Overestimates on older age populations may have led to schemes overspending on buyouts, but to what extent? Natasha Browne investigates
Buyout contracts are increasingly attractive as scheme sponsors look to get defined benefit (DB) liabilities off their balance sheets. The market has gone from strength-to-strength over the past few years, with the value of business quadrupling in the first half of 2014, according to LCP (PP Online, 11 August).
But what if sponsors found out they were overpaying for buyouts? Would this be acceptable if they were happy with the price at the time, or would it erode trust in these transactions? Schemes could be faced with these questions after a recent report from the Pensions Institute (PI) found buyouts may be overpriced.
The cause is significant flaws in mortality data calculated by the Office for National Statistics (ONS), according to the PI. These data discrepancies can be traced back to 1919 and Spanish flu.
It was the end of the Great War and soldiers were returning home. While birth rates were low in the first half of 1919, because of a Spanish flu pandemic, they were considerably higher in the second half. This is important because the ONS estimated populations on a mid-year basis.
PI director professor David Blake explains: "The ONS assumes that babies are born evenly throughout the year and that you can get a good estimate of the mid-year population by taking the estimate at the beginning of the year, the estimate at the end of the year, and halving it.
"But if you've had an uneven pattern of births during the course of the year, your mid-year estimate will be wrong and that will follow that cohort throughout its life so you will either be systemically overestimating or underestimating the mortality rate."
There was a similar distortion in birth rates after the Second World War. There was the well-documented ‘baby boom', which is the era when many of today's pensioners were born.
Essentially, the ONS overestimated the number of people born during these periods, introducing 'phantoms' into the data. This is problematic because mortality rates are determined by the number of people who die at a given age divided by the population who remain alive at that age.
Blake says: "It is quite important because the people doing the buyout are trying to project ahead how long people are going to live. If it's got the wrong estimate it could be out by a few percent."
Blake adds: "Although there aren't many people from 1919 left, there are a lot of people who were born in 1947 who are now pensioners and who would be part of these buyout deals. For 1947, the error was about 4% or 5%, whereas the normal error is around plus or minus 1%."
Hymans Robertson partner Douglas Anderson notes the impact will be greatest in mature pensioner portfolios. "Data on deaths among the oldest in society is sparse, so actuaries tend to place lines of best fit through that data. Changes in the data can change where those lines of best fit sit.
"This shouldn't have a huge effect on estimates of the baseline mortality, but it does impact on how we project mortality improvements." The Continuous Mortality Investigation (CMI) is proposing to make an allowance for this in the next version of its mortality projections model, according to Anderson.
LCP partner Ken Hardman is sceptical about the impact the flawed assumptions have on buyout prices. This is because insurers and reinsurers limit their reliance on ONS data, preferring to gather their own figures. "Now some of that ONS data might feed into some of the projections about how mortality assumptions will change in the future, but there are other sources of data," he says.
Longevity rates are only part of the picture, according to Hardman. Mortality assumptions also take into account expected medical advancements over the following two or three decades. He says: "There's inevitably some subjectivity in that. I can see why there are potential flaws in that methodology but I don't think they are necessarily directly feeding into pricing."
A fair price
JLT head of buyouts Martyn Philips asks whether schemes are really overpaying if the price matches their affordability criteria. He says: "Prices can always be lower, but if the scheme was happy to pay it and the sponsor was happy with the transaction, then a happy set of trustees have done a trade at a price they felt was fair."
Philips says insurers also guard their view of longevity risk "like the crown jewels". This is because it can set them apart from their competitors. "They have invested money in building their own books of mortality information, and therefore won't be that reliant on things like ONS information."
Legal & General says its dataset is based on exact dates of birth. This meansthe mortality rates it uses for people who were born in the ‘uneven' years is accurate. The insurer says: "With our experience within the retail and bulk annuity markets we've developed, and use, our own dataset, so we are not reliant on the ONS data."
Still, Blake points out that even an error of 1% is significant in the case of multi-billion pound transactions. "In the cases that we have discussed, the population has been estimated to be too high, which means the mortality rates are too low, which means you are predicting more people will survive. That means the buyout price is higher than it needs to be."
Philips agrees that a 1% error is material. However, he adds: "At the end of the day, buy-ins and buyouts only happen if the price that's on the table is deemed to be affordable. And it's not just mortality; there are many other factors that go into pricing as well." But more deals would become affordable if prices fell by 1% across the board, Philips says.
Ultimately, pension schemes must ensure their member data is complete and easy-to-use because this will form the backbone of insurers' pricing estimates. Hardman says: "Anything that can be done to improve the general information set that is out there on this will just increase the understanding of longevity."
The High Court has blocked the £12bn transfer of Prudential's annuity book to Rothesay Life, citing the insurer's lack of "established reputation" and differing "capital management policies".
This week's top stories included Legal & General acquiring MyFutureNow to provide a dashboard service to customers, while also agreeing a hybrid buy-in with a Hitachi scheme.
NEST has signed up to the government-backed Star Initiative, taking all of its 8 million members' pension pots with it.
It is perhaps inherently difficult to find an agreed definition of value for money, but some methodologies could act as a stopgap, argues Jonathan Stapleton.