Life expectancy is one of the largest influences on the performance of a life settlement investment, and the space continues to evolve with advances in new technology and society’s understanding of mortality. Life Risk News’ Greg Winterton spoke to Chris Conway, Chief Development Officer, ISC Services, Vince Granieri, CEO, Predictive Resources, John Lynch, Director of Actuarial and Underwriting Services, Longevity Holdings, Sean Malone, President and CEO, Longevity Services and S. Jay Olshansky, Chief Scientist and Co-Founder, Lapetus Solutions to get their views on the current state of their sector, its challenges and opportunities.
GW: Let’s start this year’s roundtable from an investor’s perspective. If you were hired by an institutional investor that was considering allocating to a life settlement asset manager, and that investor asked you: ‘What should I ask the manager about how they use life expectancy analysis?’, what would be your advice?
CC: One thing I think is very important to highlight for investors is that different managers use life expectancy reports differently, and how they use the information we provide is critical for investors to understand. Investors often think the underwriting work is driven solely by actuarial factors, that the ‘calculator is the key,’ but it isn’t. Calculators are driven by input; they don’t drive input. An incorrect risk assessment applied to the ‘best’ table will produce an invalid result. The risk assessment part of what we do determines the input and the calculators themselves all generally do the same thing in the same way. However, all underwriters do not assess risk the same way and investors need to understand the bases for a given underwriters approach.
VG: There are many considerations with this one because different managers use LEs differently. Since we are just getting started, let me provide a general comment and a specific comment. Generally, I would ask what led them to the decision to use LE estimates in the manner that they do. More specifically, I would ask for evidence that their decisions have been good ones (e.g.: their A/E experience is favorable and there are demonstrable differences between the cases they considered but rejected and ones they actually sold).
JL: I would want to make sure that the fund manager has a full understanding of the range of outcomes embedded in the reports. It’s important that fund managers understand that if mortality is understated by x%, then, for example, that means that y% of the lives might live twice as long. Understanding the full range of longevity risk is important.
Additionally, the underwriting report tries to give you a picture of mortality. It’s a snapshot; a point estimate that describes the nature of a mortality curve and embedded in that is a one-time estimate of an insured’s mortality profile. The longer you go since the most recent life expectancy report, the less accurate the previous picture becomes. Some fund managers are more aware of this than others.
SM: I would ask the manager how the life expectancy reports inform their own views of longevity. Do they focus on the mortality curve, the median LE, mean LE, or mortality rating? How does the summary of the patient’s medical history influence your view of the case? I think it’s important for life settlement asset managers to have their own unique perspective of life expectancy that is supported by the life expectancy underwriter’s reports.
JO: If I was the CIO for, say, a multi-family office, I would be most interested in how close the LE provider’s assessments come to the observed event of interest (e.g., maturation or alive). I wouldn’t care about observed differences in LE assessments of particular cases — this is expected and normal when different methods and procedures are used to generate such estimates. What I would say is that an investor needs to gain comfort with the credentials of those generating the assessment for the asset manager; ensuring that their asset manager uses LE providers that operate in the open, with no black box, where the LE assessments are explained and defended using the tools of medicine and science are what is important here. If an LE provider generates an LE without telling the manager how and why they arrived at their number means the investor should be pushing on the manager to change providers.
GW: Now let’s turn this around. If you were hired by a manager to help them ‘sell’ their life expectancy underwriting skills as a reason why the investor should allocate to them, what would you say?
VG: We’ve created a tool to test our clients’ underwriting skills (or their ability to uncover good cases) so the first step would be to evaluate their skills using this tool. Assuming they pass muster, then the asset manager can then explain to their potential investor that they have demonstrated their understanding of life expectancy underwriting, as this will have been validated by an independent, external provider. This could be used as the basis for explaining to the potential investor about how they have successfully bought and sold policies.
JL: The benefit to an asset manager of working with an external provider is that we’re providing an unbiased view. We’re not representing insureds, or carriers; we’re providing an unbiased view of a mortality profile. I’d say that it’s important to explain the value of an independent opinion to an investor.
Another important element is to use life expectancy providers that have the most data, because without credible data it’s impossible to claim accuracy statistically. Also important is to use an LE provider that admits where they might have got things wrong – and what they have done, or are doing, to reassess in a manner that makes sense.
SM: If I’m a life settlements asset / fund manager, I would tell the investor about the importance of using an independent, external provider that focuses solely on underwriting for life settlements. It’s very important in our industry to understand that general mortality tables are not appropriate for the life settlement market, because of the specific nature of the cohort that this industry focuses on. Data-driven and customized mortality tables help the fund manager to more accurately price policies, and consequently, the manager’s ability to deliver strong returns for their clients will be enhanced.
JO: I’d point out that there are only a limited number of providers and I’d explain the differences in the ways the different providers generate their estimates – which requires the LE provider to be transparent in terms of how they operate and therefore arrive at their conclusions.
Then I’d focus on how the manager can utilize this knowledge to their benefit – is there an arbitrage opportunity here? How do they view and use shorter or longer LEs relative to other companies evaluating the same patient? How do they decide whether to use a shorter or longer LE in what situation during the buying process? It’s worth emphasizing that some companies generate LEs that tend to be longer or shorter than those generated by other companies because some may not be utilizing the full medical record with a thorough understanding of the patient’s health trajectory across time and their response to treatments. For example, I reviewed a case not long ago of a patient that had what appeared to be a severe lung disease, but after a deeper inspection of the medical records, that individual had been responding well to the treatment protocol. If all one looks for are keywords for impairments that are linked to a prescribed debit to LE, they would be underestimating survival in this case. The reverse can also be true.
CC: Well, the first thing I would tell them is not to tell investors that they have “underwriting skills.” In general, they don’t. In the same way I am not an underwriter (I am an owner and operator of an underwriting firm), most asset managers are not underwriters (or actuaries, for that matter). What I would advise managers to tell prospective investors is (1). That they have a “view” (an opinion, if you will) relative to longevity risk, (2) what their view is based upon (ie: what process did they go through to develop their position), (3) how they apply their viewpoint to the external underwriting information they receive from external sources, including life expectancy assessments produced by independent underwriters, and (4) the degree to which their viewpoint is applied in their processes of buying, holding and selling longevity risk instruments.
GW: How much of a problem is false information on a medical record? Is it common, indeed, do insureds ever lie, and can this issue ever be solved for?
JL: Both our underwriting businesses, TwentyFirst and Fasano, have the default position that we are medical record-driven, and on the validated medical notes, we have a certain level of belief there that we feel is credible.
That said, we do scans for doctors that are no longer licensed, checks for social security numbers, and other scans that might provide us with insights into the validity of other data points. This all gives us a higher confidence level that a medical record is legitimate.
If our underwriters feel that there is misleading information or misrepresentation, they have the ability and authority to remove debits or include credits to account for that. And ultimately, we use this as a last resort, but we occasionally respond with a “no quote.” We saw something last year where someone wanted us to redo a LE without an updated report from the attending oncologist from the last LE report three years ago, which we responded with “no quote” and requested the updated medical notes from the oncologist before proceeding with a life expectancy report.
SM: There are different types of false information regarding medical records. A patient might not be truthful when telling their doctor how many alcoholic beverages they consume, or how often they smoke or exercise. I think this is common, and we use whatever details are in the records for our analysis. In rare occurrences, there might also be false medical records themselves. For example, someone might try to offer up a doctor’s note that is not authentic. In that case, we would inform that client that we cannot accept the document unless it is an actual medical record and it will be excluded from our analysis.
JO: There are lots of places where incorrect information can enter the LE estimation process because people may misrepresent their information, which is mostly unintentional when it happens. But largely, we have to use what is given to us and assume that it is accurate.
That said, regardless of what is placed in front of us, we still have to use good medical judgement with consistency. You might see two medical records, three months apart. One might have the insured at 200 pounds and the other at 170 pounds. That can make a big difference when you’re looking at BMI as a risk factor – we have to determine whether that’s a typo or whether the person actually lost 30 pounds in three months, and why, for example. In a situation like this, we stop our review, go back to the client and get a definitive weight on the patient. The same goes for situations where we see them listed as a non-smoker, and sometimes a current smoker. Sometimes people are listed as taking one medication for an impairment, and in the next medical record they are taking a different medication. Why? Was there a typo? Did the patient have to change medications for some reason? These are all significant variables in an assessment because different medications have different effects. When there is ambiguity in the medical records, we prefer to go back to the client to confirm what we’re seeing rather than assume anything. You have to go back and get up-to-date medical information in situations like these where you see inconsistencies.
CC: We generally do not see a copy of the original application for life insurance among the materials submitted to us for assessment. We’d like to, but clients do not provide this to us. That said, after 40 years working in and around the life insurance industry, I don’t think prospective insureds lie so much as they may be guided to present the best possible picture of themselves when they apply for new coverage. When applying to sell a policy in the secondary market, insureds may also be guided to emphasize impairments, but I think this is extremely rare given the underwriting process is very good at ferreting out any such activity and there are many other factors involved in determining eligibility for life settlement.
VG: These issues really begin with the initial life insurance application, which we don’t often see. Unsurprisingly, prospective insureds are focused on highlighting the positive aspects of their health, while potential life settlers are doing exactly the opposite. Two sides of the same coin if you will. Our assessments always search for the truth, irrespective of the biases introduced in the process.
As for the medical records themselves, it’s not so much that records are being falsified or someone else’s records are being substituted (although that does happen from time to time), it’s more errors of omission and commission. We fight this by eschewing self-reported information and requiring independent confirmation of these assertions. We also cross reference impairments with other information in the file (e.g.: the subject’s drug history and labs), to ensure that there is consistency. Someone claiming to be a diabetic should have insulin or metformin in their drug history and periodic A1c tests in their lab records.
GW: Many life settlement industry participants bemoan the time it takes to complete a sale. Moving on to electronic health records would seem, therefore, like a silver bullet for the industry. But it’s not that clear cut, is it?
SM: Electronic health records would probably be easier to source and analyse. I know that it can be difficult to get doctors or hospital staff to send medical records required for a life settlement transaction. Would that change if the records were electronic? I can say that as an LE provider, electronic health records would likely save minutes or hours per case, not days or weeks. We are just a part of the process to complete a sale.
JO: If the information in EHRs was accurate then yes, there would be a benefit in terms of time savings. Keyword searches, for example, would return results quicker than manually going through handwritten medical records. But you have to be 100% certain that handwritten notes are transformed accurately. At this time we just don’t trust the accuracy of EHR, which is why we read the medical records, especially if they are handwritten. So, for firms like ours that still read the medical records then the increased use of EHRs wouldn’t have an impact. Perhaps in the future we’ll feel more confident in using EHRs. I should point out that at Lapetus we are constantly looking for ways to improve and streamline our procedures, and we have explored various options for generating electronic medical record summaries – that is likely going to be our first move into automating one element of our assessment process.
In terms of the overall time span for a life settlement transaction, particularly in the secondary market, I doubt that speeding up this part of the LE assessment process would have a material impact on the market. There are many other components in a life settlement transaction that, if solved for, would have a much greater impact.
CC: The main thing here is that ultimately, user error is the primary risk, whether that’s paper records or electronic ones. Someone can say that they are not a smoker, and the person entering the data can check the wrong box. And correcting that data is enormously difficult. You can contest a credit report, for example. But to correct an error like this, which would have an enormous impact on your life insurance policy – and therefore on the life settlement industry, if that policy ever entered the market – is orders of magnitude in terms of time and complexity when compared to correcting a credit rating.
If every single life insurance policy that we see was in the form of an EHR then there would be a time savings, but I’m not sure this benefit would have as much of an impact on the life settlement market as it may seem on the outside. Many other factors go into the process of buying a life settlement, and they each add time to the sales cycle. EHRs are only one component of a larger process.
VG: First, I should point out that EHRs are only one part of Digital Health Data (DHD), along with claims data, RX data, portal base data and wearables. Fully understanding DHD can provide the needed holistic view.
As for EHR, the pros include speed, potential for widespread use, and a comprehensive view of the subject (including, diagnoses, prescription drug usage, laboratory and diagnostic test results, treatment plans, and more). The cons include low hit rates (35-45%), the prevalence of unstructured data (especially with respect to highly impaired lives), the need for extensive skills and technical abilities to fill in gaps and standardize available data and, of course, the cost in both time and money to remedy these weaknesses.
There are many misconceptions regarding EHR, many promulgated by wishful industry participants who deny or downplay the above cons. Further, there is a tendency to oversimplify the issue and presume the DHD presents a complete view of a subject’s health. One misconception is this idea that DHD alone can replace underwriting judgment in assessing risk. As appealing as that sounds, it just doesn’t work that way.
As for the transaction itself, we are not at the ‘silver bullet’ stage yet, and may never be. There is potential to move the needle, but premature adoption of this unproven technology will likely be painful for some investors who are lured in by the siren song of the early adopters. In the end, it may injure the credibility of this new approach and delay widespread adoption.
JL: Two things here. First, in terms of electronic health records themselves, they are still not 100% accurate. There are too many false positives, which may be beneficial for a shorter LE, but the accuracy levels are not there. We’re currently not endorsing current technology, and not using them in our underwriting.
In terms of the process to complete a life settlement, then the tertiary market obviously can move a lot quicker as you would already have more timely electronic records that can be validated or scanned. But that would only shave a few days off the total time in a secondary transaction. Is that really the biggest roadblock to quickening the buying process in life settlements? In a future state, with 100% accuracy and proper connectivity of availability of that data, and a standardised framework, there could be some pros there, but not at the current moment. This isn’t the piece that’s holding the market back.
GW: Last one for this year. In terms of your ability to provide as good of a product or service to your customers as you can, what is one thing holding you back, and one thing that enables you to do just that?
JO: The most important thing that enables us to do our job as well as we can is also the thing that holds us back, which is to be provided with the most updated, thorough, complete records on the patient as possible. If you want the best report possible from Lapetus, give us the most recent information that exists, but also provide us with a medical history so we can assess the health trajectory of the patient. Using only a snapshot in time of the patient’s medical history misses some of the most highly relevant information that can be used to assess survival prospects. Keep in mind that people move into and out of states of ill health, so health trajectories across time are an important element of the evaluation. We can only evaluate what we see. If we’re working with information that’s a year or two old, we’re not going to be able to give you as accurate of an LE as we can with data that’s only a month or so old.
CC: The one thing that holds us back the most is probably access to capital. It’s probably the most significant factor inhibiting our ability to evolve, invest, and most importantly, innovate more quickly. We do invest and we do pursue these objectives to the greatest degree possible, but there is not enough interest from investors in the underwriting sector itself despite its role in underpinning nearly all the assets and transactions across the industry. The thing we believe enables us to provide the best service we can is first and foremost our team of underwriters and the depth and breadth of their skill set, including our breadth of experience gleaned from having played many other roles in the industry. Having walked (many) miles in our client’s shoes over the years, we believe we truly understand their perspective and can therefore meet their needs where they are.
VG: As an eternal optimist, I will lead off by answering the second part of the question first – at Predictive, we have folks with many years of experience in the disciplines needed to provide an excellent life expectancy product – information technology, underwriting, actuarial, statistics, digital health data and industry knowledge. What’s holding us back? The transactional, short-term focus of the market, and its shallow understanding of the true nature of an LE.
JL: In terms of holding us back, then I’m not sure that there’s anything specific to LE providers like us, but there is a looming cloud for the life settlement market more broadly, which is that we’re now in a pivotal time with STOLI lives hitting expected mortality. A lot of these lives that are currently in the 85-95 range are expected to live to late-90s, maybe early 100s. Firms like ours can decipher which ones are more likely to and which aren’t. But a lot of buy and hold firms aren’t refreshing LEs and are relying on assessments from five+ years ago. If you think about how fast medical advancements have been just in the last few years, that means that these investors are operating with old information. There could be higher impairments which have now fallen off and you’re going to have an older yet healthier pool today than originally five years ago.
In terms of enabling, I look to the future – about 10 years’ time. Fast forward, and you’ll see Artificial General Intelligence emerging at 99% accuracy. We’ll be funnelling data through an assessment similar to how a highly trained medical director can perform assessments now, taking into consideration all current studies in reputable journals to link prior experience with future expectations. That could mean a future state where things could be turned around in real time. But 10 years is the earliest estimate in our view: I’d be wary of anyone who says they have figured it out today because LLMs are based on brute force predictive algorithms on past information only. The best quote I’ve heard describing the skillset of today’s ChatGPT and other LLMs are like having 10 highly skilled interns.
SM: Our customers enable us to provide the best service. We always appreciate client feedback or the opportunity to discuss a case that we’ve competed. However, if a customer wishes to discuss a case while we are working on it, or they are missing important documents or information, then the process could take a little longer. Our best work comes from good collaboration with our customers.
Chris Conway, Chief Development Officer, ISC Services
Vince Granieri, CEO, Predictive Resources
John Lynch, Director of Actuarial and Underwriting Services, Longevity Holdings
Sean Malone, President and CEO, Longevity Services
S. Jay Olshansky, Chief Scientist and Co-Founder, Lapetus Solutions
The views expressed in this article are those of the individuals.