In 2023, I co-wrote an article on underperformance of indexed universal life (IUL) blocks. One of the potential drivers of that underperformance was policyholder behavior. Throughout the course of time, actuaries have learned many hard lessons about misestimating policyholder behavior and the efficiency of policyholders. In this article, I expand on the potential areas of policyholder behavior risk in IUL products and where actuaries should pay particular attention to the lessons of the past.
According to LIMRA, IUL new sales premium fell 5% in 2023, though the number of policies sold increased 20%. Despite this slight decline in premium, IUL continues to dominate both variable universal life (VUL) and fixed universal life (UL). Whether or not all of those IUL sales will generate future profits to meet pricing expectations is going to depend heavily on whether the actuaries’ assumptions related to policyholder behavior turn out to be reasonably accurate.
So what are the key areas of policyholder behavior risk that actuaries need to keep in mind for IUL? I spoke to several of these behavior risks in the prior article, but there are definitely more to consider. This article delves into additional behavior risks to be aware of and hits on the risks discussed in the prior article as well as provides some suggestions on how to monitor, understand and potentially reduce these risks.
IUL is still a fairly new product, and long-term behavior experience is limited. The industry has more long-term experience on non-secondary guarantee UL and VUL, and an initial guess might be that IUL will land somewhere in between UL and VUL surrender rates given its hybrid nature. WTW’s TOAMS 5 industry experience study, which explored fully underwritten life insurance experience between 2014 and 2018, indicated that ultimate surrender rates for non-secondary guarantee UL and VUL were 3.0% and 3.5% to 4.0%, respectively; however, that study also showed that secondary guarantee ultimate surrender rates were 2% or lower, and since the onset of COVID-19, many companies have seen stronger persistency than in the past.
Many IUL products have persistency bonus features that will encourage long-term persistency. Those features, in particular if the spread goes negative, can drive the product to be lapse-supported. The situation gets exacerbated if a company also has reverse select and ultimate cost of insurance rates or expects to ultimately have negative mortality margins. Insurance industry historical experience has shown that lapse-supported products often end up with ultimate surrender rates of 1% or lower. We recommend sensitivity testing of different ultimate surrender rates to fully understand the potential impact of varying long-term surrender behavior.
Through conversations with our clients and via our industry surveys, we are aware that many companies don’t model dynamic surrenders for IUL products — the thought being that IUL products won’t be sensitive to interest rate movements.
Companies should definitely ask themselves whether or not IUL will be sensitive to interest rate movements. Considering the significant increase in new money rates and most IULs being based on a portfolio rate approach, there is a strong possibility that surrender activity could happen as rates rise, including the potential impact of increased premium financed loan rates driving additional surrender activity. Companies should also be asking whether IUL behavior could be sensitive to index credited rates and the equity returns driving them. To the extent crediting ends up being lower than expected in illustrations, will policyholders begin to surrender or lapse at higher rates? Most IUL has been sold after 2008/2009 and has had strong equity markets driving strong crediting rates. To the extent we live through a period of poor equity market performance, will this potentially drive excess surrender activity? Significant cap changes are another potential trigger for a change in behavior. We suggest companies perform stochastic modeling on their IUL business to understand the impact of volatile equity markets on their profitability and at least consider testing dynamic surrender behavior that varies depending on interest rates or equity market performance.
A meaningful portion of IUL sales have been supported by premium financing where external loans are used to help pay for deposits into the IUL contract. This can look attractive for the policyholder when borrowing rates are low and the IUL illustration is showing 6% to 7% long-term crediting rates. What happens when those borrowing rates increase significantly and the leverage benefit starts to shrink or disappears? With the recent rise in rates, in particular at the short end of the curve that drives borrowing rates, policyholders may decide to surrender their contracts. Premium financing has been commonplace in Asia in the high-net-worth UL market, and we have heard from several companies that they are seeing meaningful increases in surrender rates given the recent rise in rates, and a portion of it is driven by the premium financing issues. Premium financing can introduce additional behavior risk, and companies that are aware of this practice should determine whether their dynamic surrender assumptions need to consider this external factor.
Many IUL policies are illustrated to generate future distributions from the policy for various purposes, such as supplementing retirement income or funding college tuition. Some IUL products offer wash loans (no spread earned on the loan), and the spread earned on loans can differ materially from the spread earned on non-loaned fund value. This introduces additional behavior risk as different loan or withdrawal behavior can drive different spread earnings. Contracts with low loan rates could also experience disintermediation if alternative investments become more attractive relative to their IUL policy, in particular if caps have been lowered and performance is lagging expectations. Poor performance and the inability to loan or withdraw amounts that were previously illustrated could also lead to an increase in surrender activity. We find companies commonly use simplified loan or withdrawal assumptions, and commonly these assumptions can become stale. Similar to how companies study mortality and lapse/surrender, companies should regularly monitor their loan and withdrawal behavior relative to assumptions and update those assumptions as needed.
Many IUL companies assume some level of decrements in setting their hedge targets, as most IUL contracts only pay the guaranteed crediting rate up until the end of the index year. For many companies, combined lapse/surrender rates will start off high in early durations and then decrease over time until mortality starts to have a more material impact on decrements; however, the lapse and surrender rates can vary by multiple factors (e.g., duration, attained age, issue age, presence of a surrender charge, face amount, risk class). So, if you’re using a simplified total decrement rate in hedge targets, you could certainly introduce hedge ineffectiveness, particularly if the assumed total decrement rate becomes stale due to lack of regular updating. The use of a single decrement rate can also lead to hedge ineffectiveness across issue year cohorts, as more recent IUL sales would typically have a higher actual decrement rate than previously issued vintages. We suggest routine monitoring of hedge effectiveness and the impact of decrements as well as regular updates of the hedge decrement assumption to help reduce this risk.
The impact of a lapse (termination with no value) versus a surrender (termination with value) can cause a meaningful difference in earnings. Historically many companies priced and modeled their UL products with a combined non-death termination rate and an overall premium persistency assumption. Where there is positive cash surrender value modeled, those total non-death terminations will lead to an expected revenue source from collected surrender charges. But what happens if the total decrement rate is made up of significantly more lapses than surrenders? That situation could lead to collecting less actual surrender charges than was expected in pricing. Surrender charges can be an important source of revenue in GAAP, and whether you are collecting the full surrender charge can affect short-term earnings.
We also think that the approach of modeling policies in aggregate with a premium persistency factor can be a less accurate method. An improved approach is to model policies on a seriatim basis and to use each policy’s recent premium payment history to model future premiums. Policies would be modeled with a surrender-only assumption, and projected lapses would then result from modeled fund exhaustion. You then would need to determine an assumption using experience and expert judgment as to how policyholders will react to projected lapse situations (i.e., will they choose to lapse or minimum fund their policy?). Improving the accuracy of models to better model lapse and surrender activity, as well as performing sensitivity tests, can help companies to fully understand the risks they are exposed to relative to lapse versus surrender and varying premium payment behavior.
Our suggested actions to help monitor, mitigate and understand IUL behavior risk are all reactive. Product design and charge structure can help to influence policyholder behavior and can also be a way to reduce the risks of behavior deviations. Removal of persistency bonuses are an obvious way to reduce the lapse-supported nature of a product. Having a balance of profits from different sources (e.g., mortality margins, spread, expense margins, surrender charge) can also help to mitigate the behavior risk impacts. Building in low to no spread as well as low mortality margins will likely drive a product toward being lapse-supported and/or generating low earnings in the future.
UL products contain non-guaranteed elements (NGEs) in addition to the cap rates. Contract language can be worded to potentially allow a company to adjust its NGEs to the extent future expected behavior is inconsistent with priced expectations.
Even though many companies model IUL just like it is UL, many key differences between the two products need to be considered. Those differences will drive variations in behavior relative to UL. As actuaries, we should learn from the past and try to avoid making the same misestimations that were made on many UL products. With that historical knowledge, and the greatly enhanced computing power and modeling tools that are available today, you should be able to better understand IUL behavior risks. To the extent your company chooses to take on those risks, you will at least have advised your company of the potential consequences of actual behavior deviating from expectations.
