#321 | Midland National and the Legacy of Flat COIs

monochrome photo of man looking up

A subscriber reached out to tell me that a Midland National representative told him that the block referenced in this article was actually written by Clarica, which Midland acquired in 2003.

Despite the fact that virtually every Universal Life uses explicitly non-guaranteed and currently declared policy charges, there’s always a bit of a storm if a life insurer changes those rates for in-force policyholders. Producers and distributors think of policy charges in Universal Life as non-guaranteed-in-name-only – and they have some reason to believe that. Since the introduction of the illustration model regulation in the mid-1990s, any illustrated non-guaranteed values must be actuarially supportable based on recent and reasonable experience. The only time non-guaranteed charges should change, therefore, is when there’s been a major shift in experience.

Hence, the blowback to the spate of life insurers – Transamerica, Lincoln, John Hancock and Equitable, to name a few – who have raised Cost of Insurance charges in the face of mortality experience that is demonstrably at or better than expected. The subsequent lawsuits against the insurers surfaced the fact that Cost of Insurance charges aren’t just about mortality. They’re functionally a catch-all charge that can be used to cover other mispriced product elements or poor experience, particularly for lapse and policy funding behavior. And, consequently, those lawsuits have largely been settled in favor of the plaintiff policyholders.

Now we can add another name to the list – Midland National. It’s easy to read the headline and stop there, which is more or less what I did when I saw the notice from ITM 21st, the life insurance policy management company that has been able to spot these COI changes before nearly everyone else because of their massive block of managed policies. But an astute subscriber of TLPR forwarded the notice to me and pointed out that Midland National actually published the old COI rate and the new COI rate in a nice and neat table that was included by ITM 21st in their email for the 1997* policy that received the notice, a Security 2006 UL product written on a now-62 year old female.

So far, the COI increase at Midland is only for Security 2006 UL. And when you look at the table of policy charges, it ain’t hard to see why. The table below shows the old COI rate in dark green and the 1980 CSO in black:

The former COI rates for this Security 2006 UL policy were effectively zero, although not quite zero. They gradually increased from 0.26% of Net Amount at Risk to 0.32% by age 100, meaning that the mortality expense on $1 million of NAR would have been just $3,200 at age 100. That’s less than what you’d expect to see for a Cost of Insurance charge in the first policy year for a newly underwritten 62 year old female.

This raises a very simple question – what in the world was Midland National thinking with this product? But as the astute reader of TLPR pointed out, these sorts of things were common in the 1990s. He said he remembered UL products with full COI refunds after certain periods of time. And this Midland design reminds me of the Conseco policy that was the subject of a lawsuit a while back that, from what I recall, had zero Cost of Insurance charges on a non-guaranteed basis. It must not have been uncommon for features and designs like this Midland National policy to exist in the past, regardless of how bizarre they look today.

If the life insurer has severely underpriced the mortality element of the policy, then how are they making money? Presumably other policy charges, particularly the embedded investment spread. Think about how these illustrations must have looked in the 1990s, when interest crediting rates were 8% or higher. They probably all over-endowed even with relatively lean funding. From an actuarial standpoint, the mortality line item probably wasn’t that big compared to the asset spread. Therefore, you can kind of imagine the actuaries saying something like “assuming policyholders fund this appropriately and interest rates stay where they are today, this design is supportable” without noting that those assumptions are bogus and expose the insurer to huge rate and funding pattern risk. And, presumably, that’s exactly what happened.

But the most interesting part of this story is what Midland National did next. You would expect – and I certainly did expect – that Midland would slap a reasonable Cost of Insurance charge on this product to make up for past sins. That’s not what they did. They increased the rates by, on average, 500% (yes, five hundred percent) through age 99. Does that sound like a lot? It’s not. Take a look at the new COI rates relative to the 2015 VBT ultimate rates, which is the industry standard for underwritten life insurance policies and a fair baseline for future mortality assumptions:

On average, the new COI rates are just 20% of 2015 VBT after age 90. There is simply no way to say that the new COI rates are a reasonable reflection of the actual mortality experience for this policy. How do we explain the difference? A few ways. First, the new COI rates are actually higher than VBT for the first 5 years or so, which means that the profitability problem may be perceived as more short term than long term. Second, it’s possible that this is the first of what may be several COI rate increases on this block to bring it into line. Or, third, it’s also possible that the other policy charges, including investment spread, are kicking off enough income to shore up profitability. For now.

What makes this COI increase interesting, in my view, is that it’s a bit of a different storyline from some of the other COI increases we’ve seen. In general, life insurers have said they’ve raised COI rates for two reasons – STOLI and guaranteed interest rates. I don’t think either of those are at play with this Midland National policy. If the guaranteed interest rate was an issue, the COI rate increase would have been much, much higher. And although I’m sure that every secondary policy buyer would love to have a boatload of these policies, the reality is that they probably don’t. If they did, then Midland would have tried the Equitable and John Hancock tactic of limiting the COI increase to the policies owned by secondary buyers. So it seems as though this is just a flat out example of a mispriced COI slope – one that undoubtedly made the product a very compelling proposition when it was sold.

Early on in my career, I regularly put together what I called “spaghetti graphs,” which were basically the COI slopes ripped out of any Universal Life product I could get my hands on. The point of those graphs was twofold. First, to show the incredible variation in actual COI rates across policies, hence the spaghetti comparison. The second was to say that, all else being equal, a client would rather have a relatively flat COI rate slope than a steep one for the simple reason that when the policy inevitably is either underfunded or the crediting rate drops, the new premium to maintain coverage will be lower than if the COI rate slope is steep.

However, my analysis missed a key risk factor – low and flat COI rate slopes are more likely to be changed in the future. That’s a logical statement, not an empirical one. We haven’t seen enough COI rate changes to be able to know, for sure, which policies being sold today are most at risk of a future COI rate increase. But it stands to reason that the policies that are priced with negative mortality margin (meaning, the COI rate is less than the actual mortality rate) are the ones that will be first to have future adjustments.

And there’s still a lot of that going on in the market. Take a look at a modern “spaghetti graph” showing COI rates for about 70 Indexed UL and Variable UL policies in market over the past couple of years as a percentage of the 2015 VBT (solid black line), which is a reasonable basis for underwritten mortality for this 45 year old Preferred Male. I’ve also added the 2017 CSO rate (solid black) and the average of all of the life products (red line).

Before we draw any conclusions, let me just make a few observations about the nature of these mortality slopes. Remember that this graph is shows the COI rates as a percentage of 2015 VBT, which is why the 2015 VBT line is flat. I built the graph this way so that it’s easier to read but also because I think 2015 VBT is the fair starting point for mortality. It’s the table that most life insurers use to compare their own mortality against. It’s essentially life insurance underwritten mortality by risk class (Preferred NS, in this graph). The operative 2017 CSO table, by contrast, is Ultimate rates only (not underwritten), which is why it’s vastly different early on and then beings to converge with the 2015 VBT over time as the effect of underwriting wears off.

A few things become immediately apparent. First and foremost is the fact that the majority of life insurers sandbag their first few years of mortality. Some sandbag more than others. Allianz, for example, essentially charges the maximum 2017 CSO rate in the first 5 years, which in year 1 is 800% of the 2015 VBT. Global Atlantic isn’t far behind. On the flip side, other companies charge very close to 2015 VBT for the first few years – John Hancock, Pacific Life (PIA6), Nationwide and Lincoln, to name a few.

Second, after the first few years, the spread seems to tighten. It seems as though there’s quite a bit more consistency in COI rates after the first 15 years and before age 100. But is that really the case? The rates themselves are a lot bigger in these years, so the fact that some life insurers are charging 50% of VBT at age 80, for example, when other life insurers are at 125% of VBT is a massive difference in actual COI rates.

Third, at around age 90, companies start to do some funky stuff with their COI rates. Some drop to zero. Some simply fade off at some very low rate. To paraphrase what one senior actuary said to me a few years ago, “who cares about COI rates past age 95? No one is around to get them anyway.” Fair point – and that seems to be exactly the mantra that some of these life insurers are taking.

Fourth, it’s very clear that virtually every life insurer is pricing some level of mortality improvement into their COI slope because, on average, the COI rates fall below the 2015 VBT after age 68, just 23 years into these contracts. At age 100, the average COI rate is just 64% of 2015 VBT. That’s quite a bit of mortality improvement. But after age 100, it seems as though some life insurers essentially revert back to the CSO table and take advantage of the fact that it has a very steep slope in later years.

All of these things basically lead up to one conclusion – mortality pricing is art, not science. There are wildly different approaches to how COI rates are set for different policies and it shows up in a graph like this one. Low COI rates are a benefit unto themselves, as I’ve long argued, but only if the rates hold. Future rate changes are more of a risk in some policies than others, it seems, and we have to take that into account when making product decisions.

But there’s also a systemic risk issue. The entire industry seems to have made the same assumption about future mortality improvements. If those improvements don’t show up as expected, then virtually all of these policies are going to have a negative mortality margin. It’s possible that we could see wholesale, industry-wide shifts in mortality slopes for Universal Life that would be justifiable on the basis of actual experience. Possible, but not probable. Mortality improvements pre-COVID have been pretty consistent. The last few years have thrown a wrench into some of that and time will tell if it lasts.

Will Midland National be the last company to change the rates on a mispriced COI slope? Hardly. The reality is that all Universal Life in its current form is built on non-guaranteed assumptions. We must think of it that way and communicate to clients accordingly. Universal Life isn’t wink wink non-guaranteed. It’s for-real non-guaranteed. Actuaries make mistakes. Policies get mispriced. And sometimes some of the most attractive policies on the market are the ones most mispriced – and if that mispricing is on a non-guaranteed basis, then the presumption should be that it will eventually get fixed.

In short, the presumption for clients’ policy risk management purposes is anything particularly advantageous that can change, will change. The policy should either be funded from the start to be resilient to those changes or the client should be prepared to, at some point, change their funding pattern in order to maintain coverage – for better or for worse. The only fixed premium permanent products in market are Whole Life and Secondary Guarantee UL**. It is a huge mistake to sell non-guaranteed Universal Life – much less Indexed UL, and certainly not Variable UL – as if it is, too.

*This seems implausible – why would a 1997 policy be called Security 2006? Because, I think, there was an original 10 year angle to it. That or it’s a typo. I’m still working to find that out.

**This isn’t technically true for GUL policies that use shadow account designs, but it is practically true for clients who pay on time, every time, which is the definition of fixed premium.