#418 | The Future of Engineered Indices
Chat GPT Summary
Engineered indices, now numbering over 200 in indexed annuities and life insurance products, have evolved through three key phases: the “Brand Name Era” (2009-2015), the “Illustration War Era” (2016-2021), and the “Great Pivot” (2022-2024). Initially, these indices were marketed based on high-profile bank and asset manager names, with a focus on backtested performance, but they often failed to deliver in real-world conditions. The shift to lower volatility targets in the second phase resulted in cheap options but poor performance during rising interest rates. In response, the latest indices have moved to higher volatility targets and more sophisticated strategies like long/short positions, aiming to improve returns. However, the real challenge lies in creating indices that consistently deliver strong returns regardless of market conditions, with diversification as a potential solution. Ultimately, the future of engineered indices may depend on their integration into diversified portfolios, rather than relying on backtested returns or brand-name appeal.
Full Article
Earlier this year, engineered indices crossed over a dubious milestone – 200 indices in indexed annuity and life insurance products. It’s a staggering figure considering that there was not a single engineered index in the insurance market as recently as 15 years ago. By now, the marketing appeal of these indices is well known and well understood. Engineered indices backtest and illustrate phenomenally well, even after all of the changes to AG 49. It’s not hard to ace a test when you have the answer sheet right in front of you.
But it is hard to ace the test and to put on a convincing act that even though the answer sheet was on your desk, you didn’t look at it. That’s more or less what the purveyors of engineered indices have to do in order to lend credibility to their index concepts. The story always has to be that the reason the indices illustrate and backtest so well isn’t because they know the historical data, but instead because they simply have great concepts that have proven the test of time and will work just as well in the future. The problem is that it’s hard for anyone to tell the difference. History is known. The future is not. And in the middle lies the industrial complex spitting out engineered indices into the indexed insurance market.
I’ve written several articles in the past dissecting engineered index construction and how various factors impact the performance of a particular index, but the upshot is that there are basically two things going on in any index – the strategy and the structure. The strategy is the sexy part that everyone talks about. Each index has underlying investments that range from vanilla S&P 500 exposure to exotic alternative investments. There is always a “thesis” for the strategy of the index that drives the explanation for why a given index will do well in the future or under certain conditions. All the better if the thesis is tied to a famous academic or backed by a large asset manager or is based on some sort of hot philosophy floating around the financial world.
The structure of the index, however, often goes largely unnoticed. Linking the interest credits in an insurance product to an external index requires that the carrier hedge the credits with options. Option pricing dynamics, therefore, play a central role in structuring an index. “Wild” indices such as the S&P 500 have deep option markets that provide continuous and efficient pricing. Custom indices, on the other hand, have no natural hedge market. If a bank were to write options on a custom index, they’d have to charge an enormous premium to take into account the uncertainty around the characteristics of the index, which creates pricing risk for the bank.
The solution is to tame the indices by structuring them to constrain the various factors of option pricing. The easiest one is volatility control. By reallocating the index composition or injecting cash into the index to maintain a constant volatility target, the bank knows that the volatility is going to fall within a very tight range relative to what it would be without the volatility control mechanism. The tighter the range, the more certainty for the bank, the more competitively priced the options – and the better rates for customers. We’ll get back to that dynamic later on.
Interest rates are also a factor in option pricing. The higher the interest rate, the higher the price of the option. But if the index deducts the interest rate from the index values, then the interest rate component is removed from the option price. It goes back to the classic idea of the iceberg and what is visible above the water and below the water. By using an Excess Return index, which deducts the interest rate directly from the index itself, the bank can push interest rates from below the water (the option price) up to the surface (the index return). The same mechanism can be used to surface transaction costs and dividends into the index itself rather than the option price. The net result is a low and very stable option price that presents very little risk to the bank due to the structure of the index. The known parameters of the index allow the bank to make a market for options where a market wouldn’t otherwise exist on favorable terms for insurers and their policyholders.
Over the past 15 years, engineered indices have gone through three distinct cycles in strategies and structures – and we now have enough real data to see how these indices have performed and, even more importantly, how the purveyors of these indices have reacted to changing macroeconomic conditions.
The Brand Name Era – 2009-2015
Early on, the primary appeal of engineered indices rested on the reputation of bank or asset manager that put its name on the index. Nationwide New Heights FIA adding JP Morgan’s Mozaic index in 2015 was an inflection point in the industry. No name garners as much reverence in financial services as JP Morgan. The fact that they were even aware of indexed annuities and willing to put their name on an index to be used in one caused headturns. But when New Heights went on to propel Nationwide into the ranks of the top 3 FIA writers and made (reportedly) tens of millions of dollars for JP Morgan, every other bank and asset manager took notice.
The result was a flood of indices into market from a wide range of well-known names from 2009 to 2015, the first cohort of engineered indices. Volatility targets generally ranged from 5% to 10%. Some indices were Total Return, others Price Return and many were Excess Return. A few were equity-oriented, such as the S&P 500 Daily Risk Control series, but most were either “balanced” between stock and bonds (such as the Bloomberg US Dynamic Balance index in Allianz products) or multi-asset strategies that incorporated equities, fixed income, commodities and other alternative asset classes, such as JP Morgan Mozaic. Did the differences in structure and strategy result in different returns from 2016-2024? Of course, but probably not as much as you might think. Take a look at the CAGR for these 30 indices since 2016. Volatility targets are the color codes. Each bar is an index.
The rule of thumb is what you’d expect – higher volatility targets generally resulted in higher returns. That’s a structural distinction, not a strategic distinction, and those higher returns are offset by lower offered participation rates due to higher option prices. But the strategic differences largely got washed out. There are multi-assets that did well and multi-assets that did poorly. The same goes for balanced and equity-focused strategies. But at the end of the day, risk and return are linked up. Low risk, in the form of a low volatility target, produces low returns.
These early engineered indices taught insurers some important lessons. First, agents like high participation rates with no caps. If choosing between a 10% volatility target with a lower participation rate and a 5% volatility target with a higher participation rate, they’ll take the latter because the optics are better, even though the two options should perform identically over the long haul, all else being equal.
Second, backtested performance is a selling point. You can find old Nationwide marketing flyers showing JP Morgan Mozaic outperforming the S&P 500 through the selected backtested period of time. There is no reason on any basis to think that Mozaic could keep that up – and yet, Nationwide put it in the marketing flyer and agents bought into the idea. From 2016 to 2024, the S&P 500 produced a 12% CAGR. Mozaic produced just 3.27%. That’s about what you’d expect given the structures of the two indices, but not at all what was insinuated by the marketing brochure.
Finally, agents and distributors didn’t draw a distinction between Total Return, Price Return and Excess Return indices. Given the prevailing rate environment, that’s understandable. When interest rates are low, the rate deduction from index performance for Excess Return isn’t noticeable. But Excess Return allows for cheaper and more stable option prices, which ultimately results in better rates for policyholders. It’s the superior solution for engineered indices and, as you’ll see, it became the standard in the next phase.
The Illustration War Era – 2016-2021
We think of the illustration war in the context of Indexed UL but, as I’ve written before, what we experienced in life insurance pales in comparison to what has been happening in annuities. It is not uncommon to see double-digit illustrated returns in indexed annuity policies, including products that link the illustrated income to the illustrated return, such as in Allianz’s best selling 222 and ABC products. But the mechanics that make indices appealing for annuity illustrations also work for life insurance illustrations – cheap options with high backtested performance.
The 104 engineered indices of this era are almost entirely uniform in structure. All but handful have volatility targets between 4% and 5%. Virtually all are Excess Return. Many incorporate embedded fees. This sort of structure creates very cheap options, which is exactly what insurers wanted in a low interest rate environment. A standard Excess Return engineered index with a 5% volatility target and a 0.5% embedded fee has an option price of a little under 2.5%, including fees for the bank. A 100% participation rate, therefore, costs a scant 2.5% for this sort of engineered index whereas a 100% participation rate on the S&P 500 usually costs around 8%.
But low interest rates also changed the asset strategy of these indices as well. The broad consensus amongst banks during this era was that incorporating long-duration fixed income into the index was a necessity because no other asset class showed better risk-adjusted returns over the backtested period. In a consistently falling interest rate environment, long-duration fixed income almost continuously goes up in price. It was the magic dust for backtested performance. Until it wasn’t. The average return for this cohort of indices was -8% in 2022 as interest rates went up. Since 2016, this cohort of indices actually performed worse than the previous cohort despite the fact that many of the indices were created after 2016. Take a look:
The great irony is that these indices were designed specifically to illustrate well – and those same decisions led to sub-optimal performance in the real-world. Until 2022, long-duration fixed income had generated strong returns with very little risk. Suddenly, the opposite scenario unfolded. If these indices were calibrated with high volatility targets, then they could have still maintained a strong equity position to participate in the rally of 2023. Instead, they had low volatility targets to make the participation rate optics of the product look more appealing to agents and distributors. Those low volatility targets were almost completely chewed up by long-duration fixed income exposure, which was suddenly very volatile, with nothing left over for equities. If you looked at some of the flagship indices for major insurers, such as Allianz’s BUBDI series, many of them were carrying equity allocations of 10% or less throughout 2023 for the simple reason that they didn’t have any volatility budget leftover to spend on equities.
And, worse than that, the yield curve inverted, which means that the pure yield on long-duration fixed income was close to or potentially even less than the short term risk-free rate Excess Return deduction from the index. As a result, the net return on the long-duration fixed income sleeve was effectively zero. Here’s more or less how things played out for some indices. LDFI is long-duration fixed income.
The net result was that some indices produced negative returns in 2023 despite a ripping year for the S&P 500, as shown in the table below. This is a completely unacceptable outcome for producers and their clients. How are you supposed to sit down with a client and explain that even though the S&P 500 is up nearly 20%, the index in your FIA was down a few percentage points? Or, as is the case with the vast majority of these indices, performance was substantially less than 5%? It’s a terrible situation that was created entirely because banks and insurers were focused on backtested performance and participation rate optics rather than robust real-world performance. They were essentially driving by looking in the rearview mirror and, not surprisingly, ran smack into the brick wall of 2023. You’ll notice that all of the bar colors are showing actual engineered index performance for this year because all of the indices from this era were created prior to 2023.
The Great Pivot – 2022-2024
The response from insurers and engineered index providers has been swift. Higher interest rates have fundamentally changed the optics of the products on the FIA side and, increasingly, the same is true for IUL products due to the introduction of new portfolios. Low volatility targets produce participation rates that are north of 200% – which makes no sense to anyone who doesn’t have a good grip on option pricing. As a result, there has been a wholesale shift towards higher volatility targets for the 53 indices that have been introduced from the beginning of 2022 until now. Results since 2016 are below:
Higher volatility targets have certainly produced better backtested index performance at the cost of lower participation rates. Theoretically, the participation rate is purely a vanity metric. A 200% participation rate on a 5% volatility target index should be the same as a 100% participation rate on a 10% volatility target index* as long as the underlying portfolio remains unchanged. The real question, then, is how these new indices stack up against similarly structured indices from previous generations. In other words – is the better performance of the most recent crop of indices simply a result of pushing to higher volatility targets or have they actually started to do something different in the strategy of the indices themselves?
On average, something appears to have changed. The latest cohort of 4-6% volatility target indices have delivered an average of 102bps more in compounded return since 2016 relative to the 2016-2021 cohort. But there are two factors going on. First, there are way fewer indices with 4-6% volatility targets that have been created in the past 2 years. Second, they have the benefit of hindsight, particularly for a couple of crucial years – 2020 and 2023. If an index does well in those two years, then it’s probably going to outperform the rest of the group. As a result, indices from the latest vintage have – not surprisingly – optimized for performance in those years. Check out the performance of indices by vintage in 2020:
The indices that didn’t do well in 2020 were the ones that actually existed in 2020 and the ones that did well in 2020 were the ones that were created after 2020. If ever you wanted to see the power of hindsight optimization in one chart, this is it. Many of the new indices have sophisticated long/short overlays that allow the index to rapidly convert to a short position in stocks, bonds or both. Not surprisingly, these long/short overlays have calibrations that work brilliantly in 2020. If an index could pause on the dip in March and then catch the rally for the rest of the year, then it can outperform the S&P 500 – which is exactly what happens for all of the indices to the left of the yellow bar, which is the S&P 500. We see a similar effect in 2023 as the Mag 7 dominated S&P 500 performance. Indices that happened to be tilted in that direction did incredibly well. Again, not surprisingly, most of the indices that did well in 2023 are of the most recent vintage:
At the risk of oversimplification, the engineered indices of the past couple of years are designed to solve yesterday’s problems. They adroitly avoid getting locked into a large long-duration fixed income allocation and even take it a step further by shorting fixed income as rates rise. They tend to tilt away from traditional value-oriented equity plays and towards trendy tech indices. These indices are created for a very particular and peculiar environment – and, as it turns out, they’re already wearing thin. Take a look at YTD 2024 performance for indices color coded by vintage:
This year, with live data, there is a much lower concentration of new-breed indices at the top of the heap. In fact, some of the best performing indices this year are actually of the 2016-2021 vintage that has fared so poorly. Building today’s indices to solve yesterday’s problem is not the right formula for actual performance – no matter how good the backtest looks. What saved an index in 2020 may be exactly the same thing that drags it down in 2024. There is no such thing as a perfect index. Every single one has favorable scenarios and problematic scenarios.
The Future of Engineered Indices
So where do engineered indices go from here? In my view, the bloom is well off the rose. Some of the most celebrated, most vaunted, most sold indices have turned out to be duds. Some indices that flew under the radar, such as the S&P 500 Daily Risk Control series, have done exceedingly well. Marketing appeal, brand name and distribution support appear to be completely uncorrelated to actual results for customers. At the end of the day, there is simply no evidence that suggests that anyone can separate good indices from the bad ones, if that distinction even exists. You might as well just go blindfolded and throw darts.
That raises a bigger question – if we can’t delineate good indices from bad ones, then what’s the point in allocating to one at all? The industry has come up with all sorts of terrible answers to that question. Some folks will say that it actually is possible to choose the best index based on software or projections. Other folks say it’s all about brand names or academic research or particular asset classes. Carriers will skirt the question and simply show the outrageously high backtested or illustrated returns. Others will say that a track record of returns, however short, is proof that a strategy works. But for every index that supports an argument, there’s an index that spoils it.
There is, though, a real argument for engineered indices as a tool for diversification. The ideal solution would be for engineered indices to generally post positive returns even when the S&P 500 drops. One easy way to do that is to have a large long-duration fixed income allocation, at least that was the theory until fixed income and equity correlations flipped positive after a couple of decades of being generally negative. Another solution is to put alternatives and commodities into the mix. Some of the more modern indices are also trying to do it with long/short exposure. Even if these indices lag the S&P 500, they could serve as an important counterbalance in down years. Clients appreciate a 0% return in a down year – but they’d really appreciate a positive one.
Increasingly, it seems to me as though carriers and index providers are starting to look at engineered indices through the lens of diversification. The problem is that very few existing indices actually manage to pull off the stunt. Since 2014, the S&P 500 has posted negative calendar year price returns in 2015 (-0.73%), 2018 (-6.24%) and 2022 (-19.44%). Of the 187 engineered indices in market, only 76 posted positive returns in any year when the S&P 500 had negative returns and just 21 had positive returns in two of those years. None had positive returns in all 3 years. If diversification is the goal, then it’s fair to say that the current crop of engineered indices isn’t delivering the goods.
The challenge, I think, is that making an index that is truly diversified and doesn’t drag on returns is a very challenging engineering problem. Creating an index with cheap options? An intern can handle that. Absurdly high backtested performance? A French grandes ecole-trained quant** would have that done before his morning espresso. Putting a little bit of sizzle into an asset strategy? Easy. But non-correlation to US Large Cap equities while still checking all of the other boxes for returns as a standalone strategy? That’s a different animal altogether.
Part of the solution, perhaps, isn’t to sell these indices as standalone options. A better approach may be to package them into portfolios that include other crediting strategies where the overall return profile is the attraction. We’re already seeing this in the FIA space and Symetra is already kind of doing it in its IUL. More will come. The next evolution of engineered indices may not be about the index itself but about the portfolio of indices altogether. That sort of logic is independent of particular economic environments and doesn’t rest on ludicrous, often misleading backtested and illustrated performance.
Instead, it gets back to the core logic of indexed insurance as an asset class with certain return characteristics and the unique role that engineered indices can play. If these indices are going to prove their worth, they’re going to do it on fundamentals, not optics. Whether they can actually pull it off, however, is what remains to be seen.
*The math isn’t quite that clean but it’s close. The difference is in the practicalities of option pricing by the banks and the fact that some of the fees are denominated by notional, others aren’t.
**Yes, most quants are French.
