis below:
Here again, the correlation is very slight and very slightly positive – just 10.95%, nearly identical to the correlation of 10.21% in the base S&P 500. This is exactly what you’d expect because the S&P 500 Daily Risk Control 5% index doesn’t have a long-duration fixed income component, so why would it exhibit a correlation fundamentally different from the base S&P 500? It wouldn’t and it doesn’t.
However, a few years ago S&P created its own variant of Daily Risk Control indices that involved a long-duration fixed income component in response to the continued evolution in the proprietary index market towards using that asset class. The result was the S&P 500 Futures Daily Risk Control index, which uses TNX as the primary asset class for volatility control rather than cash. We would expect, therefore, for that index to exhibit quite a bit more correlation to movements in TNX – and that’s exactly what we see:
Here, the correlation between daily movements in TNX and movements in the index is strong and negative, sitting at -54.05%. I put a polynomial trend line into the chart to really make the story pop. That trend line shows what I’ll charitably call the cross of death – a strong tendency for the daily index returns to be inversely correlated to TNX daily changes. If the TNX yield dropped, then the index generally went up. If the TNX yield increased, then the index generally went down. In this index, generally means fully 67% of the time. Only 33% of the time did the index go the opposite direction of what TNX would dictate.
And this index is hardly alone. Another prime example of this phenomenon comes in the form of the Merrill Lynch Strategic Balance (MLSB) index. MLSB is, by proprietary index standards, positively ancient. It has been in market with AIG since 2014 and is a fairly simple structure that blends exposure to long-duration fixed income and the S&P 500. Both the sales and performance of the index has been extremely strong – no surprise in a falling rate environment. But the bloom has recently come off the rose. Take a look at the MLSB using the same analysis as the previous indices:
MLSB sports an even stronger negative correlation than the S&P 500 Futures Daily Risk Control 5% – a whopping -79.68%. In my view, it is not a stretch to say that MLSB is a thinly disguised trade on TNX returns. Where the TNX goes, MLSB follows. How many advisors who have sold MLSB or clients who bought it understand this dynamic? Certainly less than 1% and probably none. MLSB is an indicator of the first generation of proprietary indices that used (and still use) long-duration fixed income assets to juice lookback performance without taking into account duration risk. The other big index in this category, I would argue, is the similarly-ancient Bloomberg US Dynamic Balance Index, now in its second generation. Take a look at the Bloomberg US Dynamic Balance II ER index using the same daily return comparison as the others:
BUDBI is a bit less correlated to TNX than MLSB with a -50.66% correlation and 66% of observations lining up with TNX movements. Why is that? Because BUDBI doesn’t use TNX – it uses the Barclays Agg. That’s what makes this graph and the correlation even more jaw-dropping. TNX isn’t even in this index and yet it has high predictive power for the direction of index returns on any given day, especially in the high loss and high interest rate right tail of the chart. I could keep going, but suffice it to say that the majority of indices that I put through my analysis with long-duration fixed income components exhibited very, very similar behavior, with correlations ranging from -62% (Barclays TrailBlazer Sectors 5 and S&P MARC 5) to -8.47% (PIMCO Global Optima).
By the looks of it, these indices are duration time bombs. In a rising interest rate environment, particularly one where there’s a fair bit of equity volatility (like there is right now), these indices are going to take an absolute drubbing. We’re already seeing this play out in market. Some proprietary indices with volatility control mechanisms are, unbelievably enough, posting returns from 1/4/2021 until 4/6/2022 that are negative even as the S&P 500 is up more than 20% over the same period. It’s one thing to lag the S&P 500. That’s to be expected. But how did these indices deliver negative returns with equity performance that strong?
Because, simply put, these indices were sitting in long-duration fixed income while waiting to ride out the high equity volatility of 2021 and the trade turned against them when interest rates went up. How do we know? Just look at the two S&P Daily Risk Control 5% indices. The old index that uses cash is up 7.11% since the beginning of 2021, but the new index that uses long-duration fixed income is down -2.3%. Over the same period, BUDBI has dropped 1.58%, S&P MARC 5 dropped 3.3% and the MLSB is down a jaw dropping 7.05%. A whole crop of other indices with relatively high TNX correlation – JP Morgan Mozaic II, Barclays Trailblazer Sectors 5, JP Morgan Efficiente 5 – are barely breaking even. The only reasonable explanation is the duration trade.
In response, the banks and asset managers who create these indices have begun to incorporate increasingly sophisticated means to reap the benefits of exposure to long-duration fixed income assets for backtesting purposes while mitigating the risk of rising interest rates in the real world. These strategies, which can generally be called duration overlays, dictate duration exposure based on a variety of signals, usually related to interest rate momentum. For example, if interest rates are rising, the index might automatically allocate out of long-duration fixed income and into short-duration or cash. Vice versa if interest rates are falling.
These new variants of proprietary indices have been exhibiting markedly different correlations to TNX than their progenitors. Take a look at the S&P PRISM index found in Securian’s proprietary life product with Annexus, BGA IUL, which has a complex formula dictating when and how to get into TNX and then, from what I can tell, uses a more traditional cash volatility control mechanism. The net effect is less apparent connection and correlation (6.6%) with TNX:
The same goes for HSBC’s Aipex Index, which is in a suite of Athene annuities, and uses IBM Watson to select constituents. It has a cash volatility control mechanism – no long-duration fixed income. And, as you’d expect, there’s very little correlation to TNX (10.83%):
More interestingly, Allianz recently partnered with PIMCO on the PIMCO Tactical Balanced ER index, which is designed specifically to address duration risk by using the PIMCO Synthetic Bond Index with a duration overlay. Tactical Balanced ER is, essentially, doing the opposite of what BUDBI is doing by systematically increasing and decreasing duration exposure as interest rates move. Therefore, we’d expect to see Tactical Balanced ER look completely different from basically every other index in this set, with daily returns that are highly positively correlated to movements in TNX. No surprise, that’s exactly what the analysis below shows. The positive correlation between TNX daily movements and index daily movements is a whopping 30%, the highest of any index by a country mile. Take a look:
To paraphrase what one managing director at an investment bank told me recently, they’re learning from their past mistakes with duration and are becoming more sophisticated in how they’re engineering the indices. The tragedy, though, is that hundreds of thousands of clients with tens of billions of dollars linked to these old-style indices are likely going to be incredibly disappointed with the performance of their index in a rising interest rate environment, much to the embarrassment and chagrin of the advisors and life insurers who sold those products. There is a reckoning that will come due and, when it does, it won’t be pretty.
And it won’t be the last reckoning for proprietary indices. These new indices address the problem of duration, yes, but they introduce other issues. Take S&P Prism. Since the beginning of 2021, Prism has ripped at nearly 11% compounded, an unheard-of return for an index with a 5.5% volatility control. Why is that? Because one of the index constituents is the S&P GSCI ER Index, a commodities index that has increased an astounding 90% over the same period. Is that performance repeatable? Of course not, but it’s in the backtest. How about HSBC Aipex? For all of its savvy marketing about the power of AI, the index took a -1% drubbing since the beginning of 2021 for no apparent reason. That’s the problem with AI-based indices – you never really know what’s going on. And for PIMCO Tactical Balanced ER, it currently has a negative duration, which means the values will drop if rates drop. Pick your poison.
That’s not a throwaway line. I mean it – if you’re going to sell an insurance product with a proprietary indices, then you need to pick your poison. You need to know the fact patterns that will break the index and make an educated choice with the client about whether the diversification benefits and potential performance benefits of a proprietary index are worth the risks. It is impossible to do that for all 110+ proprietary indices that are currently in market. You have to pick a handful, preferably offered in insurance products written by companies you like, and do your due diligence. It takes time. It takes effort. You have to get educated. It’s not a short process. You can’t be cavalier about it. And at the end of all of that, you should probably hedge your bet by using at least a 25% S&P 500 allocation which, for all its flaws, is a transparent and broadly understood index that doesn’t need to be explained. That’s the roadmap, in my view, for responsible use of proprietary indices.