Your money or your life
Life settlements are emerging as a viable form of exposure to the longevity asset class. Martin Steward asks if there is any place for them in a pension fund portfolio
We all hate insurance. For the least cynical it buys peace of mind, but for most it is just money flowing out to protect against losses, thefts, illnesses, cancelled flights and car crashes that never happen. One might think life insurance is different - death and taxes famously being our only certainties - but some estimates put the number of US life policies that actually pay out death benefits at just 12%. Around 85% are allowed to lapse, despite the fact that US policyholders have been allowed to sell them to someone else for almost a century. So far only those missing 3% are taking advantage.
That ‘life settlement' market is growing (it now represents more than $20bn) and inroads into the rest of the trillions of dollars of US death benefit could be opened thanks to recognition from the US Senate that this secondary market is useful for ‘seniors' struggling for liquidity after the financial crisis, and new legislation in Washington State, likely to be replicated elsewhere, that requires insurance companies to inform policyholders of the life settlement option. Furthermore, policies bought for estate planning which already make up a big part of those that come to market are beginning to look redundant for those who benefited from the Bush administration's raising of the US inheritance tax threshold.
The establishment of the European Life Settlement Association (ELSA) in May - existing bodies include the US's Life Insurance Settlement Association (LISA) and Germany's Spitzenverband der Lebensversicherungszweitmarktbranche (BVZL) - sees the industry "promoting highest standards and educating investors", particularly outside Germany and the UK where secondary markets for endowment and life policies are up and running. "We have already seen interest from investors in Spain, the Netherlands, Luxemburg, Scandinavia," says Peter Jäderberg, managing partner with ArensJäderberg and an ELSA steering committee member.
The basic concept is simple. A policyholder sells to a life settlement fund manager via the agent who originally sold to them. The buyer spends about four months evaluating the policy and then pays the insurance company the premiums due until the original policyholder dies. The buyer receives the death benefit (or policy face value) from the insurer, pocketing the difference between that and the money spent buying the policy and maintaining the premiums. Typically, a buyer pays about 20% of face value and pays premiums of 2-3% per annum.
Why would pension funds want to buy this longevity risk when they lose so much sleep over the longevity risk attached to their liabilities? Demand from UK pension schemes, which are particularly exposed to longevity, has been muted compared with the Dutch market, where it is less of an issue.
"I can see a place for it," says Hewitt Associates' head of longevity solutions, Martin Bird, "but it's really a whole different conversation about alternative investments, and as soon as people hear ‘longevity' they focus on hedging - especially with the recent announcement of the Babcock deal."
However, the longevity risk in life settlements does not correlate perfectly with that of a pension fund. The latter is ‘macro' longevity - the risk associated with tens of thousands of anonymous people, priced straight off of actuarial mortality tables. Life settlements is ‘micro' longevity - specifically, the risk associated with wealthy octogenarians, whose policies have been individually medically underwritten. At this age, life expectancy is less volatile than for the middle-aged, who benefit most from medical advances. Against that, negative selection (if you are at death's door you are unlikely to want to sell your life insurance) means that a life settlements portfolio is unlikely to experience the same rate of mortality as the general actuarial tables predict for the same age range. Pricing of micro-longevity starts with those tables, but specialist life expectancy providers make significant adjustments based on the individual lives covered.
"Micro longevity is relatively non-correlated with a pension fund's macro longevity," says David Rawson-Mackenzie, managing director of Centurion Group, which runs a full suite of funds based on micro, macro and mixed longevity. "It's a bit like residential property as opposed to commercial property."
Patrick McAdams, member of the ELSA steering committee and investment director with Surrenda-Link Investment Management, which has worked with Benelux pension funds, agrees. "When you buy longevity risk you get to price and risk-adjust it," he notes. "Pension fund members' longevity goes on the books regardless - you have no control." There are opportunities for alpha beyond the simple differences between life settlement portfolios and actuarial tables. "The tables have very little information on heavily-impaired lives, which account for the majority of value," says Surrenda-Link's senior actuary, Ian Cotgias. "And we've seen cases where non-smoker rates were given to smokers as a special offer. US life insurers sometimes do things like this under commercial pressure, and you can find genuine value if you know where to look and understand how to adjust your table against the Valuation Basic Table."
In addition, micro-longevity risk does not all have to point in one direction. Another ELSA committee member, Andrew Wilkins of Catalyst Investment Group, tells of working with Australian and US pension funds to build longevity products that mix life settlements with structured settlements and annuities. If an insured person is seriously injured his insurer buys an annuity to provide an income over a set period or until death. The insured person can sell that annuity stream into the secondary market - leaving the buyer with mortality risk, a portfolio of which nicely hedges the premiums on a life settlement. Other products like endowments, reverse mortgages and life tenancies round out the longevity portfolio-construction offering.
In terms of a pension fund's two other key liability risks - duration and inflation - the picture is mixed. Policies tend to mature within 10-12 years, but back-ended cashflows offer relatively long duration. However, premiums increase with the cost of life insurance, while death benefits are fixed - presenting some inflation risk.
So investors have to be happy that yields outweigh these inevitable correlations with liabilities. For a seven-year holding, yields of 9-12% net are not unusual. "There are short-term opportunities to beat that, but that's not necessarily sustainable long term," says McAdams. "Current conditions see a lack of capital seeking policies but increasing supply because US seniors need to shore-up liquidity, having been hit by the stock markets."
Note that neither of these short-term yield-drivers is longevity-linked. Both refer to financial risk that has resulted in a fight to liquidity, and in the stress of last year they did not cancel each other out. Yields started 2008 at about 13%, peaked at 19% and have since come back. "At the end of last year a couple of funds had huge redemptions and had to sell at a very different price from what they had on their books," as Rawson-Mackenzie notes.
There is also a certain amount of credit exposure to the insurance companies that underwrite the policies. The majority of US States indemnify policyholders against the insolvency of the issuer - but not all. "State Guarantee Funds will make good the payments of insurers up to as much as $500,000," says Wilkins. "The average face value in our portfolios is $500,000 so that in the event of a failure the impact on the asset pool is minimal. In some cases we've seen guys that were happy with face values of $3-5m moving into the smaller stuff - a flight to safety, as it were - so we've seen a slight decrease in the IRRs of the small-face sector."
That is significant because much of the sales pitch talks about pure exposure to non-correlated longevity risk - which, of course, life settlements do provide when held to maturity. But last year's mark-to-market risk was clearly correlated with financial markets, and the jury is out as to how institutionalisation of the space - the development of electronic trading platforms by Life Exchange or by Goldman Sachs, Genworth Financial and National Financial Partners; the launch of Goldman's QxX.LS index; similar initiatives from Maple Life and Cantor Fitzgerald; and the general inflow of risk capital - will affect those correlations.
It is easy to overstate the importance of mark-to-market risk, though. Arguably, all it does is signal particularly good market entry points.
"Correlation is not an issue at all," insists Roland van den Brink, member of the executive board with Mn Services, which oversees $550m in life settlements for Pensioenfonds Metalektro (PME first allocated in 2006 after 18 months of preparation). "If you can get around 12% returns, why wouldn't you invest? Dutch pension funds have to mark-to-market, so it's true that PME's life settlement portfolio has suffered a loss on that basis. But PME will hold these to maturity so the yields that were targeted in 2006, 9% net, still hold as the yields we will achieve at the end of 10 years. While life settlement policies' mark-to-market valuations will fluctuate like an equity's, there is no guarantee you will see any return from an equity at the end of 10 years. The holder of a life policy will die one day, for sure, so that yield is guaranteed."
Still, even if investors can hold on for 100% longevity risk, pension funds don't need to be told that pricing that is a minefield. "They would have to get comfortable with a lot of detailed underwriting to invest," says Bird, "and the pensions industry hasn't even got to grips with analysing their own members' longevity yet." Furthermore, some players have done the life settlement industry no favours with their conduct in this key area. Most egregious was Mutual Benefits Corp, a $1bn fund for whom a medical doctor signed fraudulent letters and affidavits sent to investors claiming that independent State-licensed physicians were determining life expectancies. In fact, the company was generating these life expectancies itself - and, of course, low-balling outrageously.
"Getting LE right is the critical way to achieve value," says Cotgias. That means independent life expectancy assessment, clearly (leading names in the business include AVS Underwriting, Fasano Associates, 21st Services and ISC Services), and preferably more than one provider: even without outright fraud, whenever doctors are involved it's best to get a second (third and fourth) opinion. Insurance information provider A M Best found in 2008 that medical examiners had become more conservative in their assumptions, but were still capable of disagreeing over the same policy by as much as two years' life expectancy.
Furthermore, Cotgias points out that the independent providers tend to generate life expectancies based on all of the data they possess - covering transacting policies but also those that failed to transact because no price could be generated. In other words negative selection bias is not built in, which leads to predictions of more early deaths than would be realistic in a typical life settlements portfolio. Surrenda-Link applies that bias in-house, pushing mean life expectancy out by about 12-18% compared with VBT and tables generated by life expectancy providers.
"In 2007 we bought about 30% of our acquisitions through the tertiary market," says McAdams. "But that's a private market where everyone uses their own models and assumptions. Pension funds need to fully understand these critical factors to be sure reported performance is real. Valuation is a big challenge for the market as a whole and an important initial focus for ELSA."
Part of that challenge will be refereeing the tug of war between deterministic and probabilistic models. A deterministic approach focuses on median life expectancy and values a policy based on that single, set point in time throughout its life. A probabilistic approach uses the whole mortality curve to price the policy. It is more complex, but it enables a portfolio to be priced to absorb the inevitable statistical anomalies (the man who should have died at 87 living to be 100). Unfortunately, generating a useful mortality curve requires a big sample.
"If fund managers had portfolios of 10,000 lives they could use probabilistic models - they would have deaths occurring regularly so that actual-to-expected mortality could be matched up to the actuarial tables," says Rawson-Mackenzie. "But they don't."
The compromise most conservative managers have come up with is a hybrid which is essentially deterministic, but factors in extra life expectancy at the start and then pushes expected maturity dates further out as policies near their deterministic date, or adjusts to take account of new medical information. On top of that, some will apply stochastic modelling to stress-test cash flows - an important consideration in what can be fairly concentrated portfolios.
"In our view the best result comes from using all three of these methodologies," says Cotgias: "deterministic, for its simplicity; probabilistic for planning; and stochastic for stress testing."
Splitting hairs over life expectancy should not obscure the fact that the deep discounts paid for these policies mean an investor will only lose money on a conservatively-constructed portfolio if its insured lives exceed expectations by about 180%, on average - almost unimaginable. But valuation policy is an issue of shareholder equity as well as shareholder return. Because Mutual Benefits Corp was under-representing life expectancy so much, the escrow reserve set up to pay premiums was pitifully under-funded and the firm had to pull in new investors and liquidate newer policies to maintain older ones that had not matured. The structure looked horribly like a Ponzi scheme. Purely deterministic approaches to valuation run the same risk should longevity be seriously under-estimated.
"If you get in and out early you haven't really taken any longevity risk - or at least, you'll never know if the longevity risk was priced correctly or not," as Rawson-Mackenzie explains. "If you want to invest in longevity rather than the sales skill of the manager you need a valuation model which, like ours, is a hybrid of deterministic and probabilistic approaches."
Surrenda-Link's Cotgias agrees: "You have to avoid new money simply propping up the price and funding the original investors' gains - and in our view the only way to achieve that is by applying actuarial methodology."
Clearly, life settlements still nurses something of a Wild West reputation and requires thorough due diligence and specialist knowledge. The complex correlations with liabilities and (mark-to-market) with other assets also need to be borne in mind. But for those able to buy and hold, the guaranteed yields from this non-financial risk are difficult to ignore.