The Critical Nature of Life Insurance Policy Due Diligence in the Life Settlement Asset Class

Many investors in the life settlements asset class do not fully grasp the complexities underlying the asset before making the decision to invest in it. On its face, a life settlement appears relatively simple – it is effectively a negative carry, discounted cash flow asset that matures at some future point – not dissimilar to a zero-coupon bond. A closer look, however, reveals that life settlements are more complicated to acquire and manage than other comparable discounted cash flow assets. Hence, it is critical that careful diligence be undertaken before an investor makes the decision to invest in life settlements and begins deploying capital.

There are three important due diligence topics associated with the life settlement asset class that a prudent investor should address with the fund manager in order to gain a better understanding of the life settlement asset class.

Guaranteed/Non-guaranteed Cost Of Insurance (COI)

From a financial viewpoint, a life insurance policy is composed of two “legs,” – the “face value leg” on the assets side and the “premiums leg” on the liabilities side. While these elements are already stochastic in nature, as the actual reception/payment depends on the future realizations of a random variable (length of the reference life), buying non-guaranteed products that will be carried on a current assumptions basis adds a source of uncertainty to the investment.

Universal life insurance products, which are overwhelmingly the most purchased life settlement asset, were launched in the US in the 1980s as a more flexible (and cheaper) form of permanent life insurance as compared to whole life insurance products. While whole life products have fixed premium payments that are calculated by the insurance carrier to build up substantial cash value over time, universal life products leave it to the policy owners to decide whether they wish to build up cash value in the policy (reducing the net amount at risk and therefore premiums due in the future) or not.

Most importantly, though, for universal life products, the COI rates applied by the carrier (referred to as “current assumptions COIs”) are generally lower, and in certain cases much lower, than the guaranteed maximum COIs that are written in the policy contract. Insurance carriers may, under certain conditions and subject to regulatory supervision, adjust their COI rates upwards. (The market recognized this, especially after the COIs increase waves of the mid/late 2010s, when dozens of universal life insurance products experienced COI increases imposed by insurance carriers that in certain circumstances exceeded +200%, therefore causing the impacted products to lose value significantly or even entirely.)

As mentioned, the market shows some level of sensitivity to the COI-increase risk component and guaranteed products tend, everything else being equal, to trade at more aggressive IRR levels, even if these “IRR corrections”, may be misleading. We recognise the value of guarantees, especially taking into consideration the long-term nature of an investment into a life insurance policy, which often entails a decade(s)-long horizon.

There are different due diligence procedures that we follow to reduce the risk of suffering losses as a result of a COI increase on policies purchased and kept in force on a current assumptions COI basis. Among these, when buying policies on a non-guaranteed basis, we look for the existence of secondary guarantees in the policy contract. While these secondary guarantees may be less attractive on a present value basis as compared to keeping the policy in force on the basis of current assumptions COI rates, they represent an option that can be exercised (i.e., the option to switch from current assumption-based premium optimization to secondary guarantee-based premium optimisation) should a COI increase be implemented in the future. This would provide a better “backstop” as compared to the one foreseen under the policy contract for the “main account”, the maximum (guaranteed) COI rates.

While it is true that in previous COI increase waves, very few products were the subject of COI increases implemented by carriers to the fullest extent possible (applying maximum contractual COI rates), the option of switching to secondary, guarantees-based premiums proved valuable for increases that weren’t close to the maximum possible, which increases the importance of looking for these carefully during due diligence, even if the policy is being purchased on a non-guaranteed basis.

Medical Underwriting and Life Expectancies

Medical underwriting of the insured life – and the life expectancy estimate of that life – is the most important variable in determining whether a life settlement policy has value – and if so, how much. A significant portion of the life insurance policies shopped on the secondary market show meaningful differences in value that are derived from relatively small variations in life expectancy assumptions. The variability around life expectancy estimates is a key indicator of the riskiness of the transaction, and this variability must be specifically analysed. However, having a clear perception of said variability is often problematic, and getting a proper reward for this risk (i.e., being able to purchase at appropriate discounts) is even more so.

This problem is aggravated for investors by the sophisticated “life expectancies selection mechanism” sourcing that intermediaries have developed over time, combined with the fact that in most cases, only one life expectancy is actually presented to the market during the bidding process, even for cases that are underwritten clinically, i.e., even when the life expectancy estimate is highly sensitive to the underwriter’s judgement and therefore a secondary, additional source of uncertainty is added to the game.

Currently, a handful of life expectancy underwriters account for more than 90% of the life expectancy underwriting market. They employ different proprietary mortality tables (meaning one life expectancy underwriter could assign a 90-month life expectancy, and another an 80-month life expectancy to the same 89-year-old, non-smoking male) and different proprietary underwriting manuals, hence, they often end up with different views (i.e., different mortality multipliers and life expectancies) on the same life, even starting with the same medical information set.

For example, the combination of different tables and views on the same life (given the same medical information set) could generate a range of “potentially admissible values” that ranges from $20,000 to $250,000 on a policy with a face value of $400,000. Absurd as it might seem, this scenario is not rare. This situation embeds a high level of risk in addition to the possibility of excess longevity risk (i.e., the risk of seeing the individual surviving beyond the life expectancy, assuming the life expectancy was correctly underwritten), and simply taking the average of the extremes does not provide any appropriate reward for the extra risk, especially when in presence of a potential life expectancies selection bias. In these circumstances, greater control by the investor of the medical underwriting, however achieved, is essential.

Another interesting area is that of lives that are underwritten by life expectancy underwriters at close-to-standard mortality ratings (up to 150-175%). It appears there is a sourcing/sell-side’s push to make policies viable that might not really “price” (i.e., they would not normally be a candidate for a life settlement), and this needs to be taken into consideration each time these cases are reviewed. In the end, sourcing policies is a tough and costly job: once an intermediary has worked several weeks on a file, after many calls to the potential sellers (likely many of them ending up in the voicemail) to try to obtain medical records, illustrations, policy contracts and so on, it is understandable that they will try all available routes to make the case work.

For these policies, on top of all the other due diligence steps and the analysis of the insured’s health status, a closer look at the history of the life policy itself, including issuing assumptions, product type and vintage as well as the initial table ratings, is prudent and necessary. As an example, life insurance policies issued at super preferred mortality rates to wealthy individuals in California will likely price – even if these life insurance policies weren’t originally issued several years prior – if valued at a 150%-or-so rating applied on a table that’s based on nationwide, non-wealth specific historical survivorship data.

However, this isn’t necessarily a no-loss, stochastic arbitrage. Also, in this case, we believe that looking at how the present value of the policy behaves by changing life expectancy assumptions helps in understanding the riskiness of the transaction. Although the industry seems to lack a “unifying theory” about life insurance policies valuation and riskiness, certain aspects should, in our opinion, at least be considered in a practical way. Life insurance policies whose present value (at the buyer’s representative IRR) varies by, say, –50% or more in dollar terms for a small variation in the life expectancy (in particular, the mortality multiplier assigned by the underwriter) are in our view very risky and should, in our opinion, be avoided unless the price can be adjusted to embed a proper remuneration for such a coin flip-like game. This scenario is all but uncommon in our experience and the best way to deal with it is in our opinion to always have the ability to ditch all the cases that show similar characteristics without embedding a proper reward.

IRRs and Model Sensitivity

The industry’s standard valuation methodology relies on a discounted cash flow model that weighs future cash flows on the basis of an actuarial array that is fitted to the specific life, by the application of a mortality multiplier that forces the probability distribution to have an expected value (the expected duration of the reference life) equal to that of the life expectancy report(s) obtained from third party life expectancy underwriters. Probability-weighted cash flows should then be discounted at an appropriate discount rate.

As mentioned above, a naked investment into a life settlement policy generally implies an exposure to one single, stochastically distributed positive inflow (the death benefit, on the asset side) and to a series of outflows (the premiums due, on the liability side). On policies that are kept in-force on a current assumptions basis following a minimum premium approach, liabilities tend to significantly increase over time. However, if the selected discount rate is high enough and the LE employed is short enough, the pricing model “won’t see” distant liabilities clearly enough to properly reflect them into pricing until they become closer, i.e., after years of carrying. In short, the exponentials employed in the discounted cash flows model may amplify the effects of wrong assumptions.

Interestingly, while when using a discounted cash flows model to price an asset-only instrument (i.e., embedding only positive projected cash flows, like in bonds) adding the right amount of basis points to the base discount rate could be an effective way to better reflect volatility/riskiness in the projected future cash flows. And when a discounted cash flows model is employed to price liabilities, adding basis points to the base discount rate reduces the present value of the liabilities and therefore results in a more aggressive, and not more conservative, estimate of such liabilities.

If a single level discount rate is employed across assets and liabilities to value life settlement policies without making distinctions between their assets and liabilities components, a net mixed result is obtained. How misleading the net mixed result will depend on several characteristics of the life settlement policy at hand, and in our opinion a key variable is the shape of the projected premium streams. In our view, therefore, transaction-implied IRRs (level across assets and liabilities) should be handled with care, especially if one wants to use them to compare the relative value of life insurance policies that embed drastically different features. This is, of course, also true when it comes to ongoing book valuations.

Life insurance policies and the mathematical formulae upon which they are based, are inherently and surprisingly complicated. Policies are designed by insurance actuaries, who spend years studying mortality and other risks using mathematics, statistics, and financial theories, and who then must take a series of tests that extend over several years to become a certified actuary.  Life insurance companies have been very profitable because the actuaries who design the products that are their stock in trade have a deep and comprehensive understanding of mortality and statistics.  It is, therefore, advisable for investors in life settlements to also have an understanding of these disciplines and their application to the asset class.

Maurizio Pellegrini is Life ILS Manager at Azimut Investments, Azimut Group